Important Notice

This copy may be used only for

the purposes of research and
private study, and any use of the
copy for a purpose other than
research or private study may
require the authorization of the
copyright owner of the work in
question. Responsibility regarding
questions of copyright that may
arise in the use of this copy is
assumed by the recipient.
 UNIVERSITY OF CALGARY

"Integration of the seismic data with rock physics and reservoir modeling in the FRS project"

by

"Davood Nowroozi Charandabi"

A THESIS
SUBMITTED TO THE FACULTY OF GRADUATE STUDIES
IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE
DEGREE OF "DOCTOR OF PHYLOSOPHY"

GRADUATE PROGRAM IN GEOLOGY AND GEOPHYSICS

CALGARY, ALBERTA

JUNE, 2017

© "Davood Nowroozi Charandabi" 2017

 Abstract

This thesis is related to a CO2 injection project from the well logging to seismic modeling and

imaging, so many disciplines are involved. The reservoir is at a shallow depth of 300 m so it is in

a low temperature and low pressure state. A black-oil reservoir simulation was not appropriate for

the study, so a compositional method was used for the fluid simulation. The change of phase

possible around the anticipated pressure and temperature for CO2 injection is another limitation

for a compositional simulation, so the gas phase injection was selected for the simulation

modelling. Results show that the CO2 injection will decrease the density of formation around 3%,

and the P-wave velocity between 7 and 15%. It can also affect the S-wave velocity, and in the

seismic studies, there is enough of a change in the S-wave velocity to consider PS and SS-wave

data for the reservoir characterization. The rock physics equations solved for the pressure changes

by the Equation of State for CO2 and for the brine and a set of curves related to the fluid mixed

type were introduced. After 5 years of injection at bottom-hole pressure of 4.9 MPa, the injected

CO2 plume has a diameter of 185 m

The seismic studies based on the rock physics models show that the fluids mix type is a

determinative factor for interpretability of a reservoir. Seismic forward modelling was undertaken

using both acoustic and elastic finite difference approaches, and imaging was done using reverse

time migration. For patchy or semi-patchy saturation, mixed with a linear (or near linear)

converter, the saturation is calculated with an acceptable error by the acoustic, seismic response.

In a parabolic converter as Reuss average in a fine mixed type, the time-lapse acoustic response is

insufficient to identify saturation explicitly.

ii
 Acknowledgements

Foremost, I would like to express my sincere gratitude to my supervisor Dr. Don Lawton for

his unsparing guidance and support that helped me a lot to accomplish writing this thesis.

I thank the sponsors of CREWES for continued support and CMC Research Institutes Inc for

access to the data. This research was funded by CREWES industrial sponsors and NSERC (Natural

Science and Engineering Research Council of Canada) through the grant CRDPJ 461179-13. I

would like to thank Schlumberger for the use of OMNI, VISTA, Petrel and ECLIPSE and

Computer Modeling Group LTD. for CMG simulation software.

Also, I would like to thank Amir Ghaderi for help and supervise the reservoir simulation

section, Hassan Khaniani for significant aid in the seismic imaging, Mohammad Soroush for assist

in the geomodeling, Helen Isaac for the seismic data processing and Andreas Cordsen for some

excellent points in the seismic design.The VSP Seismic data acquisition has done by CREWES

team (Malcolm Bertram, Kevin Hall, and Kevin Bertram).The CREWES project was managed

successfully during my research by Laura Baird, that I would like thank her.

iii
 Dedication

To Ava and Alma

iv
 Table of Contents

Abstract ........................................................................................................................ ii
Acknowledgements ..................................................................................................... iii
Dedication ................................................................................................................... iv
Table of Contents ..........................................................................................................v
List of Tables ............................................................................................................. viii
List of Figures and Illustrations .................................................................................. ix
List of Symbols, Abbreviations and Nomenclature ................................................ xxiii

CHAPTER 1. INTRODUCTION ....................................................................................1

1.1 Research procedure and resources ...........................................................................1
1.2 Accomplishments and goals ....................................................................................2
1.2.1 Software .............................................................................................................5
1.2.2 Data Resources...................................................................................................6
1.2.3 Background ........................................................................................................6
1.3 CO2 sequestration...................................................................................................10
1.3.1 Introduction ......................................................................................................10
1.3.2 CO2 in the atmosphere .....................................................................................10
1.4 CO2 sequestration, a method for decreasing CO2 concentration in the atmosphere14
1.4.1 CO2 trapping mechanism .................................................................................15
1.4.2 CO2 gas injection in the brine ..........................................................................18

CHAPTER 2. THE FIELD RESEARCH STATION (FRS) PROJECT ........................19

2.1 Introduction ............................................................................................................19
2.2 Geological setting ..................................................................................................21
2.3 Petrophysical study and Interpretation...................................................................25
2.4 Core study ..............................................................................................................30
2.5 Lithology and Mineral study of the Belly River sandstone ...................................32
2.6 Core analysis ..........................................................................................................34

CHAPTER 3. SEISMIC DESIGN FOR 3C-4D PROPOSE IN THE FRS ....................36

3.1 Introduction ............................................................................................................36
3.2 Background information.........................................................................................36
3.3 Considerations for a 4D seismic design and acquisition .......................................37
3.4 Targets....................................................................................................................38
3.5 Frequency content ..................................................................................................39
3.6 Velocity-depth function .........................................................................................40
3.7 Bin size...................................................................................................................40
3.8 Box size and geometry ...........................................................................................42
3.9 Design option .........................................................................................................42
3.9.1 The acquisition parameters ..............................................................................43
3.10 PS survey design ....................................................................................................48
3.11 A criterion for fold distribution..............................................................................51
3.12 Improving PS fold coverage: randomize pattern ...................................................52

v
CHAPTER 4. THE BASELINE SEISMIC DATA, INTERPRETATION AND
GEOMODEL DEVELOPMENT..............................................................................55
4.1 Introduction ............................................................................................................55
4.2 Field acquisition .....................................................................................................55
4.3 LMO effect and fold taper .....................................................................................58
4.4 Seismic data resolution and coherency ..................................................................59
4.5 Seismic PP and PS-wave data after processing .....................................................61
4.6 Seismic interpretation: The phantom horizons methods ........................................64
4.7 Well ties .................................................................................................................65
4.8 Attributes study ......................................................................................................72
4.8.1 Generic Inversion .............................................................................................72
4.9 Fault detection attributes ........................................................................................74
4.9.1 Structural smoothing ........................................................................................75
4.9.2 Dip Deviation ...................................................................................................75
4.9.3 Chaos................................................................................................................75
4.9.4 Variance ...........................................................................................................75
4.9.5 Ant tracking .....................................................................................................76
4.10 The velocity model ................................................................................................81
4.11 Geomodel of the project area .................................................................................82
4.12 Introduction to Geostatistics: .................................................................................85
4.12.1 Variogram: ..............................................................................................85
4.12.2 Variogram models: ..................................................................................86
4.12.3 Anisotropies ............................................................................................88
4.12.4 Kriging: ...................................................................................................89
4.13 Primary models ......................................................................................................89
4.14 The isotropic geomodel for fluid simulation .........................................................91
4.15 The geomodel for the seismic modeling and imaging ...........................................94

CHAPTER 5. RESERVOIR FLUID SIMULATION FOR FRS PROJECT .................96

5.1 Introduction ............................................................................................................96
5.2 Initial state of the FRS reservoir ............................................................................97
5.3 The reservoir concepts and simulation methods ....................................................99
5.3.1 Relative permeability .......................................................................................99
5.3.2 Anisotropy of the permeability ......................................................................101
5.3.3 Formation compressibility .............................................................................102
5.3.4 Darcy’s law ....................................................................................................104
5.3.5 Simulation methods .......................................................................................105
5.3.6 Black-oil simulator.........................................................................................108
5.3.7 Specifications and Advantages of Compositional Simulator.........................109
5.4 The result of the reservoir simulation ..................................................................110
5.5 Simulation results.................................................................................................112
5.6 The CO2 gas injection effect on fluid phase in the reservoir ...............................116
5.7 Injection with different BHP ................................................................................117
5.8 Long-term prediction ...........................................................................................118
5.9 Conservation of CO2 mass after stopping the injection .......................................122
5.10 An injection/production pattern for improved gas phase injection ......................126
5.11 Injection at higher BHP .......................................................................................129
vi
CHAPTER 6. ROCK PHYSICS STUDY FOR THE FRS PROJECT ........................140
6.1 Introduction ..........................................................................................................140
6.2 The complexity of a solid or fluid........................................................................141
6.3 The physical properties of the mix phases ...........................................................144
6.4 Voigt and Reuss average ......................................................................................144
6.5 Brie’s average for fluid mix .................................................................................146
6.6 Hashin-Shtrikman (HS) Bounds ..........................................................................147
6.7 Fluid properties ....................................................................................................151
6.7.1 Brine ...............................................................................................................151
6.7.2 Carbon dioxide ...............................................................................................155
6.8 Mixed fluid properties..........................................................................................157
6.9 Effect of pore pressure on seismic velocity .........................................................160
6.10 The physical properties in the matrix...................................................................162
6.11 Fluid substitution .................................................................................................164
6.12 Model Assumption ...............................................................................................165
6.13 Practical usage of the Gassmann’s equation ........................................................167
6.14 The CO2 gas injection effect on the formation velocity in the field and lab test .169
6.15 Time delay caused by the injected fluid ..............................................................173
6.16 The formation velocity and density after CO2 injection in the FRS reservoir .....174
6.17 The velocity and density model based on the rock physics study .......................176

CHAPTER 7. SEISMIC IMAGING ............................................................................183

7.1 Introduction ..........................................................................................................183
7.2 The research method ............................................................................................184
7.3 Acoustic forward modeling strategy ....................................................................185
7.4 RTM migration strategy.......................................................................................186
7.5 Boundary Conditions ...........................................................................................187
7.5.1 The Perfectly Matched Layer (PML) .............................................................188
7.6 Low-frequency noise due to RTM procedure ......................................................192
7.7 Condition for successful 4D study .......................................................................193
7.8 Interpretability of an event ...................................................................................194
7.9 Plume size and velocity variation ........................................................................195
7.10 A diffusive and solid velocity model ...................................................................196
7.11 Seismic response of a solid and diffusive velocity model ...................................198
7.12 Acquisition geometry for 4D seismic surveys .....................................................205
7.13 Seismic interpretability of a diffusive velocity model .........................................209
7.14 Seismic imaging for FRS project .........................................................................211
7.15 FRS reservoir seismic time-lapse results .............................................................226
7.16 The elastic medium and the seismic response of the reservoir ............................232
7.17 Seismic response of CO2 injection in a complex geological setting ....................236

CHAPTER 8. CONCLUSIONS AND THE RECOMMENDATIONS FOR FUTURE

WORK 240
8.1 The research trend in the future ...........................................................................242

APPENDICES .................................................................................................................245

vii
REFERENCES ................................................................................................................250

List of Tables

Table 1-1: The large-scale CCS were operated in Canada (Source: Global CCS Institute) ......... 15

Table 2-1: The mineral types in the reservoir zone by the well log data analysis (from Figure
2-12) ...................................................................................................................................... 32

Table 2-2.Effective porosity and water and gas saturation in the core ......................................... 34

Table 2-3.The measured total porosity and water and gas saturation in the core ......................... 34

Table 3-1.The Targets were considered for the seismic design purpose. ..................................... 38

Table 3-2. The acquisition parameters. ......................................................................................... 43

Table 3-3.The variance test result for 12 different acquisition pattern. ........................................ 54

Table 5-1.The initial properties of the injection target ................................................................. 99

Table 5-2. The different BHP and related figures....................................................................... 129

Table 6-1. The fraction of the minerals in the reservoir based on the well log data analysis. .... 163

Table 6-2.The mixed minerals bulk and shear modulus calculated by three average methods. . 163

Table 7-1: Acquisition parameters and patterns ......................................................................... 207

Table 8-1: The specification of FWI study. ................................................................................ 244

viii
 List of Figures and Illustrations

Figure 1-1.The disciplines used for the reservoir integration in this thesis .................................... 2

Figure 1-2.Schematic view of the research procedure. ................................................................... 3

Figure 1-3: The concentration of CO2 and the temperature of the atmosphere from late
Precambrian to recent. The blue line demonstrates the temperature fluctuations, (Scotese
et al. 2002) (Pagani et al. 2005). ........................................................................................... 11

Figure 1-4. Temperature, CO2, and dust concentration in the atmosphere from 400,000 years
ago. Temperature has a strong correlation with the CO2 concentration in the atmosphere
and reverses relation with the dust. The loess sediments are a real proof of dust
concentration in the atmosphere during the cold stages (Petit et al., 1999). ......................... 12

Figure 1-5. CO2 in atmosphere and temperature from 50,000 years ago, by Vostok ice cores
study (Petit et al., 1999, and joannenova.com.au). ............................................................... 13

Figure 1-6: The concentration of CO2 in the atmosphere since 1958 to 2015 (Measured at the
Mauna Loa Observatory, https://www.co2.earth). ................................................................ 13

Figure 1-7: World energy consumption by source (based on data from BP Statistical review,
2014) ..................................................................................................................................... 14

Figure 1-8. CO2 trapping mechanisms and increasing CO2 storage security over time
(Pooladi-Darvish, 2009). ....................................................................................................... 17

Figure 2-1.The map of project area, it is located in 017-16-W4................................................... 20

Figure 2-2. The satellite image of the FRS, the green rectangle shows the project area and the
yellow points are the well sites with log data (Source: Google Earth). ................................ 20

Figure 2-3.The outcrop of two upper formations in the Belly River group in the Bow River
valley near the Bow City, west of project area and 7.5 km far from the well site (AGS,
2010). .................................................................................................................................... 22

Figure 2-4.The stratigraphic correlation and table of formations in the southern plain (Alberta
Geology Survey (AGS), 2010).............................................................................................. 23

Figure 2-5. The Stratigraphic chart for the well 10-22 (from well 10-22 drilling data
combined with the adjacent wells data, Schlumberger, 2015) .............................................. 24

Figure 2-6. P-wave versus s-wave slowness in the well 10-22. The color-bar shows the
gamma ray log, and the size of circles are for density porosity. The Vp/Vs ratio is
between 1.8 in the deep layers to 2.6 in the shallow formations. ......................................... 26

ix
Figure 2-7. The well log data for seismic porosity (SPHI), density porosity (DPHZ), TCMR
porosity, the average of mentioned three porosity (PHI), the permeability (KTIM:
Timure-Coates, KSDR: Schlumberger-Doll-Research model), upscaled data for PHI and
KTIM, and Vp/Vs ratio near the injection zone.................................................................... 27

Figure 2-8: The Timur-Coated model permeability (mD) vs. density porosity (v/v). The
different statistical populations are recognizable because of the different lithological and
sedimentation condition. ....................................................................................................... 28

Figure 2-9: Petrophysical interpretation result for the injection zone and around. The
minerals type was characterized by the well logs. ................................................................ 29

Figure 2-10. There is no relation between KTIM and other log data. Just a meaningful
relation is between CMFF (free fluid porosity (index) from CMR) and KTIM. The
coordinate is full logarithmic scale. For more information see Appendix A. ....................... 30

Figure 2-11. Core photo from the injection zone (sandstone) (taken by Schlumberger). ............ 31

Figure 2-12: The result of the mineral types discriminated by the well log data (Kirk Osadetz,
personal conversation) .......................................................................................................... 33

Figure 2-13: Before and after log-to-core calibration of K_INT data for the injector ................. 35

Figure 3-1. Frequency analysis: for the shallowest target (A) and the deepest targets (B) .......... 39

Figure 3-2. A linear function for velocity is used for calculation of bin size and migration
aperture. Velocity function for FRS project regards to well log data (wells 11-22-17-16
(a) and 7-22-17-16 (b) and 10-22-17-16 (c)) is V=V0+kz=2650+z. .................................... 41

Figure 3-3. Bin size for the shallow target with 80Hz max frequency (left) and the deep target
with 65Hz (right). .................................................................................................................. 42

Figure 3-4.The survey geometry. The red points demonstrates shot points, and the blue ones
are receivers. There are a dense shot and receiver points in the mid of survey with
dimension equal to 500*500 m. ............................................................................................ 44

Figure 3-5. A. The fold map. The highest numerical range for the fold is 220 for both shown
by the red color in the scalebar. The yellow circle shows the fold range between 30-40.
B. The fold map for offset 0-700 m. ..................................................................................... 44

Figure 3-6. A. The azimuth distribution in the full fold zone. It shows a proper azimuthal
coverage; the azimuth depends on the shorter offsets. B. The offset distribution in the
full fold box. The acquisition covers full offset ranges. ....................................................... 45

Figure 3-7. A. Azimuth distribution in the target range (0-700 m offsets). B. Offset
distribution in the full fold box for 0-700 m. ........................................................................ 45

x
Figure 3-8. A. Offset redundancy, the target zone demonstrated by the black lines. B.
Azimuth redundancy, the number of traces that fall in each section; gaps indicate
missing azimuth .................................................................................................................... 46

Figure 3-9. A. Histogram of Fold, the numbers of bins that fall in each range of fold values.
B. Histogram of Offset, the number of traces that fall in each range of Offset values ......... 46

Figure 3-10. A. Histogram of azimuth, the number of traces that fall in each range of
azimuth values. B. Offset versus azimuth diagram. .............................................................. 47

Figure 3-11. The PS fold map (non-asymptotic method) for a target in 400 m depth, Vp to Vs
ratio is considered equal 2..................................................................................................... 48

Figure 3-12. A. The PS azimuth distribution in the full fold boxes. B. Offset distribution for
the PS wave. .......................................................................................................................... 49

Figure 3-13. Offset redundancy for PS wave, the number of traces that fall in each section;
gaps indicate missing offsets.The curve of boomerang shape distribution is a function of
p to s wave velocities ratio. ................................................................................................... 49

Figure 3-14. Azimuth redundancy for PS wave, the number of traces that fall in each section;
gaps indicate missing azimuth.It shows a proper azimuth distribution. ............................... 50

Figure 3-15.A sample of the suggested regular survey (rectangular box) and randomized
receivers pattern. ................................................................................................................... 53

Figure 3-16. The fold map for regular rectangular box pattern (left) and randomized receivers
pattern (right) ........................................................................................................................ 53

Figure 4-1. The seismic survey map of FRS project. The blue points are the receivers and red
shows the sources. Two pipelines caused some change in the shot point coordinates.
(Satellite image source: Google Earth) ................................................................................. 56

Figure 4-2.The PP-wave fold map for (a). total nominal fold and (b). 0-300 m source-receiver
offsets. ................................................................................................................................... 57

Figure 4-3. (a). Azimuth distribution and (b). Offset distribution ................................................ 57

Figure 4-4. The PS-wave fold map for the target in the 300 m depth. ......................................... 58

Figure 4-5.The low fold zone and acquisition on the acquired seismic data. ............................... 59

Figure 4-6.The configuration of acquisition were used for the processing. Left survey shows
and right is the same survey after eliminating the dense center region. ............................... 60

Figure 4-7. The result of processing for the acquired PP seismic data in (left) and after
eliminating the dense source-receiver CMPs (right).The arrows show the main
differences between two seismic sections. ............................................................................ 61

xi
Figure 4-8. The seismic data, A. processed PP-wave data, B. processed PS wave data .............. 62

Figure 4-9. PP-wave migrated seismic section on inline 41 ......................................................... 63

Figure 4-10. PS-wave migrated seismic section on inline 41 ....................................................... 63

Figure 4-11.The TWT conversion diagram between PP and PS seismic sections. The
injection target at 233 ms for PP time and 364 ms for PS time. It was calculated from the
P and S-wave slowness logs. The velocity for no well log data zone calculated by the
seismic analysis. .................................................................................................................... 64

Figure 4-12. Flowchart that shows the simplified stages of work for a general structural
interpretation. ........................................................................................................................ 67

Figure 4-13: schematic of procedure used for well to seismic calibration ................................... 67

Figure 4-14. Ricker wavelet and the synthetic seismogram from well log data against
processed seismic data. ......................................................................................................... 68

Figure 4-15.The wavelet calculated by extended white method for the Sensor processed data
set. The reflectors correlation is acceptable in the 200-300 ms range. ................................. 68

Figure 4-16. PP seismic interpretation in time domain including the main formations and
phantom layers (time domain). The line is passing of the main well. .................................. 69

Figure 4-17. The top Basal Belly River sandstone as the top of the reservoir in the seismic
cube in time domain. ............................................................................................................. 69

Figure 4-18. Time structure of the top of the Basal Belly River sandstone.................................. 70

Figure 4-19. Depth structure of the top of Basal Belly River Formation. .................................... 71

Figure 4-20.Genetic Inversion (time slice is 235 ms near the top of reservoir) ........................... 73

Figure 4-21.Genetic Inversion on the top reservoir surface ......................................................... 73

Figure 4-22. Fault detection using the multi-attribute method. .................................................... 74

Figure 4-23.Smoothing attribute is the first step of the multi-attribute method for faults
recognition. ........................................................................................................................... 77

Figure 4-24.The single attribute study (dip deviation) for fault recognition. ............................... 77

Figure 4-25.The chaos attribute applied to the smoothed data cube............................................. 78

Figure 4-26. The variance attribute applied to the smoothed data cube. ...................................... 78

Figure 4-27. AntTrack attribute applied to the Smoothed Chaos data cube. ................................ 79

xii
Figure 4-28. AntTrack attribute on the Smoothed Variance attribute result ................................ 79

Figure 4-29.A multi-attribute map (Structural- Chaos – Ant track) for the fault
discrimination. ...................................................................................................................... 80

Figure 4-30. A multi-attribute map (Structural - Variance – Ant track) for the fault
discrimination. ...................................................................................................................... 80

Figure 4-31.The P-wave velocity in the well 10-22 with simplified gradients for time to depth
conversion ............................................................................................................................. 81

Figure 4-32.The time to depth conversion for the PP, PS and SS wave in the well 10-22........... 82

Figure 4-33: The procedure for producing a geomodel in the FRS project. ................................. 83

Figure 4-34. The upscaled well log data for building geomodel in well 10-22 before
permeability calibration. ....................................................................................................... 84

Figure 4-35: A typical Variogram diagram and main parameters. ............................................... 86

Figure 4-36: Three variogram models. ......................................................................................... 87

Figure 4-37: The histogram (to check the distribution type of the data) and semi-variance
with a spherical model fitted for the density porosity variable (The nugget effect =
0.0002080, the sill= 0.0023660, the range= 2.0700 m for h=15 cm in the well direction). . 88

Figure 4-38. A geomodel made from 11 wells and a small cell size around the injection well.
The dimension of geomodel is 3900*3000 m. ...................................................................... 90

Figure 4-39.The geometry of layers in the geomodel from the seismic interpretation. Colors
shows the main geological events and formations, dimension is 1000*1000 m .................. 90

Figure 4-40.The revised geomodel derived from log data from three wells. The dimension of
geomodel is 1000*1000 m. ................................................................................................... 91

Figure 4-41.The porosity geomodel made up with the well log data. The geomodel size is
1000*1000*240 m. ............................................................................................................... 92

Figure 4-42. A section of the xy permeability geomodel shows the grid size in the injection
layer and others. The red rectangle shows the injection layer (i.e. Basal Belly River) ........ 92

Figure 4-43. A section of geomodel for the permeability in z direction....................................... 93

Figure 4-44. A section of geomodel for the porosity. ................................................................... 93

Figure 4-45.The P-wave velocity model. This model was used for the seismic imaging. ........... 94

Figure 4-46. The density model used for seismic modelling. ....................................................... 95

xiii
Figure 4-47: The P-wave velocity oriented by seismic layers (a 2D view of Figure 4-45 in a
section passing of the 10-22 well). The section is W-E and view is to the North. ............... 95

Figure 5-1. Disciplinary contributions to reservoir flow modeling (after Fanchi, 2006) ............. 96

Figure 5-2.Hydrostatic and lithostatic pressure in the reservoir; the latter is calculated from
the density log data. .............................................................................................................. 98

Figure 5-3. Trapping efficiency in sandstone based on previous work in Alberta (Bachu,
2013) ................................................................................................................................... 101

Figure 5-4. The relative permeability curve calculated for the reservoir. .................................. 101

Figure 5-5.Geomechanical properties of the reservoir in the well 10-22-17-16. The

compressibility was demonstrated by cf (the unit is 1/GPa). .............................................. 103

Figure 5-6. Volume elements or grid block in reservoir simulation. .......................................... 105

Figure 5-7.Reservoir gridblock, coordinate and directions ........................................................ 107

Figure 5-8. Appropriate situation for Compositional and black-oil and compositional
simulators for oil and gas phases (ECLIPSE course material, 2016). ................................ 109

Figure 5-9: The phase diagram of carbon dioxide and pressure and temperature condition in
the FRS project (Phase diagram from ChemicaLogic Corporation). .................................. 111

Figure 5-10. The phase diagram of CO2 for the reservoir condition. This figure is a magnified
image of Figure 5-9 near the reservoir condition. ............................................................... 111

Figure 5-11.The simulation result for the reservoir pressure by the CO2 injection for five
years with BHP=4.9 MPa. The scale is same as Figure 5-12.The unit for the pressure is
MPa. .................................................................................................................................... 113

Figure 5-12. The CO2 gas saturation for the five-year injection by BHP=49 bar (4.9 MPa) ..... 114

Figure 5-13: Diagrams showing the result of injection for BHP=49 bar for five years (the x-
axes show the year of injection) a. Cumulative gas mass (kg), b. Cumulative gas
volume (m3), c. Daily volume (SC) injected gas rate (m3/day), d. Daily mass injected
gas rate (kg/day), e.Well bottom hole pressure (kPa), f. Well block pressure (kPa), (SC
stand for Standard Condition - 15oC and 1 bar). ................................................................. 115

Figure 5-14.Results of visual inspection of the brine+CO2-reacted sample; deposits of NaCl

crystals (salt precipitation) and calcite dissolution textures at the outer surface of the
sample in a lab test (Rathnaweera et al., 2016). .................................................................. 116

Figure 5-15. Gas saturation after five-year injection with BHP=48 bar. .................................... 118

Figure 5-16.The saturation after five-year injection by constant BHP. ...................................... 119

xiv
Figure 5-17. Predicted CO2 saturation five years after discontinuing the injection ................... 119

Figure 5-18. Predicted CO2 saturation ten years after stopping the injection ............................ 119

Figure 5-19. Predicted CO2 saturation 20 years after stopping the injection ............................. 120

Figure 5-20. Predicted CO2 saturation 40 years after stopping the injection ............................. 120

Figure 5-21. Predicted CO2 saturation 60 years after stopping the injection ............................. 120

Figure 5-22. Predicted CO2 saturation 80 years after stopping the injection ............................. 121

Figure 5-23. Predicted CO2 saturation 100 years after stopping the injection ........................... 121

Figure 5-24. The Bottom-hole pressure changes over a century. The pressure will be equal to
initial reservoir pressure after the year 2044 (28 years after beginning the injection
process). .............................................................................................................................. 122

Figure 5-25.The thickness of the plume in the injection point after stopping the injection.
After 100 years, all cells in the plume show a saturation rate around the trapping
efficiency, and so continued gas migration will change to a mainly horizontal direction. . 123

Figure 5-26. The result of long-term monitoring for the gas saturation after 233 years from
the start of injection. The maximum plume thickness is 2 m. ............................................ 126

Figure 5-27.The pressure condition in 25,50 and 200 m distance of injection well after one-
year injection. ...................................................................................................................... 127

Figure 5-28.The plume size after 1 and 5-year injection and 100-year post-injection. .............. 128

Figure 5-29. The production well(s) in X m distance to decrease the reservoir pressure, it can
help to inject more mass in the gas phase. T is the angle between the wells. X, T and the
number of wells are variable. .............................................................................................. 128

Figure 5-30. CO2 saturation at pressure of 51.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031). ................................................................................................................... 130

Figure 5-31.The pressure change in the reservoir for BHP=51.41 bar and temperature =20oC,
(a). one year after injection, (b). Five-Year injection, (c). Two years after stopping the
injection............................................................................................................................... 131

Figure 5-32.The cumulative gas mass (kg) injected at the constant BHP=51.41 bar. ................ 132

Figure 5-33.The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=51.41 bar. ......................... 132

xv
Figure 5-34. CO2 saturation at pressure of 53.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031). ................................................................................................................... 133

Figure 5-35. The cumulative gas mass (kg) injected at the constant BHP=53.41 bar ................ 134

Figure 5-36. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=53.41 bar .......................... 134

Figure 5-37. CO2 saturation at pressure of 55.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031). ................................................................................................................... 135

Figure 5-38. The cumulative gas mass (kg) injected at the constant BHP=55.41 bar ................ 136

Figure 5-39. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=55.41 bar .......................... 136

Figure 5-40. CO2 saturation at pressure of 57.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031). ................................................................................................................... 137

Figure 5-41. The cumulative gas mass (kg) injected at the constant BHP=57.3 bar .................. 138

Figure 5-42. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=57.3 bar ............................ 138

Figure 5-43. The pressure over 5 years of injection at BHP=57.3 bar (5.73 MPa), and 10
years after injection. ............................................................................................................ 139

Figure 5-44. Cumulative CO2 mass for different BHP for a five-year injection plan. An
exponential function describes the relation between BHP and injected gas mass. ............. 139

Figure 6-1. The fluid substitution procedure used in this chapter. ............................................. 142

Figure 6-2. A microscopic thin section of sandstone in the polarization microscope. It

demonstrates a variety of minerals in a rock (source: micro.magnet.fsu.edu).................... 143

Figure 6-3: The bulk modulus for the mix of a porous sand with 100% plagioclase (as
immature sand) and water. .................................................................................................. 146

Figure 6-4.Two phase material in the Hashin-Shtrikman bounds. ............................................. 148

Figure 6-5.The average bulk modulus for a mixed case of quartz and calcite. The blue curves
show Voigt, Reuss and VRH averages. The red curve is HS+ and green is HS-. The
VRH is very near to Hashin-Shtrikman averages. .............................................................. 150

Figure 6-6. Matrix Properties calculated by Voigt (blue), Reuss (red) and VRH (green)
methods for a mix of pure quartz and wet clay. .................................................................. 150
xvi
Figure 6-7. The density, bulk modulus and P-wave velocity of brine and water temperature
from 13 to 28 oC (with 5 oC steps) and salinity equal to 8000 ppm from 2 to 8 MPa. ....... 153

Figure 6-8.The density, bulk modulus and P-wave velocity of water and brine (salinity=8000
ppm) as a function of temperature. Each curve belongs to the pressure from 1 to 10 MPa
in steps of 2 MPa................................................................................................................. 154

Figure 6-9. The viscosity of brine (based on Batzle-Wang (1992)), the pressure does not have
a significant influence on the brine viscosity. ..................................................................... 155

Figure 6-10.The bulk modulus and density of CO2 at different pressures and temperatures.
EoS described by Span and Wagner (1996) were used to generate the diagram (drawn
by Yam,2011). .................................................................................................................... 156

Figure 6-11. Density of CO2 for 13 and 20oC and 23<p<57 bar. ............................................... 156

Figure 6-12. P-wave velocity of CO2 versus pressure at T= 13 and 20 oC (the velocity
calculated upper than 4 MPa at T=13 oC was unstable) ..................................................... 157

Figure 6-13. The density of the mixed fluid in T=13.8oC and pressure from 30 to 60 bar (3 to
6 MPa). ................................................................................................................................ 158

Figure 6-14. The bulk modulus for the mix of brine with 8000 ppm salinity and CO2 in 13 oC
and 4.5 MPa (45 bar). In the mixed fluid condition (as CO2 and brine), the Hashin-
Shtrikman averages (upper and lower bounds) are using the Reuss Average. ................... 159

Figure 6-15. The P-wave velocity in the mixed fluid of the brine (8000 ppm salinity) and
CO2 in T=13 oC and P=4.5 MPa (the reservoir condition during the injection
procedure). .......................................................................................................................... 159

Figure 6-16. The bulk modulus of the mixed fluid with a different fraction of CO2 and
different mixed condition in P=30 bar (3 MPa) and 45 bar (4.5 MPa). .............................. 160

Figure 6-17. The velocity change in the fluid phase of the reservoir (brine+CO2) for a semi-
patchy mixed fluid by pressure. The pressure increased from 3 to 4.5 MPa. ..................... 161

Figure 6-18: Bulk modulus estimation for different fraction of fluid mix by Reuss average
(fine mixed fluids) in T=60 C and different pressures (16 to 40 MPa) (P and T for Nisku
aquifer condition, WASP project; Nowroozi, 2014)........................................................... 162

Figure 6-19. The error in density calculated by the mineral discrimination method and well
log data in the injection horizon.......................................................................................... 164

Figure 6-20. The influence of mixing method on the P-wave velocity. Reuss average is
suitable for a fine mixed fluid, and the velocity change, in this case, is very dramatic in
the low saturation of CO2.Over 15% of CO2 saturation there is a slight increase in the
velocity of the formation (a test with the low-frequency laboratory data). The Voigt

xvii
 average is for a patchy mixing, and the velocity change is almost a linear decrease with
saturation (a test with the high-frequency laboratory data) (Smith, 2003). ........................ 170

Figure 6-21. P-wave velocity versus CO2 saturation from a field study. The blue dots show
the field data measurements from time-lapse well logs at the Nagaoka site in Japan
(Lumley, 2010). .................................................................................................................. 171

Figure 6-22. Results of lab test for the CO2 injection into a sandstone (Alemo et al., 2011) ..... 172

Figure 6-23. Laboratory and theoretical experiences for CO2 and a water flood effect on the
P-wave velocities (Wang, 2001). ........................................................................................ 172

Figure 6-24: Bulk modulus for CO2 and brine mixed phase. ..................................................... 175

Figure 6-25. P-wave velocity after CO2 injection in the reservoir calculated by Gassmann’s
equation, the shape of Vp diagram is a function of an average method for the fluid mix
(CO2+brine) properties calculation. The maximum possible CO2 gas saturation in the
FRS reservoir can reach to 50%. ......................................................................................... 175

Figure 6-26: The physical properties (S-wave velocity and density) change as a linear
function of the CO2 saturation in the reservoir condition in FRS project. .......................... 176

Figure 6-27: The density, and shear wave velocity change during the gas injection by the
constant bottom hole pressure (49.4 bar for five-year). The shear modulus remains
constant after injection, but decreasing in the density can make a small increase in the
Vs value. ............................................................................................................................. 178

Figure 6-28. The P-wave velocity model based on the Reuss average method that shows a
fine mixed fluid type after 1, 3 and five years’ injection. This model shows a uniform
velocity change in the reservoir volume. ............................................................................ 179

Figure 6-29. The P-wave velocity model based on the Brie’s average method that shows a
semi-patchy mixed fluid type after 1, 3 and five years’ injection....................................... 180

Figure 6-30. The P-wave velocity model based on the VRH average method that shows a
semi-patchy mixed fluid type after 1, 3 and five years’ injection....................................... 181

Figure 6-31. The P-wave velocity model based on the Voigt average method that shows a
fully-patchy mixed fluid type after 1, 3 and five years’ injection....................................... 182

Figure 7-1. Description of the abbreviations. ............................................................................. 187

Figure 7-2.The Boundary condition in the seismic model. The orange rectangles show the
internal boundaries. ............................................................................................................. 188

Figure 7-3. The model boundary is shown by ABCD. AB, CD, and AD have a SPML
boundary condition, and BC is a free surface boundary. .................................................... 191

xviii
Figure 7-4. The different components of the wave at the PML absorption boundaries for a P-
wave source in an elastic homogenous medium. ................................................................ 192

Figure 7-5. The internal structure of diffusive (an injective or productive point in the middle
of the ellipsoid) and solid velocity models. The reduction is linear from the center to the
outer bound in the diffusive model. The unit of velocity is m/s. The dimension of the
ellipsoid will define with the big and small diameter (a,b). ................................................ 197

Figure 7-6.The velocity model with two diffusive and solid velocity changes. The velocity
change in the center of diffusive model or whole solid shape is -7% equal -175 m/s
(2325 m/s). The size is ellipsoids are (200,40m). ............................................................... 198

Figure 7-7.The Uz component of the seismic response of diffusive and solid models in a
homogeneous medium. The left is the SM(m), and the right is SM(m)-SM(baseline). ..... 199

Figure 7-8. RTM(SM(m)) for Uz component and RTM(SM(m))-RTM(SM(baseline)) ............ 199

Figure 7-9. RTM(SM(m)) for pressure component and RTM(SM(m))-RTM(SM(baseline)) .. 199

Figure 7-10.The velocity model for a. Three-layer model as baseline b. Model a with
diffusive and solid velocity models as monitored model c. Subtracted result
(Monitored-Baseline model) ............................................................................................... 201

Figure 7-11.The pressure component of the seismic acoustic model for: a. 3-layer baseline
b. Baseline plus diffusive and solid velocity ellipsoids c. The difference ......................... 202

Figure 7-12. The Uz component (a.SM(base,Uz), b.SM(monitor,UZ), c.SM(base,Uz)-

SM(monitor,Uz)). It is similar to the pressure component but with the lower amplitudes. 202

Figure 7-13.The Ux component for (a.) 3-layer baseline (b.) Baseline plus diffusive and
solid velocity ellipsoids; (c.) The difference ....................................................................... 203

Figure 7-14. a. diff (RTM (SM (baseline, Pressure))), b. diff (RTM (SM (m, Pressure))), ....... 203

Figure 7-15. a. diff (RTM (SM (baseline, Ux))), b. diff (RTM (SM (m, Ux))),......................... 204

Figure 7-16. a. (RTM (SM (baseline, P))), b. (RTM (SM (m, P))), .......................................... 204

Figure 7-17. (RTM (SM (baseline, Ux))), (RTM (SM (m, Ux))), .............................................. 205

Figure 7-18. The diffusive velocity and density models for a 7% and 3% change in the
ellipsoid shape. The ellipsoid dimensions are180m wide and 10m in thickness ................ 207

Figure 7-19. The seismic model (a) and migrated section (b) for the surface survey. ............... 207

Figure 7-20. A seismic record and migrated section for the VSP acquisition ............................ 208

xix
Figure 7-21. a. The seismic response of cross well acquisition pattern of the model in Figure
7-18. b. After eliminating the surface and shot effects. c. The migrated data (from a). d.
The noise reduced migrated section. ................................................................................... 208

Figure 7-22: The seismic response (column b) and RTM result (column c) for a model (100 *
20 m) with a different velocity anomalies (column a). Higher velocity difference causes
greater amplitudes for the surface acquisition survey. ........................................................ 210

Figure 7-23: The seismic response (column b) and RTM result (column c) for two models
with a different plume size and 5% velocity change in the center of ellipsoids in column
a. a1: 50*20 m and a2:200*20 m. ....................................................................................... 211

Figure 7-24. The baseline P (a) and S-wave (b) velocity (m/s) and density (c) (gr/cc) models.
The reservoir saturation effect on the velocity and density after injection is included in
these models. ....................................................................................................................... 213

Figure 7-25. The migrated acoustic Uz component for the baseline velocity and density
model (surface seismic, five shots and 996 receivers). ....................................................... 214

Figure 7-26: The velocity and density models before and after five years’ injection with a
BHP=49.4 bar in the gas phase for the VRH average. The original physical properties
oriented by the seismic interpretation result. a. The base model before injection. b. The
perturbation model base on the saturation results. c. The physical properties after
injection. d. The magnified figures on the reservoir zone. ................................................. 215

Figure 7-27: The seismic model generated by the velocity and density patterns introduced in
Figure 7-26 (VRH average) for a surface seismic experience with one shot in x=500 m
and receivers with 1 m interval and from 0 to 1000 m. a. Baseline seismic model. b.
Baseline RTM result. c. Monitor seismic model. d. Monitor RTM result. e. The
difference between monitored and baseline seismic models (amplitude ten times
magnified). f. The difference between RTM results (amplitude ten times multipled). ..... 216

Figure 7-28.The amplitude of the seismic acoustic seismic modeling (Figure 7-27.A, section
AA’ on the red line) for the baseline, after one, three and five-year injection (VRH
average method). ................................................................................................................. 217

Figure 7-29: The seismic model generated by the velocity and density patterns introduced in
Figure 7-26 (VRH model) for a VSP survey with one shot at x=400 m and receivers
with 1 m interval at x= 600 and extending from 0 to 600 m depth. a. Baseline seismic
model. b. Baseline RTM result. c. Monitor seismic model. d. Monitor RTM result. e.
The difference between monitored and baseline seismic models. f. The difference
between RTM results. ......................................................................................................... 218

Figure 7-30. A VSP seismic model and image for a wide source to receivers aperture (400 m
distance). a and b show the seismic model and image for the baseline and c and d are for
the five-year injection calculated by Reuss average, d and e are the difference of the 5-
year injected model and baseline (wavelet: 55 Hz Ricker). ................................................ 219

xx
Figure 7-31. The seismic model generated by the velocity and density patterns introduced in
Figure 7-26 for a Cross-Well survey with one shot at x=400 m and 295 m depth and
receivers with 1 m interval at x= 600 and extending from 0 to 600 m depth. a. Baseline
seismic model. b. Baseline RTM result. c. Monitor seismic model. d. Monitor RTM
result. e. The difference between monitored and baseline seismic models. f. The
difference between RTM results. ........................................................................................ 220

Figure 7-32: The time lapse seismic models: a. The seismic model for one-year injection. b.
The difference between the baseline and (a). c. The difference between seismic models
after five and one year of injection. d. Migrated section of (a). e. The difference of
migration sections between the baseline and one year of injection data. f. The
difference of migrated data between five years and one-year of injection. ........................ 221

Figure 7-33. The migrated seismic data from the reservoir’s response by different kinds of
average related to the mixed fluid condition after a year injection. The left figures are
the model made by the rock physics models after one year injection for the Reuss(a),
Brie (b), VRH (c) and Voigt (d) averages. .......................................................................... 222

Figure 7-34. The migrated seismic data from the reservoir’s response by different kinds of
average related to the mixed fluid condition after 5-year injection. The left figures are
the model made by the rock physics models after five years injection for the Reuss(a),
Brie (b), VRH (c) and Voigt (d) averages. .......................................................................... 223

Figure 7-35.The variation of velocity due to injection after 5-year injection (calculated by
Reuss average) and the time-lapse seismic migrated response of it (RTM(SM(5,R)-
RTM(SM(base))) ................................................................................................................ 224

Figure 7-36.The statistical distribution of the velocity change in the reservoir cells in the
patchy mixed (Voigt average). ............................................................................................ 225

Figure 7-37. The statistical distribution of the velocity change in the reservoir’s cells in the
fine mixed (Reuss average). ................................................................................................ 226

Figure 7-38: The research routine to compare results of the seismic and reservoirs time lapse
surveys. A. shows the direct seismic time-lapse, B. seismic time-lapse based on the
reservoir time lapse ............................................................................................................. 228

Figure 7-39. The difference model (time lapse) for the p wave velocity (by VRH average) in
the reservoir between different years of injection. The result calculated according to the
CO2 saturation content and Gassmann’s equation for a semi-patchy mixed condition. ..... 229

Figure 7-40. Left: SM (R (5-year injection))-SM (R(1-year injection)) and right: SM (R ((5
year) - (1-year injection))). As mentioned previously, SM stand for seismic model
(Acoustic), and R is calculated Vp based on reservoir simulation result ........................... 230

Figure 7-41. The difference between two model in Figure 7-40, the left figure shows same
amplitude scale and the right one is 100 times magnified amplitude ................................. 230

xxi
Figure 7-42. The RTM results for the seismic models in Figure 7-40 ....................................... 230

Figure 7-43. The difference of RTM images in Figure 7-42. The left shows the difference in
natural amplitude and the right figure shows 100 times magnified. ................................... 231

Figure 7-44. The saturation and P-wave velocity change distribution after a year stopping the
injection............................................................................................................................... 231

Figure 7-45. The acoustic (left) and elastic seismic response (right) for a three-layer model
(top). .................................................................................................................................... 233

Figure 7-46. The P-wave seismic response for the acoustic wave propagation. a. shows the
seismic response for the baseline model. b. the seismic model after five-year injection
by Brie’s model. c. the difference section shows a PP response of the reservoir. .............. 234

Figure 7-47. The seismic response for an elastic model. a. baseline. b. after five-year
injection by Brie’s model. c. The difference section and PP, PS and SS seismic response
of the reservoir. As demonstrated in Figure 7-45, the SS-wave amplitude is considerable
at far offsets. ........................................................................................................................ 235

Figure 7-48.The original Marmousi model (P-wave velocity) ................................................... 237

Figure 7-49. A new physical property (Vp) defined as a CO2 injected reservoir pointed by the
red rectangle. ....................................................................................................................... 237

Figure 7-50.The seismic imaging result on the original Marmousi model. The acoustic wave
forward modeling and RTM migration method was used. ................................................. 238

Figure 7-51.The seismic imaging result for Marmousi model and the implemented reservoir. . 238

Figure 7-52.The subtract of monitor seismic model of the baseline model. The red rectangle
shows the location of the reservoir. .................................................................................... 239

Figure 8-1. The concept of FWI (Martinez, 2016). The FWI method is a suitable way for
correcting the velocity model according to the initial model and seismic acquired data. It
can be a revolutionary approach to explaining velocity change (that can be translated to
the saturation) in a reservoir by seismic 4D data in the seismic resolution range. ............. 243

xxii
 List of Symbols, Abbreviations and Nomenclature

Symbol Definition
B Bin size
BHP Bottom Hole Pressure
CCP Common converted point
CCS Carbone Capture and Storage
cf Compressibility
CGDN Core density
CKAR Core air permeability
CMFF Standard free-fluid porosity
CMP Common mid point
CPOR Core Porosity
EOR Enhanced Oil Recovery
fdom Dominant frequency
Fij Fold in bin number i*j
fmax Maximum frequency
FRS Field research site
G and µ Shear modulus
GR Gamma ray log
K Bulk modulus
KINT Intrinsic Permeability (ELAN)
krg Gas relative permeability
krg Gas relative permeability
krw Water relative permeability
KSDR Permeability (Schlumberger-Doll-Research model)
KTIM Permeability (Timur-Coates model)
LMO Largest minimum offset
PHI Average porosity
PHIT Total porosity
PIGN log-derived effective porosity
RC Reflection coefficient
RTM Reverse time migration
SPHI Porosity calculated by P-wave velocity

xxiii
TCMR Total CMR porosity
VCL Volume of Clay
Vint Interval velocity
Volume_Clay Volume of Clay
Vp Compressional wave velocity
Vs Shear wave velocity
VSP Vertical Seismic Profiling
VXBW Volume of Bound Water

xxiv
 Chapter 1. Introduction

1.1 Research procedure and resources

CMC Research Institutes (CMCRI) and the University of Calgary are conducting a research

project to monitor the behavior of subsurface CO2 injection. The project is known as Field

Research Station (FRS) which plans to study the trapping and leakage of CO2 gas injection into a

shallow target. I had been involved in the project from the initial steps of research, and so the

dissertation reflected my work and studies in different branches of the geoscience and engineering.

The thesis covers the geological studies, petrophysical analysis and interpretation, seismic design,

geomodelling, reservoirs fluid simulation, rock physics studies and seismic forward modeling and

RTM imaging for time-lapse study. Figure 1-1 and Figure 1-2 demonstrate the disciplines were

used in the study and their link together.

The various kinds of analysis were used for the project and the principal methodology for the

research includes:

1- Seismic design: PP and PS parameter design

2- Seismic interpretation: structural analysis and attributes study

3- Geomodelling: Geostatistical method, geometry from seismic and data from well log

4- Fluid Simulation: Compositional simulation

5- Rock Physics: Property average of the constituents for the solid and fluid phases, fluid

substitution, applicating Gassmann’s equation

1
 6- Seismic modeling and imaging: Seismic forward modeling (acoustic wave equation

solved by the finite difference method), Reverse Time Migration (RTM)

Geomodel

Seismic

Simulation

Rock Physics

Figure 1-1.The disciplines used for the reservoir integration in this thesis

1.2 Accomplishments and goals

In the seismic design chapter, the goal is to improve PS fold in the acquisition design and

define a criterion to compare the surveys based on the PS fold distribution. In the simulation

chapter, the condition of the reservoir (shallow reservoir) was determined, and the CO2 injection

was simulated in gas phase. Gas phase of CO2 was selected for two reasons: 1- In the gas injection,

the density will drop to half of initial density, and it can cause a small increase in the shear wave

velocity. This change will allow us to check elastic specifications of a reservoir; 2- The frack

pressure due to the shallow depth of the reservoir and low lithostatic pressure is very low, and so

pressure easily can exceed of the fracture gradient.

2
 In the reservoir simulation study a scenario for the long-term monitoring (after a limited

injection time) was formulated, and it describes the plume size after a long-term monitoring based

on the simulation results of initial years after ceasing the injection.

As mentioned, the research covers a very broad technical range including fluid simulation,

rock physics and seismic algorithms to make it possible to compare seismic migrated responses in

different conditions that are possible in the FRS reservoir. In the rock physics chapter, the physical

properties of the fluids in the reservoir, before and after injection were studied and the rock physics

study is integrated with the reservoir simulation results. Different average methods related to the

fluids mixed models (as patchy or uniform) were used to calculate the P-wave velocity after CO2

injection, and they were the basis for the seismic modeling and imaging.

Figure 1-2.Schematic view of the research procedure.

3
 The last chapter (seismic imaging) uses all processed data from geomodeling, the

reservoir simulation, and rock physics. In this step, the main advance is in the seismic modeling

of a reservoir under production/injection. For this purpose, the seismic forward modeling used the

wave equation solved by the finite difference method. The imaging step took advantage of the

Reverse Time Migration (RTM) method. The RTM is a two-way wave equation depth migration

method that can handle complex velocity models (near the reservoir), and suitable for steep dips

(to investigate the possible image of the plume in reservoir) and accurate amplitude estimation. In

this chapter, the different models of geometries and velocity variation due to the fluid substitution

and plume size in a reservoir were studied, and the seismic response of them was compared. The

solid and diffusive velocity variations are two concepts in the reservoir study that are introduced.

Also, the acquisition configuration influence was checked by surface seismic and well seismic

(Vertical Seismic Profile (VSP) and Cross-Well survey type) that they could generate the different

amplitudes and imaging conditions for the reservoir. Also, I engaged the different rock physics

models (for the mixed type of the phases) to generate the seismic model and images to compare

interpretability of models. Briefly, the main contributions of the thesis addressed the following

subjects:

a- A criterion for PS fold evaluation in different surveys and test randomly located receivers

point in the PS fold improvement.

b- Very long-term plume size estimation with the short-term simulation results.

c- The seismic response of solid and diffusive velocity variation due to reservoir activity.

d- The plume size and the velocity variation in the seismic response.

e- RTM noise reduction in a synthetic 4D seismic study.

4
 f- The acquisition configuration influence on the reservoir study.

g- The rock physics and the influence of different mixing models on seismic results and

interpretability.

h- The statistical condition of the velocity variation in the different fluid mixed models and

relation between the velocity variation’s statistical distribution and seismic inversion for

saturation estimation.

i- The S-wave velocity changes by the density drop in a CO2 injection project. The amplitude

study of SS-waves can provide useful density information.

1.2.1 Software

The following software packages were used to this thesis:

1- Seismic design: OMNI (GEDCO, Schlumberger); this software is useful to study 2D and 3D

seismic surveys with different acquisition parameters.

2- Seismic processing: VISTA (GEDCO, Schlumberger), Promax (Halliburton)

3- Seismic interpretation and data integration, geomodelling: Petrel (Schlumberger)

4- Reservoir simulation: ECLIPSE (Schlumberger), CMG (Computer Modeling Group)

5- Petrophysics: Techlog (Schlumberger)

6- Mapping: Google Earth and Map, AutoCAD Map

7- Log data: Accumap (IHS)

8- Programming: Matlab: This software was used for the following purposes:

a- Seismic Forward Modeling

b- Reverse Time Migration

5
 c- Rock Physics calculation (average methods and fluid substitution)

d- Managing big data transferring between two software (file over 2 GB that is not

manageable by the conventional software) and make suitable format for the next software

9- Numerical calculation: Excel (Microsoft)

For the rock physics study, a series of Matlab codes were prepared. In the seismic modeling

chapter, the main code for the seismic forward modeling and RTM were in Matlab. Some codes

also prepared for importing data from Petrel, CGM, and ECLIPSE to the seismic forward modeling

and RTM imaging code.

1.2.2 Data Resources

The data and information that used in this research belong to CMCRI, Field Research Station

project (FRS) and the dataset was generously available for my research purposes.

1.2.3 Background

In this section, a brief explanation about the available studies and papers are introduced about

Carbon Capture and Sequestration (CCS), CO2 related rock physics, 4D seismic and case studies.

A CCS study always needs a different level of science and engineering for the CO2 capture

and injection compared to hydrocarbon reservoir production. In this project, the following

disciplines have used for the reservoir study and 4D seismic research:

1- The physical properties of CO2 and the reservoir fluid (at the FRS it is brine with low salinity

level).

6
 2- The fluid simulation (a compositional method was used)

3- Rock physics of CO2 injection in the reservoir. It contains the mixed fluids and matrix

condition (for parameters average calculation) and fluid substitution.

4- Seismic response of CO2 injection in the reservoir. The seismic models in the acoustic and

elastic medium condition were generated, and the RTM method was used to image the

acoustic.

5- Seismic time-lapse studies are examed through the difference between the synthetic monitor

seismic data and the baseline data.

There are many papers for calculation of the physical properties of CO2. Span and Wagner

(1996) introduced a new Equation of State for CO2 from the triple-point temperature to 826.85 oC

(1100 K) at pressure up to 800 MPa (8000 bar). Sun (2009) modeled the velocity of CO2 for

temperature down to -10 oC and up to 200 oC and pressure up to 100 MPa. The first reference has

a complex formulation and the second paper used simple equations that are easy to calculate. For

the current research, the Span and Wagner methods were used for the calculations.

In the geomodelling stage, data from the well log and seismic structural interpretation were

available. The geomodelling uses the geostatistical method (included variography and Kriging) for

the best estimation of the properties of each cell. Geostatistics is a part of statistics that the

coordinates of the samples (the spatial distribution of data) were considered. The main root of this

science was the mining industry, and the initial concept was introduced by a South African mining

engineer in 1950s (D.J. Krige). G. Matheron (A French engineer) expanded and formulated this

method and established a branch of statistics that is useful in the mine, earth and atmosphere

science and oil and gas industry as a powerful estimator.

7
 In my prior work reviewing for the fluid simulation and also the WASP project, a black-

oil simulator were used for the reservoir simulation considering PVT parameter tuning

(Hassanzadeh et al., 2008). For the simulation formulation and explanation of the terms related to

reservoir in the research, I used Fanchi (2006). The current project is a very shallow depth reservoir

(very low pressure and temperature), and after checking both simulation methods (Black-Oil and

Compositional) and given the limitation of the simulators, we decided to use the Compositional

Simulator for this research.

In the rock physics study, the main goal is to determine the average physical property of the

reservoir; e.g. Voigt and Reuss (Reuss, 1929) as the upper and lower bound limit with an average

case of them (Voigt-Reuss-Hill) (Hill, 1952). Also, Hashin-Shtrikman’s narrow bounds is a proper

method for a better estimate of property fluctuation in a mixed material. Fluid substitution has

many formulations based on the reservoir condition. Gassmann’s relation at low frequencies

(Gassmann, 1951) is a well-known method for this purpose, and we used the isotropic form of it

for the velocity estimation after CO2 injection. Kuster and Toksöz formulation (1974) is the other

method for the P and S wave velocity estimation for a mixed material (after the fluid substitution

procedure). It uses a long-wavelength first-order scattering theory (Mavko et al., 2009).

There is rock physics research related to the CO2 injection and sequestration. Li et al., (2006)

characterized rock physics properties for a CO2 sequestration study. Kim et al., (2016) did a lab

test and studied the seismic velocity change in a heterogeneous sandstone by the CO2 drainage and

imbibition conjugate with the resistivity measurements. Also, a lab test with a sandstone sample

was done by Alemo et al. (2011) and a field work by Lumley (2010) that showed good

8
compatibility of the rock physics study for the velocity estimation by the VRH average. Also,

the theoretical calculations were compared with our study (Smith et al., 2003).

For the seismic modeling, the acoustic and elastic wave equations are solved by the finite

difference method in MATLAB. The main base of the seismic studies in the research is the acoustic

modeling and migration. The algorithm and formulas in the modeling section and the related code

are from Brekhovskikh (1960), Zakaria et al., (2000) and Zho (2003).

Wave Equation Migration was introduced in 1982 by Whitmore in the 52nd SEG meeting.

Loewenthal (1983) published an algorithm that he called Reverse Time Migration (RTM) based

on a two-way wave equation migration solution. Also, Baysal (1983) published a paper about the

RTM advantage compared with other methods. Levin (1984) also described this approach. In this

research, to describe the methodology of RTM, we used Withmore (2012) and for describing the

noise generated by the RTM algorithm, the research by Khalil (2014) was studied.

Seismic studies for CCS has expanded recently due to sequestration and EOR activities. Raji

et al., (2017) discussed the tomography method for the CO2 monitoring. Previously a very

introductory paper about time-lapse imaging by Full Waveform Inversion (FWI) method was

published for the velocity estimation of a reservoir (Ansari, 2012). The case histories of time-lapse

study for CO2 injection and sequestration shows many successful experiences done by VSP data

(Chadwick et.al. 2009, Geng et. al. 2011, John et. al. 2004, Thomas et. al.2008). A walkaway VSP

data on the SACROC field for EOR purpose tried to estimate velocity changes in the reservoir due

to CO2 injection by RTM and FWI method (Yang et al., 2014).

9
1.3 CO2 sequestration

1.3.1 Introduction

In this section, I present the carbon cycle in the atmosphere from Precambrian to the Recent
12
age. Carbon is an element with atomic number 6 and three natural isotopes that C and 13C are
14
stable and C is radioactive. In the earth crust and atmosphere, carbon is 15th most abundant

element. In chemistry, the main element to generate organic molecules is carbon. It is an essential

atom for the biological structures and life. It also contributes in the non-organic molecules,

minerals, and sediments as graphite, diamond, calcite (limestone), dolomite and coal.

1.3.2 CO2 in the atmosphere

The CO2 concentration in the atmosphere has fluctuated through earth’s history. A full record

of CO2 concentration in the atmosphere from late Precambrian is shown in Figure 1-3. Figure 1-4

shows the CO2 concentration in the atmosphere and temperature in the early Quaternary that was

obtained by ice core data (Petit et al., 1999). This work has shown that the CO2 in the atmosphere

shows a strong correlation with temperature, as shown in Figure 1-4 and Figure 1-5 (Scotese et al.

2002). CO2 is a gas that can change the radiation rate of the heat from the planet and trap the energy

in the atmosphere (greenhouse effect). The records show an increase in global temperature by 0.6

to 0.9 oC since 1906 (Earth Observatory, NASA). Figure 1-6 shows a dramatic increase in the CO2

concentration in the atmosphere due to human activities after the industrial revolution in the 19th

century. After the industrial revolution, the consumption of the fossil fuels (coal in 19th and oil in

20th centuries) increased shown in Figure 1-7. It means more CO2 has been emitted into the

10
atmosphere by the chemical reaction of combustion. The estimation of the CO2 annual

emissions in the world was around 1.5 (Gt) in 2011 (International Energy Administration (IEA)).

Figure 1-3 shows the CO2 concentration and temperature of the atmosphere from Precambrian

to the Recent age. The CO2 had a high concentration in the Paleozoic, and it decreased in the

Silurian eras (as late Carboniferous). The CO2 concentration had been decreased after Jurassic-

Cretaceous border gradually, and it has made a proper environment for mammals to evolve with

the higher rate in Cenozoic.

Figure 1-3: The concentration of CO2 and the temperature of the atmosphere from late
Precambrian to recent. The blue line demonstrates the temperature fluctuations, (Scotese et
al. 2002) (Pagani et al. 2005).

11
 Figure 1-4. Temperature, CO2, and dust concentration in the atmosphere from 400,000
years ago. Temperature has a strong correlation with the CO2 concentration in the
atmosphere and reverses relation with the dust. The loess sediments are a real proof of dust
concentration in the atmosphere during the cold stages (Petit et al., 1999).

12
Figure 1-5. CO2 in atmosphere and temperature from 50,000 years ago, by Vostok ice cores
study (Petit et al., 1999, and joannenova.com.au).

Figure 1-6: The concentration of CO2 in the atmosphere since 1958 to 2015 (Measured at the
Mauna Loa Observatory, https://www.co2.earth).

13
 Figure 1-7: World energy consumption by source (based on data from BP Statistical review,
2014)

Figure 1-4, Figure 1-5 and Figure 1-6 demonstrate more detail about CO2 in the atmosphere

from the early Quaternary, 50,000 years ago, and after 1958.The relation between temperature and

CO2 concentration is very significant and also as related to the Milankovitch cycles (Hays et al.,

1976).

1.4 CO2 sequestration, a method for decreasing CO2 concentration in the atmosphere

CO2 reduction from the biosphere environment include methods and technologies that can

help to reduce the emission of CO2. Carbon Capture and Storage (CCS) (sequestration) in a

geological trap is a well known globally and efficient method for reducing the concentration of

this gas in the atmosphere.The CO2 is injected into a formation for two reasons, a- storage for

environmental reasons and b- for EOR (enhanced oil recovery). Sequestration programs usually

inject CO2 in a reservoir with reliable cap rock and fluids that are mainly brines.
14
 There are 38 undertaked CCS projects (in operating, execute, define, evaluating stages)

with 15 projects that have been undertaken in the world, with3 of them beeing in Canada, as listed

in Table 1-1 (Global CCS Institute). The first attempt for enhanced oil recovery by CO2 injection

in an oil reservoir was in 1972, in Texas. However, the concept of CCS as an environmental

method to prevent of CO2 emission and to reduce the greenhouse gas concentration at the

atmosphere was introduced in 1977 (IEA Greenhouse Gas R&D Programme).

Table 1-1: The large-scale CCS were operated in Canada (Source: Global CCS Institute)

Operation Capture Transport Primary storage

Project name Industry Capture type
date capacity (Mtpa) type type
Great Plains Synfuel Plant Synthetic Natural Pre-combustion
2000 3 Pipeline Enhanced oil recovery
and Weyburn-Midale Project Gas capture (gasification)
Boundary Dam Carbon Post-combustion
2014 Power Generation 1 Pipeline Enhanced oil recovery
Capture and Storage Project capture
Hydrogen Dedicated Geological
Quest 2015 Industrial Separation 1 Pipeline
Production Storage

1.4.1 CO2 trapping mechanism

Many papers have explained geological and reservoir concepts of CO2 sequestration and

trapping options. The gas-water relative permeability hysteresis and trapping efficiency have a

significant role in the reservoir capacity; these parameters were studied in Alberta previously

(Bachu et al., 2013). For example, in the FRS project, the permeability is low, and according to

Bachu (2013) the trapping efficiency is around 50%.

The primary concern about the CO2 sequestration in high volumes is the leakage risk into

shallow aquifers or into the atmosphere that can cause significant damage and environment

15
problem for living creatures and the biosphere. However, some researcher worked on the

methods to decrease the CO2 sequestration risks (Burton et al., 2009).

In a CO2 sequestration process, four mechanisms are known to trap CO2 (Figure 1-8):

1. Stratigraphic and structural trapping: In a reservoir, in the short term after injection,

the CO2 is free phase and can circulate in the reservoir space upward because of the

density and gravitation effect and be trapped under the cap rock.

2. Residual Trapping: some injected CO2 phase is trapped in the pores space and they do

not move by gravitation.

3. Solubility Trapping: Water has a high capacity to dissolve CO2 gas, and a considerable

amount of CO2 will dissolve in the brine right after injection (Eq. 1-1). A common

sample of CO2 dissolved in water is soda water.

4. Mineral Trapping: this stage is the most secure trapping method for CO2. The minerals

in the reservoir reacte with CO2 with the water existence in the aquifer as a catalyzer,

and it can precipitate the gas to a solid phase (as Eq. 1-2 and Eq. 1-3).

The first and second mechanisms are just a physical action; mechanism no. 3 is a physical

action with a partial chemical change (Eq. 1-1) in brine, and the fourth trapping mechanism is an

entirely chemical reaction between carbonic acid and hydrogen ion produced by mechanism no. 3

and minerals (Eq. 1-2 and Eq. 1-3).The chemical reaction of CO2 with water can be explained as:

H 2O  CO2  H 2CO3  HCO3  H 

Eq. 1-1

16
 With quartz and silicates:

SiO2  4H +  2H 2O  Si 4
Eq. 1-2

With kaolinite as a typical clay mineral

Al2 Si2O5 (OH )4  6H   5H 2O  2 Al 3  2SiO2

Eq. 1-3

Figure 1-8. CO2 trapping mechanisms and increasing CO2 storage security over time (Pooladi-
Darvish, 2009).

According to the trapping mechanism and storage stage (as shown in Figure 1-8) in the primary

years of injection, structural and stratigraphic trapping play the main role, but after decades (or

centuries), mineral trapping and solubility trapping are most important and thus CO2 storage in

aquifers will ultimately be secure.

17
1.4.2 CO2 gas injection in the brine

During the injection of CO2 gas in a reservoir with primary brine fluid in, some changes

happen in the brine near the injection well (Hurter et al., 2007):

1- CO2 dissolves in brine, because of some chemical action, this increases the density of the

brine.

2- Dissolved CO2 in the brine makes carbonic acid (Eq. 1-1).

3- The water can vaporize into the CO2 gas; this procedure can decrease the immiscible water

value and increase the salinity of the brine and make salt deposits near the well (dry and salting

out).

The solubility of CO2 in the water is significant, and for brine, the CO2 solubility is a function

of salinity. When salinity increases, the solubility of carbon dioxide decreases. Industrial

simulators do not have an option for the salt or water vapor variation in the reservoir during the

injection. For a detail simulation result, they should be considered in the reservoir’s fluid

simulation procedure.

More discussion about dry-out and salting-out effects are discussed in the reservoir simulation

chapter.

18
Chapter 2. The Field Research Station (FRS) project

2.1 Introduction

This project is a CO2 injection project, led by CMCRI and the University of Calgary. The

project area is located southwest of Brooks City, west of the Newell Lake in southern Alberta. The

geographic map of the project was shown in Figure 2-1. The satellite image of FRS project area

(green square shows the 1*1 km area available for the project) and wells near the project area

(yellow pins) are shown in Figure 2-2. The objective of this research is CO2 injection in the shallow

targets (300 m and 500 m depth) and study the behavior of the CO2 migration, movement, and

leakage and monitor the gas by the seismic method coupled with reservoir studies in the FRS

project. Other geophysical methods (as microseismic, electrical methods and microgravity) will

test for the gas monitoring. In this project, there is an emphasis on the leakage of CO2 and detection

methods.

The CO2 injection procedure into the target needs some initial preparations. The field

operations began in 2014. The preparation set-up included following:

1- A 3D seismic acquisition operation was done in May 2014 as the baseline seismic data.

2- An injection well drilled for the injection purpose (March, 2015).

3- A full log data set was gathered after drilling.

4- Electrical resistivity tomography data were collected from the project area (2015)

5- New seismic surveys (included surface and borehole seismic acquisition) was done after

removing of drilling equipment (July, 2015).

19
 Figure 2-1.The map of project area, it is located in 017-16-W4

Figure 2-2. The satellite image of the FRS, the green rectangle shows the project area
and the yellow points are the well sites with log data (Source: Google Earth).
20
2.2 Geological setting

The project area is in the southern plains of Alberta. The stratigraphic column for the southern

plain is shown in Figure 2-4 (AGS, 2010). The information about the formations in well 10-22

were combined with the adjacent well data and the results are included in Figure 2-5

(Schlumberger, 2016). The formations in the project area include the following:

Glacial sediments: The Quaternary glacial sediments are on the surface. The thickness of

glacial till reach 30 m in the well site.

Belly River Group: The glacial till (Quaternary deposits) overlies the Belly River (BR)

Group. This group is a unit of upper Cretaceous and mid-Campanian stage. So, there are a

significant hiatus between Cretaceous and Quaternary sediments in the well site. This group is

subdivided to the three formations in the southern plain as: Dinosaur Park, Oldman, and Foremost

formations. Figure 2-3 is a picture from the Bow River’s valley near Bow City (near the project

area) that shows the exposure of two upper formations in this group. Entirely the thickness of the

BR Group is 272 m in the well 10-22, but it can reach a maximum thickness of 1300 m follow to

the west. The upper formation in the group is Dinosaur Park Formation with 69 m thickness. The

lithology of the formation is sandstone in the lower level and smaller grain size sediments as

siltstone and mudstone in the upper sections. The reported lithology for this formation is sandy

shale based on the well log data and the drilling report.

The Oldman and Foremost are two other formations in the group. They are detrital formation,

mainly fine grain sandstone in the Oldman and shaly sandstone with coal layers in the Foremost

Formation. The Basal Belly River Sandstone is the base member of Foremost, and it was selected

as the first target for the CO2 injection. It is a regressional shoreline sediment (Hamblin et al.,

21
1996). The target layer is a sandstone with around 14-16% porosity and the permeability

between 0.1 to 2 mD. In the research, the focus for geomodelling and the fluid simulation is on the

Basal Belly River Sandstone.

In the petrophysical study of the 10-22 well, three coal layers were detected above the cap

rock in the Foremost Formation from 285 to 295 m (Figure 2-9).

Figure 2-3.The outcrop of two upper formations in the Belly River group in the Bow River
valley near the Bow City, west of project area and 7.5 km far from the well site (AGS, 2010).

Pakowki Formation: Pakowki is a detrital formation that gradationally underlays the

Foremost Formation (here Basal Belly River member). The lithology is gray mudstone, olive

siltstone, and very fine-grained sandstone. The base of the formation is marked by a thin pebble

conglomerate.

22
Figure 2-4.The stratigraphic correlation and table of formations in the southern plain
(Alberta Geology Survey (AGS), 2010)

23
Figure 2-5. The Stratigraphic chart for the well 10-22 (from well 10-22 drilling data
combined with the adjacent wells data, Schlumberger, 2015)
24
2.3 Petrophysical study and Interpretation

The well (CMCRI COUNTESS 10-22-17-16) is of primary importance for the research as it

is planned to inject CO2 through the well into the shallow target (i.e., Basal Belly River sandstone)

and a wide range of the log data and core studies are available for the well and the injection zone.

As the core pictures demonstrate (Figure 2-11), the target zone is mainly sandstone between

two layers in the up and bottom formed by the smaller grains size detrital sediments that can make

a suitable trap for the CO2 injection. The petrophysical study introduces the main parameters for

the geomodel, including porosity and permeability and the physical properties of the formation.

For the seismic study the P and S-wave velocity and density data were derived from the well log

data, and will be used to generate synthetic seismic images. Some information and parameters as

the mineralogy, the salinity of the brine and the free fluid amount in the injection layer are also

predicted by the well log data.

In the initial log studies, the distribution of P and S-wave slowness are shown as scatter

diagram in Figure 2-6 and Figure 2-7 shows Vp/Vs ratio for the well. In the diagram (Figure 2-6)

the color bar shows the gamma ray content, and the circle size is for the density permeability.The

shale in the rock sample shows a lower permeability, so the blue ellipsoid shows an area with the

high velocity, low porosity, low permeability and shaly zone. As is shown in Figure 2-7, the Vp to

Vs ratio is from 1.8 to 2.6. In the reservoir zone (295-302 m) this ratio is 2.0. Figure 2-7 also shows

the well logs for porosity values estimated by the different methods (seismic porosity (SPHI),

density porosity (DPHZ), TCMR porosity, the average of mentioned three porosity (PHI)) and the

permeability calculated by Nuclear Magnetic Resonance (NMR) by two methods (Timur-Coates

(KTIM) and the Schlumberger-Doll-Research model (KSDR)), neutron porosity and porosities

25
determined by NMR logging. For the geomodeling, I used the average porosity estimated by

all of these methods.The range of the porosities is 15 to 20% and they are available in the Figure

2-7.

Figure 2-6. P-wave versus s-wave slowness in the well 10-22. The color-bar shows the gamma
ray log, and the size of circles are for density porosity. The Vp/Vs ratio is between 1.8 in the
deep layers to 2.6 in the shallow formations.

26
 Depth (m)

Figure 2-7. The well log data for seismic porosity (SPHI), density porosity (DPHZ), TCMR
porosity, the average of mentioned three porosity (PHI), the permeability (KTIM: Timure-
Coates, KSDR: Schlumberger-Doll-Research model), upscaled data for PHI and KTIM, and
Vp/Vs ratio near the injection zone.

27
 Figure 2-8: The Timur-Coated model permeability (mD) vs. density porosity (v/v). The
different statistical populations are recognizable because of the different lithological and
sedimentation condition.

In reservoir studies, the relation between permeability and porosity is a useful relation to

estimate permeability from the porosity. Figure 2-8 shows the relation between porosity (density)

and permeability (KTIM).There is no significant relation between two parameters in the whole

well but it is possible to define a correlation between them in some formations and horizons.

Just there is a linear correlation in the fully logarithmic diagram between KTIM and CMFF

(free fluid porosity (index) from CMR) (Figure 2-10).

28
Figure 2-9: Petrophysical interpretation result for the injection zone and around. The
minerals type was characterized by the well logs.

29
 Figure 2-10. There is no relation between KTIM and other log data. Just a meaningful
relation is between CMFF (free fluid porosity (index) from CMR) and KTIM. The coordinate
is full logarithmic scale. For more information see Appendix A.

Figure 2-9 is a full interpretation of the well log data near the injection zone (BBR). The most

useful part of the interpretation of the reservoir and seismic study is the lithological and

mineralogical content of the target.

2.4 Core study

The core samples make it possible to do some measurements directly on the rock sample. In

this section, the mineral discrimination study based on spectral gamma ray (combined with log

data), and permeability and porosity of the rock sample were introduced.

Figure 2-11 shows core samples obtained from the core in the injection zone. The injection

zone will be in depth of 295-302 m in the Basal Belly River Sandstone.

30
 Sandstone
Silt and Sand

Figure 2-11. Core photo from the injection zone (sandstone) (taken by Schlumberger).

31
2.5 Lithology and Mineral study of the Belly River sandstone

The mineralogy and lithology information were extracted from the well log data. Some useful

well logs for mineralogy and lithology analysis are:

Mineralogy: gamma ray, density, neutron porosity, neutral radioactivity, spectral gamma ray,

nuclear magnetic resonance, acoustic log, caliper.

Lithology: Laterolog, induction, Microlaterolog, SP log, electromagnetic propagation, gamma

ray, density, neutron porosity, neutral radioactivity, spectral gamma ray, nuclear magnetic

resonance, acoustic log, caliper.

The main goal of mineral study and discrimination was for calculating the bulk modulus of

minerals in the fluid substitution process. In this part, the available well logs were used to

distinguish the mineral types in the reservoir zone.

Table 2-1: The mineral types in the reservoir zone by the well log data analysis (from Figure
2-12)

Mineral discriminated Fraction

by well log analysis
Quartz 40%
K Feldspar 4%
Albite 8%
Kaolinite 15%
Chlorite 7%
Illite-Smectite 11%
Siderite 15%

32
Figure 2-12: The result of the mineral types discriminated by the well log data (Kirk
Osadetz, personal conversation)

33
2.6 Core analysis

The core study results make it possible to calibrate the well log data (permeability) with the

directly measured data from the rock samples. The calibrated permeability uses core data and well

log data, and it was used to make a geomodel (Chapter 4). The procedure for calibration for the

well data was described by J. Dongass (2016). The final data for the reservoir zone in geomodel is

demonstrated in Figure 2-13, and the used data are from the core study in Table 2-3.

Table 2-2.Effective porosity and water and gas saturation in the core

Core Depth As Received As Received Effective Dry Gas Fielled Hydrocarbon Effective Effective Effective Gas Effective Oil
Bulk Density Grain Density Grain Density Porosity (% of Fielled Porosity Porosity (% of Water Saturation (% of Saturation (% of
(gr/cc) (gr/cc) (gr/cc) BV) (% of BV) BV) Saturation (% of PV) PV)
293.34 1.401 1.414 1.469 0.96 0.96 12.35 PV)
92.2 7.8 0
294.37 2.381 2.455 2.57 2.99 2.99 10.11 70.46 29.54 0

Table 2-3.The measured total porosity and water and gas saturation in the core

34
Figure 2-13: Before and after log-to-core calibration of K_INT data for the injector
Well (10-22), from Swager (2015).

35
Chapter 3. Seismic design for 3C-4D propose in the FRS

3.1 Introduction

The primary objective of this section is to evaluate the 3D-3C seismic survey for time-lapse

and reservoir studies to monitor CO2 injection and map the underground layers and structures. The

main targets for the seismic acquisition are two porous layers for the CO2 injection in 300 m and

500 m. Technically two stages were done for seismic design: The first part is data gathering and

analysis results for velocity functions and desired and dominant frequency content of targets

(shallow and deep) and the second part is the parameter estimation for preventing spatial aliasing

and to check the best acquisition parameters for proper horizontal and vertical seismic resolution.

For the bin size and migration aperture estimation, constant and linear velocity methods were

considered. Two seismic surveys were introduced, and their attributes (fold map for PP and PS

data with different offset, offset and azimuth distribution) were compared. Finally, for improving

PS fold, we tested a random receiver distribution. The concepts and formulas for this part are from

Cordsen (2000), Vermeer (2002), Stone (1994) and Liner (1997,1999,2004).

3.2 Background information

For a regular onshore seismic design project, consideration related to the area’s geological

condition and the surface access that can make acquisition footprint are important. We introduce

the required data and information for a successful design as the following items:

1- Geology of area (surface, subsurface, and structural condition as layers’ dip angle)

36
 2- Terrain conditions (topographic, permit)

3- Frequency contents (Max and dominant) in the targets and required resolution

4- Velocity and velocity as a function of depth

5- The objective of acquisition (image, reservoir study)

6- Full fold Image zone for structural or reservoir studies to estimate acquisition boundary and

area by calculating migration aperture and fold taper

7- Seismic data (row shots for a better frequency analysis and sections for interpretation and

evaluation and both for controlling quality of data and problems)

8- Technical part and existence technology (recording system)

9- Financial conditions and limitations

3.3 Considerations for a 4D seismic design and acquisition

For a seismic design in a conventional 3D project, geophysicist consider important points as:

1- Uniform PP fold in the image area

2- Full fold on the target horizons

3- Appropriate bin size to prevent aliased data acquisition

For a 4D seismic design, we need to repeat the baseline seismic acquisition geometry. To

prevent for acquisition error and footprints, in the monitor surveys, the CMP points should be

exactly at the baseline surveys CMP locations. This requires that the source and receiver points

are exactly in the same place as the baseline seismic acquisition survey. So, for a 4D study in a big

37
field, the seismic data should be acquired with the same survey and parameters as a baseline,

and for the small fields and surveys, the geophones can be cemented in place.

3.4 Targets

As mentioned previously, the study area is in the southern Alberta plain. According to the well

tops and existing old 2D and 3D seismic data, subsurface layers are almost flat in the target

zone.The shallow target is the Basal Belly River sandstone and Pakowki Formation at 295-302 m

depth (Table 3-1), mid-target is Medicine Hat Formation (~500 m) (the second injection option),

and the deep target is the top of 2WSS (or Base_Medicine Hat Formation) at ~700 meters. A proper

parameter design should guarantee full fold data and appropriate azimuth and offset distributions

at the target horizons.

Table 3-1.The Targets were considered for the seismic design purpose.

Depth (m) Formation Top Period

189.74 Top of the available well log data

295 Basal Belly River Sandstone

Cretaceous

Target

302 PAKOWKI
Upper

357 MILK RIVER

441.5 COLORADO SHALE
478.5 MEDICINE HAT SANDSTONE
SECOND WHITE SPECKLED
711
SHALE
785 FISH SCALES

38
3.5 Frequency content

For bin size estimation and design, the maximum and dominant frequency of seismic data

from previous seismic acquisition surveys (VSP,2D or 3D) should be analyzed. The relation

between frequency (f), dip angle (θ), interval velocity (V) and bin size (B) to prevent aliasing

phenomena in spatial sampling is:

Vint
B
4 f max sin 
Eq. 3-1

There are many old 2D and 3D seismic surveys in the area as used for the frequency analysis

in Figure 3-1. According to frequency content analysis at the shallow and deep targets, as shown

in Figure 3-1, the dominant frequencies for the target formations are between 30-60 Hz, and the

maximum frequency is 80 Hz.

Figure 3-1. Frequency analysis: for the shallowest target (A) and the deepest targets (B)

39
3.6 Velocity-depth function

Well log data was used for compressional and shear wave velocity profiles, and Figure 3-2

shows the general relation between depth and velocity. For bin size and migration aperture

estimation, it is possible to use constant or linear velocity function which can decrease the

migration aperture size and acquisition area so that method can optimize the acquisition cost.

3.7 Bin size

Appropriate bin size can guarantee a data set acquisition without aliasing problem; small bin

size can ensure unaliased data, but also can decrease S/N ratio (Cordsen et al., 2000). Here we

directly use anti-aliasing bin size formula (Eq. 3-1) for the constant and linear velocity functions.

The project area is situated on a flat plain and no subsurface structure, and layers have a gentle

dip angle less than 2 degrees. Figure 3-3 demonstrates the calculation result for the bin size by

constant and linear velocity methods. The linear function for the velocity is calculating by the well

log data as Figure 3-2. The bin size estimation by linear velocity method uses the velocity as a

function of depth. The linear velocity concept is also useful in the migration aperture calculation.

The dip angle is also relevant to migrating diffracted energy, even though the layers are nearly flat.

40
 a b

Figure 3-2. A linear function for velocity is used for calculation of bin size and migration
aperture. Velocity function for FRS project regards to well log data (wells 11-22-17-16 (a) and
7-22-17-16 (b) and 10-22-17-16 (c)) is V=V0+kz=2650+z.

41
Figure 3-3. Bin size for the shallow target with 80Hz max frequency (left) and the deep target
with 65Hz (right).

3.8 Box size and geometry

The box size (receiver line interval*shot line interval) and geometry can introduce the LMOS

(largest minimum offsets) concept as an important parameter for the shallow target acquisition

design. As mentioned, the target depth is from 300-700 m, and for acquiring data with the suitable

fold on the target depth, LMOS should be smaller than the first target depth, because it results in

a no data zone equal to the LMOS two-way time. Another problem that decreases fold at shallow

depths is NMO stretch and mute so for the project, and this should be considered in the design.

3.9 Design option

Analysis and parameters calculation in the last section, and necessity to have a high-resolution

seismic profile for the research purpose, led us to suggest a dense seismic survey. For the design

quality control, the option was loaded and analyzed in OMNI (design and survey control software).

42
3.9.1 The acquisition parameters

The acquisition parameters are listed in Table 3-2. This option has a dense acquisition pattern

in the middle part as shown in Figure 3-4. In this figure the red spots are shot points, and the red

dots are receivers. The bin size for this option is 5m*5m and receiver and shot line intervals are

both 50 m in the central part of the survey (500*500 m) and 100 m in the outer parts. The nominal

fold is up to 185 in this central part. The acquisition parameters was designed for a high-resolution

image for the shallow reservoir.

Table 3-2. The acquisition parameters.

Parameters Main Central part

Bin size (m) 5 5
Receiver interval 10 10
Receiver line interval 100 50
Shot interval 10 10
Shot line interval 100 50
Total Survey area 1000*1000 500*500
Maximum Offset 1407
minimum offset 7
Largest minimum offset (LMO) 134
The highest fold (PP) 185
Maximum inline offset 1000
Maximum xline offset 1000
Aspect ratio 100%
Total shots 1434
Total live geophones 1400

43
Figure 3-4.The survey geometry. The red points demonstrates shot points, and the blue ones
are receivers. There are a dense shot and receiver points in the mid of survey with dimension
equal to 500*500 m.

Figure 3-5. A. The fold map. The highest numerical range for the fold is 220 for both shown
by the red color in the scalebar. The yellow circle shows the fold range between 30-40. B. The
fold map for offset 0-700 m.

44
 Figure 3-6. A. The azimuth distribution in the full fold zone. It shows a proper azimuthal
coverage; the azimuth depends on the shorter offsets. B. The offset distribution in the full fold
box. The acquisition covers full offset ranges.

Figure 3-7. A. Azimuth distribution in the target range (0-700 m offsets). B. Offset
distribution in the full fold box for 0-700 m.

45
Figure 3-8. A. Offset redundancy, the target zone demonstrated by the black lines. B. Azimuth
redundancy, the number of traces that fall in each section; gaps indicate missing azimuth

Figure 3-9. A. Histogram of Fold, the numbers of bins that fall in each range of fold values. B.
Histogram of Offset, the number of traces that fall in each range of Offset values

46
Figure 3-10. A. Histogram of azimuth, the number of traces that fall in each range of azimuth
values. B. Offset versus azimuth diagram.

Figure 3-5 shows the fold map. Figure 3-5.A. is a nominal fold map that covers all acquisition

offsets and azimuths. The yellow circle shows the fold range between 30-40 and out of this circle,

the fold decreases less than 30. The fold at the well point is over 200. This fold distribution is

normal for an area without any complexity in the geology with horizontal layers with no diffraction

events. Figure 3-5.B. shows the fold for source-receiver offsets from 0 to 700m. The fold

distribution in this range that can guarantee a suitable image at the target depth.

Figure 3-6 demonstrates offset (A) and azimuth (B) distribution. The distribution diagrams

show a perfect offsets from 64-1407 m (they are LMOS and maximum offset) and azimuths from

0-360 degree.The offset and azimuth distribution are suitable for the shallow target range (Figure

3-7).

The offset and azimuth redundancy diagrams were determined for each bin of the survey

(Figure 3-8). They show excellent response in the target range for offset redundancy (A) between
47
the black lines with a perfect azimuth distribution (B). Figure 3-9 shows a statistical result

for the fold versus bin count (left) and offset range versus trace count. As seen in the diagram, the

offset coverage for the shallow to deep targets are suitable.

Figure 3-10 shows azimuth versus trace count and offset. Both diagrams demonstrate a proper

distribution for the azimuth that can make a suitable database for amplitude variation with azimuth

(AVAZ) study.

3.10 PS survey design

The base of PS designing is the concept of CCP (Common Conversion Point). In this chapter,

a non-asymptotic method is used for determining fold and calculating other attributes. For the

design attributes for the PS acquisition, a flat target is considered at 400 m depth.

Figure 3-11. The PS fold map (non-asymptotic method) for a target in 400 m depth, Vp
to Vs ratio is considered equal 2.

48
Figure 3-12. A. The PS azimuth distribution in the full fold boxes. B. Offset distribution for
the PS wave.

Mid-Core bins

Offset with proper

fold coverage

Figure 3-13. Offset redundancy for PS wave, the number of traces that fall in each
section; gaps indicate missing offsets.The curve of boomerang shape distribution is a function
of p to s wave velocities ratio.

49
Figure 3-14. Azimuth redundancy for PS wave, the number of traces that fall in each section;
gaps indicate missing azimuth.It shows a proper azimuth distribution.

The fold is concentrated mainly in the dense, central part and fold map reveals that just 50%

of acquisition area will reach to the fold more than 30 (Figure 3-11). The maximum nominal fold

is 185, and the mid-core high-density acquisition zone guarantees high fold as >100 for the offset

0-700 m. Because aspect ratio is 100% and box and patch are symmetric, azimuth and offset

distribution maps show excellent design parameters. Azimuth-offset histogram indicates a good

coverage for offsets less than 1 km and 360-degree azimuth, (also there are a lack of data for some

azimuth for offsets greater than 1 km, but this part is not in our interest zone). The offset

redundancy diagram, as expected, shows a zigzag pattern that is due to the orthogonal geometry.

It shows a high redundancy for the offsets 300-700m.

For calculating PS fold, OMNI uses non-asymptotic PS conversion point between shot and

receiver. It considers a flat target layer that is 400m for the project. The PS fold and offset; azimuth

distribution maps (Figure 3-11, Figure 3-12, Figure 3-13 and Figure 3-14) show a good design

attributes for the PS data acquisition.

50
3.11 A criterion for fold distribution

The fold is an important parameter in seismic design. Sufficient and uniform fold distribution

in a survey is the priority that geophysicists deal with it in the parameter design. Prevention of

striped fold pattern or lack of fold because of acquisition field barriers in PP-wave acquisition and

smooth fold in PS acquisition are two challenges for designers. For an optimum fold condition and

distribution in a 3D seismic survey, mathematically it can be described as a parameter with the

lowest possible variance. For a discrete parameter as xi, the variance (Var) and the expected value

or average (µ) in a vector are demonstrated as Eq. 3-2 and Eq. 3-3:
𝑛
1
𝑉𝑎𝑟 (𝑥) = ∑(𝑥𝑖 − 𝜇)2
𝑛
𝑖=1

Eq. 3-2
Where:
𝒏
𝟏
µ = ∑ 𝒙𝒊
𝒏
𝒊=𝟏

Eq. 3-3
For a 3D seismic survey with totally m*n bins in x and y directions, (as two-dimension

matrices), The fold in a i, j th cell is demonstrated by Fij. So the variance of the fold for all bins

can explain as following formula:

𝑚 𝑛
1
𝑉𝑎𝑟 (𝐹) = ∑ ∑(𝐹𝑖𝑗 − 𝜇)2
𝑚𝑛
𝑖=1 𝑗=1

Eq. 3-4
And the average can describe as Eq. 3-5.

51
 𝑚 𝑛
1
µ= ∑ ∑ 𝐹𝑖𝑗
𝑚𝑛
𝑖=1 𝑗=1

Eq. 3-5
Fold taper: in traditional design, fold taper is the area out of full fold zone. Formerly, the area

out of full fold zone does not have a suitable fold amount for quantitative and qualitative

interpretation. Now with improving acquisition techniques and increasing fold, geophysicists can

work in the acquisition marginal zone because of reasonable fold value around the full-fold region.

In a seismic survey with the high-fold acquisition, a new definition of fold taper is used and fold

less than desired for interpretation will be considered as fold taper. In the current project, the

suitable fold is >30 and bins with the fold less than 30 have been deleted for statistical analysis

and variance calculations.

3.12 Improving PS fold coverage: randomize pattern

In this section, I attempt to improve the PS-fold coverage by changing the acquisition

geometry. The patterns tested have randomized source or receiver points with a different

configuration, and the criterion is the introduced variance test.

For the study purpose, I checked 12 different patterns as presented in Table 3-3. For each

model the following variation were evaluated:

1- A movement direction was selected for each geophone. It can be a radial movement

or in the cross-line or Inline direction.

2- The displacement distances are equal a bin size to half of receiver line interval.

52
 3- The PS fold map was generated for each pattern and the high PS fold (or PP full

fold area) zone selected for variance test.

4- The result of variance tests are demonstrated in Table 3-3.

Figure 3-15.A sample of the suggested regular survey (rectangular box) and
randomized receivers pattern.

Figure 3-16. The fold map for regular rectangular box pattern (left) and randomized receivers
pattern (right)

53
 Table 3-3.The variance test result for 12 different acquisition pattern.

Figure 3-15. shows the survey configuration of randomly located receiver points up to half

receiver line interval moved from the standard rectangular pattern. Figure 3-16 is the fold

distribution map for the rectangular pattern and random pattern (Figure 3-15). The variance of PS-

fold is 142.1 for the regular survey. Apparently, pattern number 11 (Figure 3-15) with cross-line

movement direction and half receiver line interval displacement can improve the PS fold

distribution.

54
Chapter 4. The baseline seismic data, interpretation and geomodel

development

4.1 Introduction

The seismic acquisition parameters and the calculation method were introduced in the last

chapter. Design option was approved and used for the 3D-3C data acquisition as baseline data. As

mentioned in the introduction, the seismic field work was done in the summer (May) 2014 by

Tesla Exploration. The recorded data sample rate was one ms. The source used was two mini

vibrators with a linear sweep from 8 to 150 Hz and two sweeps of 16 s length per each point.

INOVA-7 3C analog geophones was used as receivers in the survey.

The data were processed for PP-wave in two different versions with 2 ms sampling interval

by CREWES (Isaac, 2015) and 1 ms sampling by Sensor Geophysical. Also for the interpretation

and inversion purpose, a PS wave data processing was done in 2015 by CREWES. Post-stack

migration method was applied for the both processing flows.

4.2 Field acquisition

The field is almost flat and accessible by truck and vibroseis. The only limitation are two

pipelines that passing southwest to northeast as shown in Figure 4-1. For HSE purposes there is a

25 m setback from source points to the pipelines. The fold map and azimuth and offset distribution

do not show any problems in fold or azimuth and offset distribution so we do not expecting any

footprint of the field acquisition. The acquisition attributes for the actual acquisition are shown in

55
Figure 4-2 (PP fold), Figure 4-3 (offset and azimuth distribution) and Figure 4-4 (PS

fold).The results are very similar to the design attributes that described in the previous chapter.

Figure 4-1. The seismic survey map of FRS project. The blue points are the receivers and red
shows the sources. Two pipelines caused some change in the shot point coordinates. (Satellite
image source: Google Earth)

56
Figure 4-2.The PP-wave fold map for (a). total nominal fold and (b). 0-300 m source-receiver
offsets.

Figure 4-3. (a). Azimuth distribution and (b). Offset distribution

57
 Figure 4-4. The PS-wave fold map for the target in the 300 m depth.

4.3 LMO effect and fold taper

Always there is the fold fluctuation came from the acquisition geometry in a seismic survey

according to the layout limitation, seismic design, and processing flow. The fold taper and

migration aperture effect in the outermost area of a seismic survey; e.g. LMO effects the very

shallow target and the mute function effect in the deeper targets (Figure 4-5).

In the field, there are two pipelines that because of setbacks, the shot points were removed

from this zone and some extra shot points were added to compensate for the dropped shots (Figure

4-1).

58
 Figure 4-5.The low fold zone and acquisition on the acquired seismic data.

4.4 Seismic data resolution and coherency

Parameters was designed for a high-resolution dense central area that covers 500*500 m with

source and shot line interval equal to 50 m. We checked the influence of the dense survey for data

quality. For this purpose, we processed the seismic data, and in the second case, without the dense

central area (Figure 4-6). For this purpose, all CMP points related to the dense zone were removed

in the processing stage.

The results are shown in Figure 4-7 for PP-wave (inline no 101, that is in the middle of survey,

an east-west line passed over the 10-22 well). The red rectangle shows the location of the dense

source and receiver zone. Comparison of the results shows a better coherency in the shallow targets

(750 ms and including the reservoir) that is pointed out by the green rectangle. The red arrows

identify reflectors that have improved coherency with the denser acquisition grid. For a successful
59
reservoir study, high resolution and high fold acquisition are vital for discriminating the small

changes in reservoir CO2 saturation. As we will see latter in the seismic modeling and imaging

chapter, the resolution and coherency for the reservoir layer are crucial especially in the early years

of the gas injection.

Figure 4-6.The configuration of acquisition were used for the processing. Left survey shows
and right is the same survey after eliminating the dense center region.

60
 Figure 4-7. The result of processing for the acquired PP seismic data in (left) and after
eliminating the dense source-receiver CMPs (right).The arrows show the main differences
between two seismic sections.

4.5 Seismic PP and PS-wave data after processing

The acquisition method was 3D-3C, and so raw data of both PP and PS set were available.

Figure 4-8 shows the PP and PS seismic data cubes and Figure 4-9 and Figure 4-10 illustrate 2D

sections for the PP and PS-wave data that used for conventional analysis and the structural

interpretation and geomodel development. The data quality is excellent for quantitative and

qualitative interpretation. The two-way time conversion between PP and PS seismic sections is

illustrated in Figure 4-11, that was calculated for the P and S-wave slowness logs in the 10-22

well.

61
Figure 4-8. The seismic data, A. processed PP-wave data, B. processed PS wave data

62
TWT (ms)

Figure 4-9. PP-wave migrated seismic section on inline 41

TWT (ms)

Figure 4-10. PS-wave migrated seismic section on inline 41

63
Figure 4-11.The TWT conversion diagram between PP and PS seismic sections. The injection
target at 233 ms for PP time and 364 ms for PS time. It was calculated from the P and S-wave
slowness logs. The velocity for no well log data zone calculated by the seismic analysis.

4.6 Seismic interpretation: The phantom horizons methods

The geometry of layers and faults are key inputs for the geomodel frame. In this chapter, two

goals for the FRS project from the seismic interpretation were:

1- The geometrical form and discontinuities of the layers and formations (structural

interpretation and fault study).

2- Interpretation of the homogeneity of the sediments in the reservoir horizon around the

injection well.

64
 In the seismic interpretation, we interpolate the well tops and log data between the wells

with acceptable accuracy and geometry. A common structural interpretation framework is shown

in Figure 4-12. The fault study was done by the seismic qualitative interpretation and attributes

study. The selected horizons for the detailed interpretation and geomodelling are shown in Figure

4-16.

The well tops are the first information package for the layers’ geometry assessment, and

seismic data interpretation can reveal a better accuracy of the geometry in comparison to a well

top interpolation method. The seismic data will help to find formations fluctuations in the area and

mapped the discontinuities as faults.

For a better estimation and accuracy, intra-formation layers were also interpreted as phantom

horizons. This information was used for making surfaces of the formation, sub-formation, layers,

faults and fractures in the geomodel.

4.7 Well ties

The synthetic seismogram is a tool to correlate well tops with the seismic data. For reflectors

match between synthetic seismogram and seismic data, a wavelet was derived from the seismic

and the well log data (Figure 4-14 and Figure 4-15).The seismic data are in the time domain, and

the well log data are in the depth domain, so to produce a synthetic seismogram, we need a depth

to time converter that is coming from velocity log, check shot data or velocity analysis in the

seismic processing or interpretation stages. The overburden thickness without well log data in the

well (10-22-17-16) is 225 m, so for the seismic analysis, the P-wave velocity in this was selected

2550 m/s.

65
 Finally, the selected wavelet was convolved with the derived reflection coefficient from

the sonic log to generate the synthetic seismogram in the well site. The similarity of the

seismogram and the migrated seismic data around the well, with the formation tops will present

the markers for the interpretation purpose.

The following routine was used for generating a synthetic seismogram and well tie (also as

explained in Figure 4-13):

1- Extract wavelet from seismic and well log data

2- Generate the reflection coefficient, acoustic impedance, and synthetic seismogram

3- Time shift and match events

4- Stretch-squeeze the synthetic seismic data

For the interpretation, the reservoir layers (Basal Belly River sandstone) mapped (Figure

4-17, Figure 4-18 and

Figure 4-19) and the other layers interpreted as phantom layers to an accurate analysis and

geomodel structure and geometry.

66
Figure 4-12. Flowchart that shows the simplified stages of work for a general structural
interpretation.

Initial well log construction

(P-wave, S-wave and Density)

P-P traveltime computation/

synthetic calibration
using individual wells
Estimate wavelet
Estimate wavelet
using all wells

Convolve with extracted wavelet

Calibrate traveltime &

QC check shots

Figure 4-13: schematic of procedure used for well to seismic calibration

67
 Figure 4-14. Ricker wavelet and the synthetic seismogram from well log data against
processed seismic data.

Figure 4-15.The wavelet calculated by extended white method for the Sensor processed
data set. The reflectors correlation is acceptable in the 200-300 ms range.

68
 XLine 101
XLine 101
IL 101 101 101 101 101 101 101 101 101 101 101
XL 1 21 41 61 80 100 120 140 160 180 200

0XL

-100

-200
TWT (ms)

-300

-400

-500

Figure 4-16. PP seismic interpretation in time domain including the main formations and
-600
phantom layers (time domain). The line is passing of the main well.

-700

10-22
-800

-900

-1000

-1100

-1200

-1300

-1400

-1500
Figure 4-17. The top Basal Belly River sandstone as the top of the reservoir in the
seismic cube in time domain.

69
 420000 420200 420400 420600 420800
5589800

5589800
-232.5
-235
5589600

5589600
-235
-2
3
7.5

-232.5
5589400

5589400
-230
5589200

5589200
-237.5

-23
7.5
5589000

5589000
-235
5588800

420000 420200 420400 420600 420800 5588800

Near Top Reservoir 0 50 100 150 200 250m Elevation time [ms]
Canada Contour inc -228.50
1:6796 -229.00
0.5 -229.50
FRS Time Domain -230.00
-230.50
-231.00
Near Belly River Date -231.50
12/30/2016 -232.00
Scale D.Nowroozi -232.50
-233.00
1:6796 -233.50
-234.00
-234.50
-235.00
-235.50
-236.00
-236.50
-237.00
-237.50
-238.00
-238.50
-239.00

Figure 4-18. Time structure of the top of the Basal Belly River sandstone.
70
 420000 420200 420400 420600 420800
5589800

5589800
-2
9 5
5589600

5589600
-300

-295
5589400

5589400
90
-2
-3
00
5589200

5589200
5589000

5589000
-300
5588800

420000 420200 420400 420600 420800 5588800

Near Top Reservoir 0 50 100 150 200 250m Elevation depth [m]
-289.00
Canada Contour inc 1:6796 -290.00
1 -291.00
-292.00
FRS Depth Domain -293.00
-294.00
Near Belly River Date -295.00
-296.00
12/30/2016
-297.00
Scale D.Nowroozi -298.00
1:6796 -299.00
-300.00
-301.00
-302.00
-303.00
-304.00

Figure 4-19. Depth structure of the top of Basal Belly River Formation.
71
4.8 Attributes study

Seismic attributes are parameters calculated from the seismic data based on time, amplitude,

frequency and attenuation (Sheriff, 2002). They can reveal the structural properties, reservoir

parameters or discontinuities and faults. In the next section, the attributes were used to certify any

faults and the discontinuities and interpret the homogeneity of the reservoir layer.

4.8.1 Generic Inversion

Generic inversion is an attribute that correlates well data with seismic data to generate P-wave

slowness, and it is demonstrated in Figure 4-20 for the seismic cube and Figure 4-21 for the top

horizon layer. This attribute can show the channels, changes in the lithology, faults and dense

fractured zones. However, in this integration, Figure 4-21(the slowness in the BBRS) shows a very

smooth change in the reservoir layer near the well.

72
Figure 4-20.Genetic Inversion (time slice is 235 ms near the top of reservoir)

Figure 4-21.Genetic Inversion on the top reservoir surface

73
4.9 Fault detection attributes

Using single attribute and a combination of them were also used for possible fault and fracture

recognition. The primary attributes used for fault detection were 3D curvature, variance and chaos

attributes, which will show lineation of geological events some of which can be associated with

the existence of faults. For accurate estimation, a combination of attributes was also used, as

demonstrated in Figure 4-22. Software such as Petrel has some tools to assist extracting and

mapping possible faults using introduced attributes.

Figure 4-23 to Figure 4-30 show the results of multi-attribute on the seismic data for fault

detection.

Seismic data

Structural
Smooth

Variance Chaos

Ant Tracking

Figure 4-22. Fault detection using the multi-attribute method.

74
4.9.1 Structural smoothing

It can eliminate a local noise from the data set; through the use of a mean or median filters.

Figure 4-23 shows overall pattern in the smoothed amplitudes for a time slice near the BBRS top.

4.9.2 Dip Deviation

This attribute maps edges and truncations, and it is useful for identifying faults (Figure 4-24).

The attribute does not show any truncation events around the injection well.

4.9.3 Chaos

The chaos attribute can be described as Eq. 4-1 (Randon et al., 2000) as:

22
cx  1
1  3
Eq. 4-1

Where λ1 and λ2 are the eigenvalues of the Gradient Structure Tensor (GST) matrices.

Note that if λ1 >> λ2, the coherence is high and cχ goes to –1. If λ1 ≈ λ2 ≈ λ3, cχ goes to 0. Finally,

if λ1 ≈ λ2, but λ3≈ 0 [Bakker’s (2003) lineament attribute], cχ goes to +1.

The result of chaos attribute on the structural smoothed cube are shown in Figure 4-25, for a

time-slice of the mid reservoir level.

4.9.4 Variance

The variance attribute is another way to search faults. This attribute measures the

dissimilarity of the seismic data. Figure 4-26 shows the variance attribute on the FRS seismic

75
data. For a better result, a combination of attributes included variance were used for fault

recognition.

4.9.5 Ant tracking

This attribute is useful for fault detection. The ant-tracking method was introduced and

algorithm organized by Randen et al. (2001). By the algorithm, the coherency of the seismic data

is probed by an iterative scheme (Chopra et. al, 2007). The input data for the ant tracking attribute

are variance or chaos attributes. These two attributes can present an accurate result compared to

the seismic data after ant tracking. Figure 4-22 shows the usage of multi attributes methods for the

fault detection.

Figure 4-27 is a smoothed-chaos-ant tracking attributes study on the seismic cube (Figure 4-29

shows the BBR surface). Figure 4-28 shows smoothed-variance-ant tracking on the seismic data (

Figure 4-30 shows the BBR surface). There is no evidence of any discontinuities near the 10-22

well at the reservoir level.

76
 10-22

Figure 4-23.Smoothing attribute is the first step of the multi-attribute method for faults
recognition.

10-22

Figure 4-24.The single attribute study (dip deviation) for fault recognition.

77
 10-22

Figure 4-25.The chaos attribute applied to the smoothed data cube.

10-22

Figure 4-26. The variance attribute applied to the smoothed data cube.

78
 10-22

Figure 4-27. AntTrack attribute applied to the Smoothed Chaos data cube.

10-22

Figure 4-28. AntTrack attribute on the Smoothed Variance attribute result

79
 Figure 4-29.A multi-attribute map (Structural- Chaos – Ant track) for the fault
discrimination.

Figure 4-30. A multi-attribute map (Structural - Variance – Ant track) for the fault
discrimination.

80
4.10 The velocity model

The velocity model generated was based on the seismic velocity analysis and well sinic log.

For this purpose, the log of velocity in the 10-22 well defined as a first-degree linear function

(Figure 4-31). The velocity over 225 m is calculated by the analyzing the first breaks from the

seismic data (Isaac, 2015). Based on these picks, the velocity from the ground surface to 225 m

depth found to be 2525 m/s.

3600

3500

3400

3300

3200

3100
p wave velocity (m/s)

y = -1.295x + 3763.6
R² = 0.2156
3000

2900

2800 y = 1.5834x + 2300

R² = -0.033 y = -1.3x + 3200
2700
R² = 0.1039
2600

2500

2400

2300

2200
220 270 320 370 420 470 520
Depth (m)

Figure 4-31.The P-wave velocity in the well 10-22 with simplified gradients for time to
depth conversion

81
 Figure 4-32 shows the time to depth conversion (or vice verse) for the PP, PS, and SS-

wave arrivals calculated from the well log data. This data is used in the next section to change

the domain from time to depth for building the geomodel.

Depth to time conversion for pp, ps and ss

wave
TWT (pp) TWT (ps) TWT (SS)

850
800
750
700
650
600
550
500
TWT (ms)

450
400
350
300
250
200
150
100
50
0
225 250 275 300 325 350 375 400 425 450 475 500 525 550
Depth (m)

Figure 4-32.The time to depth conversion for the PP, PS and SS wave in the well 10-22.

4.11 Geomodel of the project area

A geomodel contains physical properties extended in 3D encompassing the reservoir. In a

geomodel, the reservoir divided to the cells as a 3D matrix that oriented by the formation geometry.

The model can be generated mainly based on well log data and seismic interpretation according to

statistical or geostatistical analysis. For a better estimation, the results from qualitative
82
interpretation of seismic data (like horizons and faults) and quantitative interpretation (pre-

stack or post-stack inversion results) can be used.

The procedure to make a geomodel is shown in Figure 4-33.

Figure 4-33: The procedure for producing a geomodel in the FRS project.

In the current research, the geomodel was used for three purposes:

1. Reservoir fluid simulation; the permeability (x, y, z directions) and porosity were

determined.

2. Fluid substitution effects on the elastic modulus.

3. Seismic modeling and imaging (velocity and density or acoustic impedance).

83
 Figure 4-34 shows the upscaled well log data in the initial stage of the geomodelling for

the FRS project.

Figure 4-34. The upscaled well log data for building geomodel in well 10-22 before
permeability calibration.

84
4.12 Introduction to Geostatistics:

4.12.1 Variogram:

A variogram is a tool to analyze the structural form of the spatial distribution of the variables

in a statistical population. Eq. 4-2 and Eq. 4-3 are general forms of variogram formula.

var Z (s)  Z (u )    Z (s)  Z (u )  f (s, u )dsdu

1 1 2
 (h ) 
2 2
Eq. 4-2

2
1 N (h)
ˆ (h )   z(si )  z(si  h)
2 N (h ) i 1
Eq. 4-3

γ (h): Semi-Variogram

Z(s): The variable (can be porosity, permeability, etc.)

f (s, u): the joint probability density function of Z(s) and Z(u)

h: the distance separating sample

N (h): the number of distinct data pairs

Note: The real data (as porosity and permeability) usually skewed and do not have Gaussian

or normal distribution. For geostatistical analysis, the normal distribution of variables are needed.

The normalization procedure can be done by logarithm transformation or Box-Cox method as

skewed data distribution has a high impact on the variogram.

Definitions:

In a standard variogram, three parameters are recognizable as shown in Figure 4-35:

85
 1- Nugget effect: The variable and variance change in its neighborhood. It is estimated

from the empirical variogram at h = 0.

2- Sill – The asymptote volume in a variogram.

3- Range – The distance that a variogram reaches to the sill, it shows the range of influence

of each point on the others. So, there is no correlation between samples with distance

bigger than the range.

Figure 4-35: A typical Variogram diagram and main parameters.

4.12.2 Variogram models:

For kriging propose and estimate the value of the sill and range in an empirical variogram, the

variogram models are fitted to the empirical models. Some models are as following:

1. Linear model

2. Spherical model

3. Logistic model

4. Cauchy model
86
 5. Power model

6. Exponential model

7. Gaussian model

8. Matern model

For the current research, the most used model is spherical (Figure 4-36). The formulation of

this model is:

 ( h)  c0  c1 
3 h 1 h 3
 ( )  0ha Eq. 4-4
2 a 2 a 
 ( h)  c0  c1 ha

C0 is nugget effect and (C0+C1) shows the sill.

Figure 4-36: Three variogram models.

87
4.12.3 Anisotropies

The model and structure of variograms can change in the directions and different azimuth. For

a detailed study and accurate interpolation by the Kriging method, the azimuthal variograms (X,

Y, and Z directions) are needed. The anisotropy is a reflection of the sedimentation layering and

geological lineaments (as fractures and faults). The different sill and range in the different azimuth

can make an anisotropy ellipsoid, and the anisotropy is then considered in Kriging stage.

In the FRS project, the seismic study and geological condition of the area demonstrate a similar

values in the X and Y direction and so it was considered a homogeneous variogram, but with a

different structure of the variogram in the Z direction because of layering and it is calculated from

the well log data (Figure 4-37).

Nugget effect

Figure 4-37: The histogram (to check the distribution type of the data) and semi-variance with
a spherical model fitted for the density porosity variable (The nugget effect = 0.0002080, the
sill= 0.0023660, the range= 2.0700 m for h=15 cm in the well direction).

88
4.12.4 Kriging:

In the final step of the geostatistical analysis, Kriging method was used for the interpolation

purpose. It is a way to distribute permeability and porosity in the prepared geometry model and

cells. Kriging (introduced by D. G. Krige) is a local estimation technic which provides the Best

Linear Unbiased Estimator (BLUE) of the unknown characteristic studied (Journel,1991).

4.13 Primary models

The first tested geomodel has dimension of 3900*3000*473 m, and the grid sizes were fine near

the 10-22 well and coarse grid sizes in the other part of the model as shown in Figure 4-38. The

surfaces fixed by the seismic interpretation and well tops. The simulation time for this geomodel

is long, and because we needed to test some features and properties for the simulation, we change

it to smaller and simpler model, with dimension of 1000*1000*473 m.

This geomodel uses data from three wells and has a geometry derived from the interpreted horizons

of the seismic data as shown in Figure 4-40.

89
Figure 4-38. A geomodel made from 11 wells and a small cell size around the injection well.
The dimension of geomodel is 3900*3000 m.

Figure 4-39.The geometry of layers in the geomodel from the seismic interpretation. Colors
shows the main geological events and formations, dimension is 1000*1000 m

90
 Figure 4-40.The revised geomodel derived from log data from three wells. The dimension of
geomodel is 1000*1000 m.

4.14 The isotropic geomodel for fluid simulation

As mentioned, because the injection rate was found to be quite limited (Chapter 5), the

geomodel that finally used for the simulation is used only the 10-22 well for geomodelling. The

geometry is shown in Figure 4-39, and Figure 4-41 is a 3D figure of the geomodel showing porosity

distribution and Figure 4-42, Figure 4-43 and Figure 4-44 are a 2D figures of geomodel showing

vertical permeability and porosity, respectively near the injection horizon with more detailed

information about the grid size.The permeability in Z direction considered as 10% of permeability

in XY directions (personal discussion with Schlumberger reservoir expert).

91
Figure 4-41.The porosity geomodel made up with the well log data. The geomodel size is
1000*1000*240 m.

Figure 4-42. A section of the xy permeability geomodel shows the grid size in the injection
layer and others. The red rectangle shows the injection layer (i.e. Basal Belly River)

92
Figure 4-43. A section of geomodel for the permeability in z direction.

Figure 4-44. A section of geomodel for the porosity.

93
4.15 The geomodel for the seismic modeling and imaging

For seismic modelling using finite difference approach, the code is based on the equal grid

sizes, and it is defined in the code by 2D matrices. In the next step, we need to import and convert

the reservoir geomodel for input into the seismic code.

The cell size in the geomodel for the fluid simulation is variable from the injection zone to the

rest of the formation. The cell size for the seismic modeling and RTM imaging code is small size

as 1*1 m (Figure 4-45 (3D geomodel for Vp), Figure 4-46 (3D geomodel for density) and Figure

4-47 (2D geomodel for Vp)). The seismic code for part of the 2D seismic modeling and imaging,

so it uses 2D geomodel, extracted from the 3D geomodel.

Figure 4-45.The P-wave velocity model. This model was used for the seismic imaging.

94
 Figure 4-46. The density model used for seismic modelling.

Well 10-22

Figure 4-47: The P-wave velocity oriented by seismic layers (a 2D view of Figure 4-45 in a
section passing of the 10-22 well). The section is W-E and view is to the North.
95
Chapter 5. Reservoir fluid simulation for FRS project

5.1 Introduction

Reservoir simulation is a direct numerical method to model fluid flow in a porous medium

containing one or more fluids. Fluid simulation has the goal of managing production/injection rate

and for optimizing the operational cost. For simulation purposes data from many other disciplines

are gathered as shown in Figure 5-1.

Seismic
Interpretation Petrophysics Fluid Properties

Geological Numerical
Model Wells
Simulation Model

Model GRID
Facilities Tubing Curves Effects

Calibration of Observations & Production Data Interpretation

Figure 5-1. Disciplinary contributions to reservoir flow modeling (after Fanchi, 2006)

96
 For a fluid simulation, one needs to have a geomodel with valid geometrical data of the

reservoir with suitable grids and appropriate cell size. These grids make a 3D matrix such that each

component such as porosity or permeability were estimated and interpolated by a geostatistical

method. For this stage, the information and data are determined from petrophysics and well log

data combined with the geological model and seismic interpretation results.

In this project, the objective is to simulate CO2 gas injection in a shallow target (300 m) that

is a low pressure and temperature reservoir, and predict the behavior of the gas in the reservoir.

The results of the simulation are integrated with a rock physics study to translate the reservoir

parameters to seismic properties for monitoring purposes. In this chapter, I calculate the static

physical parameters of the reservoir to input in the simulation procedure, and the final output in

this chapter will be dynamic parameters (saturation and pressure, with the CO2 plume geometry).

In the previous section (4.11), I prepared a horizontally isotropic geomodel from the available

well log data (10-22), seismic interpretation results and geological data as shown in Figure 4-41.

The physical properties of the fluids (brine and carbon dioxide) are discussed in the next chapter

(Chapter 6) using an Equation of State for CO2 by Span and Wagner (1996) and formulas for the

brine properties by Batzle and Wang (1992).

5.2 Initial state of the FRS reservoir

The first step of fluid simulation study is related to the physical and chemical properties of the

reservoir. For the simulation purpose, the aquifer is considered a brine with 8000 ppm of salt. Also,

the fluid phase is assumed to be isotropic, homogeneous and isothermal in the research. The

temperature of the reservoir was measured to 13oC, and it will be constant during the gas injection

97
(isothermal). Figure 5-2 shows the hydrostatic and lithostatic pressure in the well 10-22. The

lithostatic pressure was calculated from the density log with a primary density equal to 2200 kg/m3

for the no data zone in the well shallower than 225m (depth of surface casing).Table 5-2 list the

initial properties in the reservoir.

There are many methods and approaches for calculation of the physical properties of fluids in

different phases. Batzle and Wang (1992) introduced approximations for density, viscosity and

bulk modulus of oil, gas, and brine and this approach is usually used by geophysicists. Cho (1970)

and Kestin et al. (1981) discussed other approximations for the physical properties of brine. In this

dissertation, Batzle and Wang equations were used for density, compressibility (= 1/bulk modulus)

and viscosity of the reservoir fluids. The calculation method is described in Chapter 6.

Figure 5-2.Hydrostatic and lithostatic pressure in the reservoir; the latter is calculated
from the density log data.
98
 Table 5-1.The initial properties of the injection target

Depth (m) 295-302

Thickness (m) 7

Lithostatic Pressure at aquifer top (MPa) 6.6

Temperature (oC) 13

Permeability (md) 0.1-2.0

Porosity (%) 14-18%

Salinity of formation water (mg/l) 8000

Density of formation water (kg/m3) 1005.2

The viscosity of formation water (centipoise) 1.16

5.3 The reservoir concepts and simulation methods

5.3.1 Relative permeability

When there is more than one fluid phase in a reservoir, the effective and relative permeability

can play a significant role in the simulation. Effective permeability is a value for the conductance

of a porous medium for a specific fluid phase when the reservoir has more than one fluid. Relative

permeability can be explained by Eq. 5-1 and it is equal to effective permeability (ki) normalized

by the single phase absolute permeability (k).

ki
kr 
k
Eq. 5-1
99
 Models prepared by Brooks-Corey (1964) that rewrote for the CO2 injection case by

Bachu (2013), were used for the relative permeability calculation in the current study (Eq. 5-2 and

Eq. 5-3).

SCO 2  SCO
C
krCO 2  k max
( 2
)n
1  SCO 2  Sb
CO 2 C irr

Eq. 5-2

1  SCO 2  Sbirr m
kr b  k rb
max
( )
1  SCO
C
2  Sb
irr

Eq. 5-3

Where: irr : irreducible, c : critical, S : saturation, max : Maximum, m , n : Corey’s coefficient

irr
for both the drainage and imbibition cycles and SCO 2 is calculated by Eq. 5-4 as:

max
SCO
S irr
 2

1  CSCO
CO 2 max
2
Eq. 5-4

C is a coefficient related to the trapping efficiancy, and C=0 iff all CO2 trapped and infinity if

it is not trapped.

Studies about relative permeability in Alberta’s sandstone and limestone formations by

Burnside (2014), Bachu (2013) and Bennion (2010). Bachu (2013) studied Alberta’s sandstone in

different formations with a range of permeability. The permeability of the Belly River sandstone

is between 0.1 to 2 mD. So, it is in low k value and based on the previous studies on the formations

with sandstone lithology in Alberta; the trapping efficiency will be 49-55% (Figure 5-3), so it

means the CO2 saturation can be alike similar to the trapping efficiency range.

Finally, the relative permeability curve for the BBRS was calculated based on Corey’s

equation for the reservoir’s sandstone as shown in Figure 5-4.

100
Figure 5-3. Trapping efficiency in sandstone based on previous work in Alberta (Bachu, 2013)

Relative permeability for FRS reservoir

1
0.9
Relative permeability

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.1 0.2 0.3 0.4 0.5 0.6
CO2 Gas Saturation

krg krw

Figure 5-4. The relative permeability curve calculated for the reservoir.

5.3.2 Anisotropy of the permeability

Permeability is influenced by stratigraphy, and rock fabric. Usually, it has a higher value

parallel to sedimentary layers compared to the perpendicular to the layers. Also, tectonic

101
phenomena as faults, joints and fractures can play a dominant role in the formation’s

permeability through fracture orientation. The FRS reservoir does not have any evidence for

significant faults, as determined from the seismic interpretation and attribute study.

For the current study, there is no measurement for permeability in the Z direction, and so by

the industrial protocol, it was considered 10% of the permeability in X and Y directions.

5.3.3 Formation compressibility

This parameter is necessary to measure the change in pore volume due to a change in reservoir

pressure. It can be expressed by Eq. 5-5.

1 V
cf   ( )T
V p
Eq. 5-5

Where:

𝑐𝑓 is formation compressibility, 𝑉𝜑 is the pore volume of rock, p is pressure on the formation.

T as a subscript in the last term shows the isothermal process.

For the reservoir, the P and S-wave slowness and the density were determined from the 10-

22 well log data. The velocity is calculated by the following formulas (Eq. 5-6):

1
Vp 
P  slowness
1
Vs 
S  slowness
Eq. 5-6

102
 The elastic moduli have a direct relation with the density, P and S-wave velocity shown

in Eq. 6-21 and this was used for the elastic modulus calculation. Figure 5-5 showing the result of

calculation for the velocities, elastic modulus and compressibility.

Depth (m)

Figure 5-5.Geomechanical properties of the reservoir in the well 10-22-17-16. The

compressibility was demonstrated by cf (the unit is 1/GPa).

103
5.3.4 Darcy’s law

The basis of reservoir simulation is the mass movement and its relation to the permeability

and reservoir pressure. In a medium with a particular phase in it, Darcy’s law can estimate the fluid

flow rates as Eq. 5-7 and with absolute permeability value in existence of just one phase. Darcy’s

law in one dimension can be expressed as:

k P
u 
 x
Eq. 5-7
For a medium, saturated by two phases (here: gas and oil) with considering effective

permeability definition for each phase can be determined from Eq. 5-8 and Eq. 5-9.

kkrg Pg
ug  
 g x
Eq. 5-8
kk P
uo   ro o
o x
Eq. 5-9

Where

un = volumetric flow rate for a particular phase n

krn = relative permeability of phase n

n = fluid viscosity for phase n

104
5.3.5 Simulation methods

Equations used in hydrocarbon simulation are based on the continuity equation. The

conservation law in the reservoir (conservation of mass, energy, and momentum) is essential for

mass balance and the continuity equation. In simple form, for each cell (Figure 5-6 and Figure 5-7)

a combination of Darcy’s law (Eq. 5-7), the material balance (Eq. 5-11) and flow equation are

solved (Fanchy, 2006 and ECLIPSE course material, Schlumberger, 2016).

Figure 5-6. Volume elements or grid block in reservoir simulation.

Material Balance:
Mass flux =Accumulation + injection/production

105
 Nc Nc

 Ckg  1 and
k 1
C
k 1
ko 1

Eq. 5-10
Mass balance of component k in one dimension:
 

x
 Ckg  g ug  Cko ouo     Ckg  g S g  Cko o So  
t
Eq. 5-11
Pcog  Pg  Po

Pcow  Po  Pw

So  S g  1

With considering Darcy’s law:

 kkrg Pg kk P  
 Ckg  g  Cko o ro o     Ckg  g S g  Cko o So 
x   g x o x  t
Eq. 5-12

For a reservoir’s element and in three dimensions it can be explained as:

J x J y J z C
   q  1
x y z t
Eq. 5-13
or:

  M     Q
t
Eq. 5-14

  M is the mass flux,   is the accumulation and Q is injection/production term
t
(ECLIPSE black-oil reservoir simulation, Schlumberger, 2009).

106
 Figure 5-7.Reservoir gridblock, coordinate and directions

Black-oil and compositional simulators are two methods for undertaking reservoir fluid

simulation. The Black-oil simulator is suitable for three components (oil, gas, and water) and

properties of the three phases as a function of pressure. It is usable for cases with recovery

mechanics not sensitive to composition changes in the reservoir fluids such as primary recovery,

solution gas drive, gravity, drainage, gas cap expansion, water drive, water, and gas injection

without mass transfer. The density of each phase is necessary for material balance equation. The

density also relates to pressure and temperature. PVT properties are required to estimate and

convert the volume of phases in a different environment as reservoir condition or production part.

The principal assumptions in the black-oil simulation are:

1. Darcy’s law governs the velocities of the fluids.

2. The void porous is filled by water/oil and gas

3. Capillary pressure = gas pressure – oil pressure

4. Phase mobility = phase permeability / phase viscosity

107
 A Compositional Simulator can support multi-component and multi-phase reservoirs

based on the Equation of State (EoS). Also, it can model the simulation and mass movement when

a new component was created because of chemical reactions. The Compositional method is

expensive and takes more time compared to black-oil Simulator methods.

For a Black Oil simulation, the parameters needed are (a) geometry, (b) matrix properties, (c)

fluid property and (d) well production/injection plan. Geometry and properties are input into grids

and cells with size and static properties for each of them (permeability and porosity). The fluid

properties are needed for the simulation, and they include viscosities, the solution gas-fluid (here

CO2 and water) ratio (Rs) and relative permeability curve for drainage and imbibition conditions.

The production/injection schedule and strategy are the final part to complete the simulation data

sequence, as the effect of production or injection will enter to simulation as our last term of the

material balance or simulator flow equations.

Briefly, as shown in Figure 5-8, the compositional simulator is suitable for oil and gas

reservoirs when the thermodynamical condition is near the critical point. The Black-Oil and

following requirements can help to select the right simulator:

5.3.6 Black-oil simulator

The black-oil simulator is useful for the following cases:

•Three-phases (oil, gas, and water)

•Three components (oil, gas, and water)

•When all fluid properties are functions of pressure

108
 Suitable for black-oil simulators are cases with recovery mechanics not sensitive to

composition changes in the reservoir fluids such as primary recovery, solution gas drive, gravity,

drainage, gas cap expansion, water drive, water injection and gas injection without mass transfer.

Black oil

Compositional

According to condition one

of simulators are usable

Black oil

Figure 5-8. Appropriate situation for Compositional and black-oil and compositional
simulators for oil and gas phases (ECLIPSE course material, 2016).

5.3.7 Specifications and Advantages of Compositional Simulator

Multi-component and multi-phase reservoir simulators are based on EoS modeling. Suitable

for compositional simulation are cases sensitive to compositional changes in reservoir fluids such

as primary depletion of volatile oil, gas condensate reservoirs and pressure maintenance in such

reservoirs. Also multiple contact miscible gas injection, CO2 and N2 injection.

109
5.4 The result of the reservoir simulation

The FRS reservoir is at shallow depth with low pressure and temperature. The temperature of

the formation is 13oC in the target zone, and the pressure is 30 bar. At this temperature, the injected

CO2 will change from gas to the liquid phase at a pressure equal to 49 bar. The goal of the research

is to inject CO2 in gas phase, and so the strategy will be a constant bottom hole pressure equal 49

bar for five years. After five years, the injection will be stopped, and the monitoring will continue

for a decade. According to the reservoir’s PVT table, simulation has been chosen to be undertaken

by the compositional simulator (Figure 5-9) for CO2 injection in the gas phase. The result of

simulations for the gas saturation and pressure are outlined in Figure 5-11.

The compositional simulation is a complex and time taking procedure compare to the black-

oil simulator. The black-oil simulator has been used in the WASP project, that CO2 were injected

in the supercritical condition (see WASP reservoir simulation by the black-oil method, Nowroozi,

2013). Figure 5-9 shows the phase diagram for the CO2 and the FRS project condition. Figure 5-10

is a detailed phase diagram for T= 0 to 30 oC and P=0 to 80 bar (8 MPa).

110
 Figure 5-9: The phase diagram of carbon dioxide and pressure and temperature
condition in the FRS project (Phase diagram from ChemicaLogic Corporation).

Figure 5-10. The phase diagram of CO2 for the reservoir condition. This figure is a
magnified image of Figure 5-9 near the reservoir condition.
111
5.5 Simulation results

As explained previously, the simulation was undertaken using a compositional method. As

mentioned, the injection is in 10-22-17-16 well, with BHP~4.9 MPa for a five-year period. The

geomodel has 1 km*1 km dimension, so the boundary of the geomodel was considered to be as

unlimited and open boundary for a reliable and accurate results of the simulation. In the unlimited

and free boundary, the pressure increase due to injection may transfer out of geomodel without

any effect inside the reservoir.

Figure 5-11 (pressure during injection) and Figure 5-12 (gas plume during the injection) are

the result of the simulation. The plume size after five-year injection is 184 m*10.6 m.

112
 Pressure(MPa)

200 m

One-year injection

Two-year injection

Three-year injection

Four-year injection

Five-year injection

10.6 m

Figure 5-11.The simulation result for the reservoir pressure by the CO2 injection for five years
with BHP=4.9 MPa. The scale is same as Figure 5-12.The unit for the pressure is MPa.

113
 CO2 gas saturation

Figure 5-12. The CO2 gas saturation for the five-year injection by BHP=49 bar (4.9
MPa)

114
 Figure 5-13: Diagrams showing the result of injection for BHP=49 bar for five years
(the x-axes show the year of injection) a. Cumulative gas mass (kg), b. Cumulative gas
volume (m3), c. Daily volume (SC) injected gas rate (m3/day), d. Daily mass injected gas
rate (kg/day), e.Well bottom hole pressure (kPa), f. Well block pressure (kPa), (SC stand
for Standard Condition - 15oC and 1 bar).
115
5.6 The CO2 gas injection effect on fluid phase in the reservoir

CO2 gas injected into the brine can have some effects in the reservoir. The mechanisms that

may occur during the injection include:

1- Evaporation of brine into the CO2. The evaporation can increase salinity in the water and make

a new phase in the reservoir (vapor water phase). Also, this effect can decrease the value of

the immiscible water near the injection zone.

2- In a brine with a high amount of dissolved salt a desiccation can convert salt to a crystalline

form. It can reduce the porosity and permeability and so the injectivity (Figure 5-14).

The drying-out effect can change the simulation parameters during the injection process. The

simulation software does not consider the alteration of the parameters in the drying-out zone. For

controlling the permeability loss due to salt precipitation near the well, injecting a slug of fresh

water before the commencement of CO2 injection can be a solution (Karsten et al. 2009).

Figure 5-14.Results of visual inspection of the brine+CO2-reacted sample; deposits of NaCl

crystals (salt precipitation) and calcite dissolution textures at the outer surface of the sample
in a lab test (Rathnaweera et al., 2016).

116
 The gas injection also causes a pressure change in the reservoir. The pressure can have a

direct effect on the permeability of the gas (Rathnaweera et al., 2016). For a high-pressure change

in the reservoir, the CO2 permeability variation during the injection procedure is not negligible,

and so it should be considered in the fluid simulation.

5.7 Injection with different BHP

In this section, the behavior of the CO2 plume is studied for different injection pressures.For

this purpose, the BHP equal to 48, 51.41, 53.42 bar were simulated.Due to the fast pressure changes

in the reservoir, it can be possible to inject the of CO2 in liquid phase, and by decreasing the pore

pressure, the phase change from liquid to gas will happen.However, the current simulation

software does not support the phase change conditions at low pressure and temperature so the

simulation results may be biased. Figure 5-15 shows the five-year injection for the BHP=4.8 MPa

(48 bar). As a result, the cumulative gas mass in each case were obtained and it introduced an

exponential function between cumulative mass and BHP as shown in Figure 5-44.

117
 165 m

Figure 5-15. Gas saturation after five-year injection with BHP=48 bar.

5.8 Long-term prediction

The reservoir modelling process was continued for a century to observe the CO2 plume

behavior (after stopping the injection). As the expectation, during the monitoring, the gas plume

migrates to the top of the reservoir and it makes a high saturation plate in the top layer (saturation

can reach up to the reservoir efficiency value that is equal 0.5 in this type of sandstone

(Bachu,2013)). The results of the fluid simulation are demonstrated in Figure 5-16 through Figure

5-23 for up to 100 years post-injection. Figure 5-24 is the pressure variation for a long-term

prediction. It shows the reservoir pressure reduces to the initial value after 28 years from the first

injection day. The pressure will fall after stopping injection, and after 23 years, it will decrease to

the initial reservoir pressure.

118
 Plume Diameter = 186 m

Thickness = 10.5 m

Figure 5-16.The saturation after five-year injection by constant BHP.

Figure 5-17. Predicted CO2 saturation five years after discontinuing the injection

Figure 5-18. Predicted CO2 saturation ten years after stopping the injection

119
Figure 5-19. Predicted CO2 saturation 20 years after stopping the injection

Figure 5-20. Predicted CO2 saturation 40 years after stopping the injection

Figure 5-21. Predicted CO2 saturation 60 years after stopping the injection

120
Figure 5-22. Predicted CO2 saturation 80 years after stopping the injection

Plume Diameter = 280 m

Thickness= 2.5 m

West East

Figure 5-23. Predicted CO2 saturation 100 years after stopping the injection

121
 Figure 5-24. The Bottom-hole pressure changes over a century. The pressure will be equal to
initial reservoir pressure after the year 2044 (28 years after beginning the injection process).

5.9 Conservation of CO2 mass after stopping the injection

After five years’ injection, the mass of CO2 will be constant in the reservoir. If the injected

CO2 mass is considered to be equal M, so based on the cells porosity, saturation and the pressure

of the reservoir, it can be explained using the following formula:

n
Ct M i  t 
i 1
Eq. 5-15

That M is CO2 mass in a cell as a function of time (t), and n shows the cells number with CO2

saturation more than zero and C is a constant value.

For a cell, Mi be explained as:

M i  i siCO 2Vi CO2

Eq. 5-16

122
 Moreover, so for the entire reservoir:

n n
Ct   M i   i siCO2Vi CO2
(T , P)
i 1 i 1
Eq. 5-17

Where: Vi  xi yi zi and x, y and z are the three dimensions of each cell.

This formula can help to predict the expansion of CO2 plume in a sealed reservoir after

stopping the injection procedure (if the gas solution in the brine and chemical reaction between

minerals, CO2, and carbonic acid is ignored). The prediction will be near the real fluid simulation

result after a long time (stabilization time). This time is a function of pressure, temperature, and

permeability.

The plume thickness

16 All cells reached to the almost
14 trapping efficiency rate
The plume Thickness (m)

12
x and z Mainly x
10 directions direction
8 migration migration
6
4
y = 33.203x-0.513
2
R² = 0.8425
0
0 20 40 60 80 100 120
Time (year)

Figure 5-25.The thickness of the plume in the injection point after stopping the injection. After
100 years, all cells in the plume show a saturation rate around the trapping efficiency, and so
continued gas migration will change to a mainly horizontal direction.

123
 As mentioned previously, the plume size is a function of the permeability vector in the

xy and z directions. The plume thickness at the injection point decreases after stopping the injection

as the gas migrates to the higher levels in the reservoir. This decreased rate after stopping the

injection is an exponential function of time that is unique for each reservoir. For the FRS reservoir

in the Basal Belly River sandstone zone, it is illustrated in Figure 5-25 that shows a good fit after

40 years. The form of the CO2 gas plume is like an inverted cone because of gas density and

gravitation effect. So, the average thickness of the plume can be explained as half as the central

thickness. The relation between time and the plume thickness is an exponential function.

Z center
Za 
2
Z center  33.2* t 0.513
Z a  16.6* t 0.513
Eq. 5-18
that Za is average thickness and Zcenter is the thickness of the plume in the injection point, and

t is time after stopping the injection.

To test the equations Eq. 5-15 to Eq. 5-18, I simulated the fluid behavior for injected CO2 for

233 years. The result is shown in Figure 5-26. The central thickness of the plume after 233 years

of the injection day can be determined by the function fitted to the curve in Figure 5-25 and Eq.

5-18 is 2 m, and the average plume thickness is 1 m.

V= 32 m3 (the size and volume of each cell in the reservoir)

 = 16 % (for the top layer, the porosity is the same in all cells in a layer)

S= 52% (the maximum possible CO2 saturation by considering the trapping efficiency)

124
 CO =69.555 kg/m3
2
(@13oC and 30 bar) (after stopping the injection the pressure

drops to the reservoir’s natural pressure that is around 3 MPa (Figure 5-13.e)).

M
i 1
i  Mt = 735560 kg (see Figure 5-13.a.)

So, for similar cells size Eq. 5-17 can be explained by Eq. 5-19:

n n

 i SiCO2Vi CO2 N SCO2V CO2

(T , P) (T , P)
M i 
i 1 i 1
Eq. 5-19
That N shows the number of uniform cells.

With the available data in this project, Eq. 5-19 can be used to calculate number of cells with

CO2 gas inside:

32*0.16*0.52*69.555*N = 735560

Finally, the CO2 will spread in N=3972 cells. If we consider the plume thickness equal to 1 m,

the coverage area is a=124006 m2, equal to 397 m in diameter. It is similar to the simulation result

that shows the plume diameter at 376 m.

CO2 is not completely free gas in the reservoir. The first absorbance is the capability of the

CO2 solution in brine. It is a function of the physical condition of the reservoir and can be described

as:

M1=g (P, T)

The chemical reaction between carbonic acid and the minerals can absorb CO2 into a solid

phase. The amount of this mass is a function of time that can be explained as:

M2=f(t)

125
 For a long-term plume size fate, this two absorbance factors should be considered for a

better estimation as:

n n

 i SiCO2Vi CO2  (M1  M 2 )

(T , P)
M i 
i 1 i 1
Eq. 5-20

376 m

Figure 5-26. The result of long-term monitoring for the gas saturation after 233 years from the
start of injection. The maximum plume thickness is 2 m.

5.10 An injection/production pattern for improved gas phase injection

The purpose of the injection in the previous section was the gas phase injection that should

not exceed of 49.41 bar @13 oC (reservoir temperature). Decreasing the pore pressure in the

reservoir can increase the mass injection into the porous medium. A brine production plan in a

pattern of wells around the injection well can reduce the pore pressure under a phase change

condition.

A production well that produces brine from the reservoir level, decrease the pore pressure in

the reservoir. Figure 5-27 and Figure 5-28 show the pressure and plume size in the area based on

126
the fluid simulation results for the 10-22 well. Figure 5-29 shows four production wells at a

specific distance X meter from the injection well. The plume shape with these production wells

will be a function of production rate and the location of the wells.

For increasing the injection rate in the gas phase, a plan could be drill four production wells

that remove the brine out of the BBRS aquifer; they can reduce the pore pressure up to 6.5 bar in

the reservoir if they locate in 50 m distance of well (10-22). The maximum injection pressure could

then increase up to 5.6 MPa (56 bar) in the gas phase. Another advantage of these production wells

is estimating real permeability in the directions of the wells. The permeability estimation in four

directions could show the possible anisotropy in permeability or fractures around the injection

well.

Figure 5-27.The pressure condition in 25,50 and 200 m distance of injection well after one-
year injection.

127
 Figure 5-28.The plume size after 1 and 5-year injection and 100-year post-injection.

Figure 5-29. The production well(s) in X m distance to decrease the reservoir pressure,
it can help to inject more mass in the gas phase. T is the angle between the wells. X, T and
the number of wells are variable.
128
5.11 Injection at higher BHP

As mentioned, the simulation software (ECLIPSE and CMG) do not support a phase change

from gas to liquid (it was tested with both software packages). Thus, it is not possible to simulate

the CO2 behavior through a phase change from gas to liquid. The research is just in the gas phase,

and for the testing upper pressures, we changed the physical properties to stay in gas phase. For

this part, we considered a higher reservoir temperature of 20oC, as makes it possible to increase

the BHP to 57.3 bar. The pressures examined were 5.141, 5.341, 5.541 and finally 5.73 MPa at

20oC. This maybe the case of the reservoir temperature around the well is increased by heating the

CO2. The results are described in the following pages as Table 5-2.

Table 5-2. The different BHP and related figures.

Plume After a Plume After 5- Plume After ten years The Cumulative The Cumulative
BHP(bar) year injection year injection of stopping injection Gas Mass Gas Volume
51.41 Figure 5-30. a. Figure 5-30. b. Figure 5-30. c. Figure 5-32. Figure 5-33.
53.41 Figure 5-34. a. Figure 5-34. b. Figure 5-34. c. Figure 5-35. Figure 5-36.
55.41 Figure 5-37. a. Figure 5-37. b. Figure 5-37. c. Figure 5-38. Figure 5-39.
57.3 Figure 5-40. a. Figure 5-40. b. Figure 5-40. c. Figure 5-41. Figure 5-42.

Figure 5-31 shows the reservoir pressure simulation result for the BHP=5.141 MPa. The

pressure drops after stopping the injection (Figure 5-43). Also, Figure 5-44 shows the final

summary of the cumulative mass injection for the different BHP @20oC. An exponential

function explains the mass injection value as a function of BHP in the FRS project as predicted

by Eq. 5-21.

Total injected CO2 Mass = 24.613e0.0689BHP

Eq. 5-21

129
 a

72 m

10.6 m

Figure 5-30. CO2 saturation at pressure of 51.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031).

130
 a

Figure 5-31.The pressure change in the reservoir for BHP=51.41 bar and temperature =20oC,
(a). one year after injection, (b). Five-Year injection, (c). Two years after stopping the
injection.

131
 Figure 5-32.The cumulative gas mass (kg) injected at the constant BHP=51.41 bar.

Figure 5-33.The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=51.41 bar.

132
 a

Figure 5-34. CO2 saturation at pressure of 53.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031).

133
 Figure 5-35. The cumulative gas mass (kg) injected at the constant BHP=53.41 bar

Figure 5-36. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=53.41 bar

134
 a

Figure 5-37. CO2 saturation at pressure of 55.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031).

135
 Figure 5-38. The cumulative gas mass (kg) injected at the constant BHP=55.41 bar

Figure 5-39. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=55.41 bar

136
 a

Figure 5-40. CO2 saturation at pressure of 57.41 bar @ 20oC. (a). after one-year
injection(2017), (b). after five-year injection(2021), (c). after ten years of stopping the
injection(2031).

137
 Figure 5-41. The cumulative gas mass (kg) injected at the constant BHP=57.3 bar

Figure 5-42. The cumulative gas volume (m3) in the standard condition (red graph) and the
reservoir condition (blue graph) injected at the constant BHP=57.3 bar
138
Figure 5-43. The pressure over 5 years of injection at BHP=57.3 bar (5.73 MPa), and 10 years
after injection.

Cummualtive CO2 injected in 5 years (T=20 C)

1300
1250
1200 y = 24.613e0.0689x
1150 R² = 0.9992
Injected CO2 (ton)

1100
1050
1000
950
900
850
800
750
700
49 50 51 52 53 54 55 56 57 58
BHP (bar)

Figure 5-44. Cumulative CO2 mass for different BHP for a five-year injection plan. An
exponential function describes the relation between BHP and injected gas mass.

139
Chapter 6. Rock physics study for the FRS project

6.1 Introduction

This chapter engages rock physics to calculate physical parameters of the formation, the fluid

and rock matrix in the reservoir due to CO2 injection into the aquifer. Rock physics is a bridge

between seismic data and reservoir properties that integrate geological uncertainties. The output

and the primary results of reservoir simulation are the distribution of pressure and saturation. Fluid

substitution formulas and Gassmann’s equation are a part of rock physics studies about the effects

of fluid changes on the bulk modulus and consequently on the seismic velocity in the formation.

It was first introduced and discussed by Gassmann (1951), and provides a base equations of fluid

substitution in rock physics.

The density and seismic wave velocities in a fluid is a function of the pressure and temperature.

For the physical properties of CO2, work by Span-Wagner (1996) was used and, Batzle and Wang

(1992) provided the physical properties of the brine. In the reservoir, the fluid is a mix of various

fractions of CO2 and brine, and the velocity of the formation is a function of the CO2 saturation

value and the mixed fluid condition of brine and the gas. Finally, the seismic velocity variation in

the reservoir is determined from the different mixed fluids.

Gassmann’s equations were used for estimating the saturated bulk modulus in the formation

after injecting the CO2 gas. As the CO2 gas saturation is obtained from the reservoir simulation,

the velocity of each cell in the reservoir model is determined from the rock physics calculations,

so each cell has physical properties as a function of the pressure and the injection time.

The workflow for the seismic parameter estimation over the injection period are:

140
 1. Reservoir fluid simulation for CO2 injection at constant bottom hole pressure rate in the

target formation, using a compositional simulation method.

2. Input all property values (depth, porosity, saturation,) from log data, geomodel, and fluid

simulation results.

3. Calculate the initial mineral bulk modulus with the different mineral composition for the

target sandstone in Belly River sandstone. The detailed mineralogy study as described previously

in section 2.5.

4. Use Batzle-Wang and EoS equations to calculate bulk modulus and density for brine water

and CO2 and mix fluid in each cell.

5. Calculate the initial bulk modulus (Ksat) for saturated rock (before injection) by using log

data and P-wave velocity data.

6. Estimate the saturated bulk modulus and P-wave velocity for each cell during injection.

Figure 6-1 is a brief flowchart shows the work flow for the velocity and density estimation in

the formation after the gas injection, described in this chapter.

6.2 The complexity of a solid or fluid

Rocks and fluids in nature are not homogeneous, isotropic and mono component. Figure 6-2

is a polarizing microscopic image of sandstone in which each color represents a different mineral.

It shows a mixing medium of minerals that formed the rock. In the reservoir characterisation, we

need to have a realistic estimation of physical properties of the mixed fluid and solid part and

141
together as the formation properties. For calculating the physical properties (such as elastic

modulus) of a mix (in solid or liquid phase), some methods and equations are introduced.

Estimate effective
Bulk modulus of
elastic modulus of
the minerals
mixed minerals

Estimate effective The elastic

Bulk modulus of Fluid
elastic modulus of modulus after
the fluids substitution
mixed fluids injection
The velocity
after injection

The density
Rock density
after injection
Porosity

Elastic modulus
P and S velocities
estimation of
from log
reservoir

Figure 6-1. The fluid substitution procedure used in this chapter.

142
 Figure 6-2. A microscopic thin section of sandstone in the polarization microscope. It
demonstrates a variety of minerals in a rock (source: micro.magnet.fsu.edu).

To estimate a physical property of a substance that is made of mixed material (solid and

liquid), the physical properties of the components individually (such as elastic modulus and

density), the volume fraction, and components arrangement and geometry.

Precise estimation can be determined when all defined parameters described in the last

paragraph are available, but there is always some uncertainties due a lack of data. Effective

medium theory makes it possible to define upper and lower bounds of the property. It helps to have

knowledge about the maximum and minimum limits of the physical parameters (as bulk modulus

or velocity). In this section, the main challenge is determining the bulk modulus and density of the

formation in the reservoir before and after the gas injection.

Commonly three methods are used for calculation of density and bulk modulus for reservoir

fluids at different pressure and temperature:

1. Calculated from equation described and derived by Batzle and Wang, (1992).
143
 2. Measurements of the fluid, that are recovered from the reservoir or formation.

3. The equation of state (EoS) is the best method for calculating the fluid properties.

(McCain, 1990; Span and Wagner, 1996; Danesh, 1998).

6.3 The physical properties of the mix phases

Some physical properties are related to the geometrical distribution of the components in the

medium, and some are not. For example, the density is a simple property that is not related to the

distribution form and homogeneity of the mixture. For this kind of properties, we can explain an

average value by Eq. 6-1.

n
M Average   f i M i
i 1
Eq. 6-1

Where: M is a physical property, fi is fraction of i th component, and n is number of

components.

This formula is the Voigt average that will be explained in the following sections (Figure

6-13). Some properties are sensitive to the components geometry distribution or homogeneity and

isotropic specification in the mass. The velocity and elastic modulus are properties in this category.

In the next section, we examine average properties for these parameters.

6.4 Voigt and Reuss average

There are many methods for calculating a physical property of a mixed phase (such as

mineral’s bulk modulus). The Reuss lower bound and Voigt upper bound can be described through

average by Eq. 6-2:

144
 n
K   ( fi Kim )m
i 1
Eq. 6-2

Where n is the number of components, i is component’s number in the mixed phase and fi is

fraction of i th component. If m= -1 it is Reuss average and if m=1 it defines the Voigt average.

Voigt-Reuss-Hill (VRH) average (Eq. 6-3) is an approximation that uses Voigt and Reuss

estimation as:

1
KVRH  ( KVoigt  K Re uss )
2
Eq. 6-3

Figure 6-3 shows a sample of bulk modulus calculation for a mix of sand (plagioclase) and

water, solved by Reuss and Voigt averages and the physical concept of them. The Voigt average

is a formula for an iso-strain model and Reuss is an average that solves an iso-stress model. When

the water saturation is over 60%, effects of suspension are important. It is usual in rocks that are

not compacted and not cemented in environments with a high-water content.

145
 Figure 6-3: The bulk modulus for the mix of a porous sand with 100% plagioclase (as
immature sand) and water.

6.5 Brie’s average for fluid mix

Brie’s average is an empirical fluid mixing law, introduced by Brie (1995) and can be

explained by Eq. 6-4 :

K Brie  ( Kliquid  K g as )(1  S g as )e  K g as

Eq. 6-4

Where Kliquid is the bulk modulus of the liquid phases calculated from the Reuss average (for

the mixed case) in the reservoir, K g as is the gas bulk modulus and S g as is gas saturation. As

mentioned previously, the Voight bound is appropriate for a patchy mixed fluid condition, but this

occures rarely, so technically, the Brie’s average is preferred to be considered as the upper bound

for a mixed fluid.

146
 For the velocity estimation after fluid substitution, the fluid properties need to be

calculated. Figure 6-14 shows the bulk modulus of the mixed fluid (brine+CO2) with a different

fraction of CO2 calculated by Voigt, Reuss, VRH and Brie’s average methods.

6.6 Hashin-Shtrikman (HS) Bounds

The Voigt-Reuss bound introduces a wide range in average physical properties (Figure 6-3).

The Hashin-Shtrikman bounds are a better way to make the narrower bounds for a property (such

as the bulk modulus) estimation. Figure 6-4 shows the distribution and geometry of two

components in the Hashin-Shtrikman bounds calculation. For the maximum and minimum HS

boundary, the position of the minerals were changed in the calculation. Eq. 6-5 and Eq. 6-6 are for

calculation bulk and shear modulus in the HS average method.

f2
K HS   K1 
4
( K 2  K1 ) 1  f1 ( K1  1 ) 1
3
Eq. 6-5

f2
 HS   1 
( K1  21 ) 1
( 2  1 ) 1  2 f1[ ]
4
51 ( K1  1 )
3
Eq. 6-6

Where Ki is the bulk modulus and µi is the shear modulus of the material with fi fraction.

147
 Figure 6-4.Two phase material in the Hashin-Shtrikman bounds.

For the upper boundary (HS+) calculation, the hard material is considered as the first

component ( K1 , 1 ), and for the lower boundary (HS-), the softer material is selected as the first

component.

Walpole (1966) introduced a new form of the Hashin-Shtrikman approach called the Hashin-

Shtrikman-Walpole method, as Eq. 6-7 and Eq. 6-8 (after Mavko, 1998):

f2
K HS   K1 
4
( K 2  K1 ) 1  f1 ( K1  m ) 1
3
Eq. 6-7

f2
 HS   1 
 9 K  8m 1
( 2  1 )1  f1[ 1  m ( m )]
6 K m  2m
Eq. 6-8

148
 Where subscript m refers to the maximum bulk and shear modulus values for the upper

bound and minimum for the lower bound calculation. A general form of Hashin-Shtrikman-

Walpole equations for more than two-phase material (Berryman (1995)):

K HS   ( max )
K HS   ( min )
Eq. 6-9

 HS   ( ( K max , max ))

 HS   ( ( K min , min ))
Eq. 6-10
Where:

1 4
( z )    1  z
4 3
K (r )  z
3

1
( z )    1  z
 (r )  z

 9 K  8
 (K ,  )  ( )
6 K  2

The brackets indicate an average over the medium, which is the same as an average over the

constituents weighted by their volume fractions.

Figure 6-5 shows the bulk modulus calculated for a mix of calcite and quartz by using the

Voigt, Reuss, VRH and HS averages. As can be seen, the VRH average approximately is equal to

the HS bounds, and HS+ is near to HS- in this sample for two minerals. Figure 6-6 is another

149
sample of mixed minerals (quartz and wet clay) for which the bulk modulus was calculated

by the Voigt, Reuss and VRH average methods. For two mixed fluids (shear modulus of fluids is

zero), the HS+ and HS- are Reuss averages.

Figure 6-5.The average bulk modulus for a mixed case of quartz and calcite. The blue
curves show Voigt, Reuss and VRH averages. The red curve is HS+ and green is HS-. The
VRH is very near to Hashin-Shtrikman averages.

Figure 6-6. Matrix Properties calculated by Voigt (blue), Reuss (red) and VRH (green)
methods for a mix of pure quartz and wet clay.

150
6.7 Fluid properties

During the injection or production, the pressure changes in the reservoir due to the injection.

Also, if the injected fluid’s temperature is not equal to the formation’s temperature, it will be

affected. As mentioned in the Chapter 5, the aquifer salinity can change locally by CO2 injection

around the injection point. A secondary effect of these changes can affect the seismic velocity and

formation density and consequently seismic responses. In the project, there is assumed to be no

temperature change during the injection.The pressure increases a little during the injection (from

3 to 5 MPa). In this section, I discuss the fluid phases properties and the mixing case.

6.7.1 Brine

Water has high capability to dissolve salt and ions. Like other liquids, the pressure and

temperature changes have an effect on the physical properties of the water. For seismic modeling

of the reservoir, I calculate the bulk modulus, seismic velocities and density of the brine. There

are some approaches for this purpose, and for a geophysicist, Batzle-Wang (1992) equations are

the well-known method. Also, another alternative introduced by Rowe and Chou (1970) for the

brine density and bulk modulus.

As mentioned, the density of pure water is a function of temperature and pressure. By a

polynomial (Batzle-Wang, 1992) it is possible to calculate the density of water in the various

temperatures (T) and pressures (P) as Eq. 6-11.

w  1  106  80T  3.3T 2  0.00175T 3  489 P  2TP  0.016T 2 P  1.3 105 T 3 P  0.333P 2  0.002TP 2 

Eq. 6-11

151
 For the brine, salinity is another parameter that should be considered in the density

calculation. So, the density of brine can be represented as Eq. 6-12:

b   w  S 0.668  0.44 S  106 300 P  2400 PS  T  80  3T  3300S  13P  47 PS  

Eq. 6-12

In Eq. 6-11 and Eq. 6-12, ρw and ρb are water and brine density in g/cm3, P is pressure in MPa,

T is the temperature in Celsius and S is the weight fraction of salt (NaCl) in (ppm/1000000).

The bulk modulus of the brine is predictable by a simplified function as Eq. 6-13 (Chen et al.,

1978):

VB  Vw  S (1170  9.6T  0.055T 2  8.5 105 T 3  2.6 P  0.0029TP  0.0476 P 2 )

 S 1.5 (780  10 P  0.16 P 2 )  820S 2
Eq. 6-13

Where V is the velocity of P-waves in the brine (VB) and water (VW). The water velocity can

be estimated by Eq. 6-14 up to 100oC and about 100 MPa (Wilson, 1959):

4 3
VW   wijT i p j
i 0 j 0

Eq. 6-14

Where the coefficients (wij) are:

w00  1402.85 w50  1.524 w02  3.437 103 w03  1.197 105
w10  4.871 w11  0.0111 w12  1.739 104 w13  1.628 106
w20  0.04783 w21  2.747  104 w22  2.135  106 w23  1.237 108
w30  1.487  104 w31  6.503  107 w32  1.455  108 w33  1.327 1010
w40  2.197  107 w41  7.987  1010 w42  5.230 1011 w43  4.614 1013

152
 Figure 6-7 and Figure 6-8 show diagrams for density, bulk modulus and velocity of brine

(salinity=8000 ppm in BBRS reservoir) and pure water. These properties are a function of pressure

in Figure 6-7 and temperature in Figure 6-8. The relation between the properties are nearly linear

with the pressure between 2 to 8 MPa.

Figure 6-7. The density, bulk modulus and P-wave velocity of brine and water temperature
from 13 to 28 oC (with 5 oC steps) and salinity equal to 8000 ppm from 2 to 8 MPa.

153
 Figure 6-8.The density, bulk modulus and P-wave velocity of water and brine (salinity=8000
ppm) as a function of temperature. Each curve belongs to the pressure from 1 to 10 MPa in
steps of 2 MPa.

The viscosity of brine is also a necessary parameter for the fluid simulation. Kestin et al.,

(1981) derived a formula for the brine viscosity:

  0.1  0.333S  (1.65  91.9S 3 )exp{[0.42(S 0.8  0.17)2  0.045]T 0.8}

Eq. 6-15

Where T is temperature and S is salinity. The pressure effect is negligible for the viscosity

change in water and brine, so there is no influence of it in the formula. Figure 6-9 shows the

viscosity of the brine as a function of temperature. The blue arrow in Figure 6-9 indicates the

viscosity for the FRS aquifer.

154
 Figure 6-9. The viscosity of brine (based on Batzle-Wang (1992)), the pressure does not
have a significant influence on the brine viscosity.

6.7.2 Carbon dioxide

The equation of state for carbon dioxide can predict actual physical parameters for it at

different temperatures and pressures. Span and Wagner (1996) described a very detailed

formulation for CO2 properties from using the equation of state (EoS) that were used in this

research. Other articles introduced the CO2 properties using simple calculation methods (e.g.

Vargaftik (1975) and Sun (2009)). Two diagrams in Figure 6-10 demonstrate the bulk modulus

and density of carbon dioxide as the function of pressure and temperature. In the FRS project, the

reservoir temperature is 13oC. To inject CO2 in gas condition into the reservoir at a higher BHP,

some simulations were tested for a higher temperature as 20oC in the last chapter. The reservoir

pressure changes from 3 MPa to 5.5 MPa. Figure 6-11 and Figure 6-12 show the density and P-

wave velocity respectively at reservoir conditions.

155
 Figure 6-10.The bulk modulus and density of CO2 at different pressures and
temperatures. EoS described by Span and Wagner (1996) were used to generate the
diagram (drawn by Yam,2011).

Figure 6-11. Density of CO2 for 13 and 20oC and 23<p<57 bar.

156
 Figure 6-12. P-wave velocity of CO2 versus pressure at T= 13 and 20 oC (the velocity
calculated upper than 4 MPa at T=13 oC was unstable)

6.8 Mixed fluid properties

A mixed fluid of brine as a liquid and carbon dioxide in the gas phase will decrease the density

and bulk modulus as a function of the CO2 saturation. The mixed condition of brine and carbon

dioxide (that can be fine mixed, semi-patchy or patchy mixed condition) has a significant effect

on the P-wave velocity and bulk modulus (see sections 6.3, 6.4, 6.5 and 6.6).

Garcia (2001) reported a density increase equal 2-3%, when CO2 dissolves in the water (or

brine). In this research, the effect of the CO2 solution in the brine is ignored. The bulk modulus

and the density of fluid phase are calculated in this section to estimate the P and S-wave velocity

and density at the formation. Also, the all possible fluids mixed patterns (from fully patchy to fine

mixed) are considered for the 4D seismic modeling. The phases properties (for brine and CO2)

157
generated previously (from Batzle-Wang equations and CO2 Equation of State formulation)

by using the different averages (as Voigt, Reuss, VRH and Brie) were used to calculate the mixed

fluid properties shown in Figure 6-13 through Figure 6-16.

Figure 6-13. The density of the mixed fluid in T=13.8oC and pressure from 30 to 60 bar
(3 to 6 MPa).

158
 Figure 6-14. The bulk modulus for the mix of brine with 8000 ppm salinity and CO2 in
13 oC and 4.5 MPa (45 bar). In the mixed fluid condition (as CO2 and brine), the Hashin-
Shtrikman averages (upper and lower bounds) are using the Reuss Average.

Figure 6-15. The P-wave velocity in the mixed fluid of the brine (8000 ppm salinity) and CO2
in T=13 oC and P=4.5 MPa (the reservoir condition during the injection procedure).

159
 Figure 6-16. The bulk modulus of the mixed fluid with a different fraction of CO2 and
different mixed condition in P=30 bar (3 MPa) and 45 bar (4.5 MPa).

6.9 Effect of pore pressure on seismic velocity

For a gas (as CO2) the pressure has a significant effect on the velocity and density (Figure 6-11

and Figure 6-12). Figure 6-17 demonstrates the influence of the pressure on the velocity of mixed

fluid with a different fractions of CO2 and brine. The pressure change in this example is for the

maximum case from 3 to 4.5 MPa (equal to a 1.5 MPa change) that will increase velocity between

3-7% (Figure 6-17). As we demonstrated in the last chapter, the pressure in the reservoir will

decrease after ceasing the gas injection and it will return to the initial reservoir pressure after nine

years.

In the reservoir system, CO2 flooding has an effect on seismic velocity by changing the pore

or effective pressure. Higher pore pressure directly impacts the effect of CO2 injection on the

160
seismic velocities, as lab experiences show a 2-6.9% decrease in Vp for a maximum 12 MPa

increase in the pore pressure (Wang et al., 1998). For the FRS project the injection has low BHP,

so the velocity change due to the pressure change is negligible. However, a basic calculation on

the pressure change effect on the fluid phase velocity in the reservoir was undertaken and the result

is demonstrated in Figure 6-17. The velocity (or bulk modulus) change in the fluid shows big

variation with a large pressure change (Figure 6-18) or in the CO2 change in phase.

Figure 6-17. The velocity change in the fluid phase of the reservoir (brine+CO2) for a
semi-patchy mixed fluid by pressure. The pressure increased from 3 to 4.5 MPa.

161
 Figure 6-18: Bulk modulus estimation for different fraction of fluid mix by Reuss
average (fine mixed fluids) in T=60 C and different pressures (16 to 40 MPa) (P and T for
Nisku aquifer condition, WASP project; Nowroozi, 2014)

6.10 The physical properties in the matrix

For the P-wave velocity calculation after the gas injection (by Gassmann’s equation), we need

the bulk modulus of the minerals. For this purpose, mineral components of rock should be

distinguished. Some laboratory technics as X-ray diffraction or Fourier transform infrared analysis

are possible when core sample is accessible. Another method as well logging and Clay value

analysis are suitable. As mentioned in the section 2.5, the well log data was used for mineral

discrimination. The final result of the mineralogical study were demonstrated in Figure 2-12 and

Table 2-1.The modulus and density of the mineral are introduced in Table 6-1. The elastic modulus

of a combination of minerals was calculated in Table 6-2 by Voigt, Reuss and VRH average

methods.

162
 Table 6-1. The fraction of the minerals in the reservoir based on the well log data
analysis.

Mineral Fraction Bulk Modulus (Gpa) Shear Modulus (GPa) Density (gr/cm3)
Quartz C 40% 37.4 41.14 2.65
K Feldspar 4% 65.41 27.54 2.64
Albite 8% 55.94 30.17 2.61
Kaolinite 15% 46.01 23.89 2.439
Chlorite 7% 165.02 52.1 2.839
Illite-Smectite 11% 35.72 17.8 2.546
Siderite 15% 116.01 48.06 3.75

Table 6-2.The mixed minerals bulk and shear modulus calculated by three average
methods.
Average Bulk Modulus Shear Modulus
Method (Gpa) (Gpa)
Voigt Ave. 61.84 36.37
Reuss Ave. 48.09 32.52
VRH Ave 54.96 34.44

The average density value is calculated by the Eq. 6-1, and it is equal 2.87 gr/cc for a zero-

porosity sample. For the average 15% porosity in the reservoir, the formation bulk density is 2.439

gr/cc.

For quality control of the mineral discrimination, the bulk density was derived based on the

mineral study and porosity(density) well log and compared with the density from log data. The

result is shown in Figure 6-19. It shows the difference between density calculated by the well log

data and mineral discrimination study and the overestimation of density can be because of higher

percentage estimation of heavy minerals (as siderite).

163
 Error (%) of calculated density from
mineral analysis

Error (%)
0 1 2 3 4 5
295
295.5
296
296.5
297
297.5
Depth (m)

298
298.5
299
299.5
300
300.5
301
301.5
302

Figure 6-19. The error in density calculated by the mineral discrimination method and well
log data in the injection horizon.

6.11 Fluid substitution

Gassmann’s equation is a theoretical approach that relates saturated bulk modulus to bulk

modulus of the mineral matrix (mono mineral), bulk modulus of the fluid, bulk modulus of the

porous rock frame and porosity. The first introduction of Gassmann’s equation can explain as Eq.

6-16.

K dry
(1  )2
K min
K sat  K dry 
 1   K dry
  2
K fl K min K min
Eq. 6-16

164
 Where:

𝐾𝑠𝑎𝑡 = The saturated bulk modulus (undrained of pore fluids)

𝐾𝑑𝑟𝑦 = The bulk modulus of the dry porous rock = frame

𝐾𝑚𝑖𝑛 = The bulk modulus of the solid rock matrix material

𝐾𝑓𝑙 = The bulk modulus of the fluid saturating the porous rock

𝜑 = The porosity of the rock.

6.12 Model Assumption

There are some considerations for successful use of Gassmann’s theory. These assumptions are:

1. The porous rock is homogeneous and isotropic. It means frame must be formed of one

mineral or if the frame has more than one mineral, they should have a near elastic

stiffness (Berge, 1998).

2. The pores are interconnected (no isolated pores). The pore space is completely

connected, and fluid should be moveable, and fluid pressure must be uniform. It

considers one pores type, and more types of pore need to use more complex model

(Berryman and Milton, 1991).

3. Skeleton grains, fluids obey Hooke’s law (stress is proportional to strain), and the pore

fluid is frictionless (low-viscosity fluid).

4. Relative motion between fluid and solid during the passage of an elastic wave is

negligible (low frequencies only)

5. The pore fluid does not interact with the solid material (the matrix elastic moduli are

unaffected by fluid saturation).

165
 6. The system should be closed, and no fluid leaves the rock volume. No cavitation

occurs, no separation at contact boundaries.

p and s-wave velocities are controlled by shear (µ) and bulk modulus (K) as Eq. 6-17 and Eq.

6-18.

4
K 
vp  3

Eq. 6-17

vs 

Eq. 6-18

For two last formulas, if velocity is km/s and density in gr/cc, K and G will be in Gpa.It is

assumed that in the fluid substitution procedure, the shear modulus of the formation stays constant

as the fluids shear modulus are always zero.Another form of Gassmann’s equation is useful for the

direct velocity calculation for the fully fluid saturated porous rock is as Eq. 6-19 and Eq. 6-20 (for

Vs) (Gerritsma, 2005):

12
 K dry  4b 3  n 2 M 
vP   
  sat 
Eq. 6-19

12
 
vS   b 
  sat 
Eq. 6-20

Where:

166
  K dry  1
n  1   and M ;
 K min   1   K dry
  2
K fl K min K min
in which

K dry , K min or K min eral , K fl ,  were described above (Eq. 6-16)

dry  sat  the shear modulus of the dry porous rock

sat =  b = the density of the saturated rock;  sat   fl  1     m

 fl = the density of the fluid saturating the porous rock

 m = the density of the solid matrix material.

6.13 Practical usage of the Gassmann’s equation

The first parameters for a successful and correct usage of Gassmann’s equation are the wave

velocities (Vp and Vs) and density. These three parameters lead us to the shear and bulk modulus

calculation (Eq. 6-21) with a variable displacement in Eq. 6-17 and Eq. 6-18 as:

  Vs2
4
K   (V p2  Vs2 )
3
Eq. 6-21

As we know, shear modulus for fluids are zero, and it remains constant during fluid

substitution (Eq. 6-22).

167
  fl  0
1  2
Eq. 6-22

Figure 5-5 shows the result of modulus calculation by the well log data in the reservoir zone.

Kdry is a parameter that is unknown. To solve Gassmann’s equation, we need to remove this term.

Gassmann’s equation can be revealed as Eq. 6-23.

K sat K fl K dry
 
K min eral  K sat  ( K min eral  K fl ) K min eral  K dry
Eq. 6-23

In this equation, the last term is a combination of Kdry and Kmineral. It is supposed that Kdry and

Kmineral are constant during fluid substitution procedures, so the last term remains constant within

the fluid substitution procedure. This point can help us to explain Eq. 6-23 as Eq. 6-24 that is easily

applicable for the fluid substitution.

K sat1 K fl1 K sat 2 K fl 2

  
K min eral  K sat1  ( K min eral  K fl1 ) K min eral  K sat 2  ( K min eral  K fl 2 )
Eq. 6-24

So, with this simple equation, the p wave velocity is available by calculation of new saturated

bulk modulus. The bulk density before the injection is known with log data, and it is a combination

of mineral and fluid (here brine) density as:

168
  fluid  min eral (1   )  bulk
Eq. 6-25

Moreover, mineral density is calculable by using porosity and density log data, and brine

density for the initial reservoir condition (P=30 bar, T=13oC) explained in the section 0.

6.14 The CO2 gas injection effect on the formation velocity in the field and lab test

In this section, I review the results of velocity change due to CO2 injection from lab test

(Figure 6-20 (Smith,2003), Figure 6-22 (Alemo et al., 2011) and Figure 6-23 (Wang, 2001)), time-

lapse result from a field work(Figure 6-21 (Lumley, 2010)) and the velocity variation by

Gassmann’s equation (Figure 6-20and Figure 6-21) in the previously published papers. I use and

compare our result for the FRS project with these research results. The velocity calculation is

possible by three different methods:

1- Fluid substitution equations as Gassmann’s method (Smith, 2003).

2- In field estimation by injection CO2 in the real reservoir and seismic test (Lumley, 2010).

3- The lab experiences on CO2 injection on core samples (Alemo et al., 2011).

169
 Figure 6-20. The influence of mixing method on the P-wave velocity. Reuss average is
suitable for a fine mixed fluid, and the velocity change, in this case, is very dramatic in the low
saturation of CO2.Over 15% of CO2 saturation there is a slight increase in the velocity of the
formation (a test with the low-frequency laboratory data). The Voigt average is for a patchy
mixing, and the velocity change is almost a linear decrease with saturation (a test with the
high-frequency laboratory data) (Smith, 2003).

170
Figure 6-21. P-wave velocity versus CO2 saturation from a field study. The blue dots show the
field data measurements from time-lapse well logs at the Nagaoka site in Japan (Lumley,
2010).

171
Figure 6-22. Results of lab test for the CO2 injection into a sandstone (Alemo et al., 2011)

Figure 6-23. Laboratory and theoretical experiences for CO2 and a water flood effect
on the P-wave velocities (Wang, 2001).

172
 The results of the field time-lapse and lab test show a P-wave velocity change due to CO2

injection near to VRH or Brie’s averages (in the 0-10% CO2 saturation range it is the lower

boundary, see Figure 6-21) that shows a semi-patchy mixed fluid type (Lumley, 2010), (Wang,

2001), (Alemo et al., 2011). So, in the FRS time-lapse seismic estimation, it is expected to see a

semi-patchy mixed type in the reservoir according to the research. For the synthetic seismic

models, we considered all mixed models for the P-wave velocity. Comparing the acquired field

4D seismic data with the synthetic seismic data in the various mixed conditions can reveal a

realistic fluid mix type in the reservoir.

6.15 Time delay caused by the injected fluid

As it was demonstrated in the last section (as Figure 6-22), the P-wave velocity decreases

with the CO2 injection in the reservoir with brine content initially (except over 50% CO2

saturation for the fine mixed fluid). The time delay for a P-wave passing from the reservoir

horizon can explain as Eq. 6-26.

V2  V1
T  Z ( )
V2V1
Eq. 6-26

Where Z is the thickness of the injection target, V1 is the initial velocity of the formation and

V2 is the formation velocity after injection. For a reservoir with n horizons (or cells) can be

described as a summation of Eq. 6-26 as Eq. 6-27 (i is the number of the horizon).

n
V2i  V1i
T   Z i ( )
i 1 V2iV1i
Eq. 6-27
173
 The time delay caused by the CO2 injection in the below levels of the FRS reservoir is

clearly demonstrated in Figure 7-27.

6.16 The formation velocity and density after CO2 injection in the FRS reservoir

The injection of the CO2 can change the acoustic attributes. The compressional wave velocity

is decreasing by two effects: a- The bulk module of the injected CO2 is lower than the primary

pore fluid (brine). b- The effective pressure has a reverse relation with the velocity; CO2 injection

can decrease the effective pressure and velocity.

The bulk density of the fluid and consequently the formation decrease after gas injection. This

change in the density can cause a slight increase in the shear wave velocity. The density calculation

method is based on the well log density and porosity and calculated fluid density. Figure 6-26

shows the variation of the density and Vs by the injection as a function of CO2 saturation (fraction).

The diagrams that demonstrate the result of the fluid substitution by Gassmann’s method is in

Figure 6-24 for Voigt average and Figure 6-25 for the four average methods. Based on the upper

and lower boundaries (Voigt and Reuss) it is possible to estimate the maximum and minimum

velocity of the formation versus the gas saturation. Also, we calculated two models in the middle

of the Reuss and Voigt boundaries (VRH and Brie) in this section, and we will introduce the 2D

model for Vp in each year based on the mixed fluid type (Reuss (fine mixed) as lower boundary,

VRH, Brie, and Voigt (patchy) as the upper boundary).

174
 Figure 6-24: Bulk modulus for CO2 and brine mixed phase.

Figure 6-25. P-wave velocity after CO2 injection in the reservoir calculated by
Gassmann’s equation, the shape of Vp diagram is a function of an average method for the
fluid mix (CO2+brine) properties calculation. The maximum possible CO2 gas saturation in
the FRS reservoir can reach to 50%.

175
Figure 6-26: The physical properties (S-wave velocity and density) change as a linear function
of the CO2 saturation in the reservoir condition in FRS project.

6.17 The velocity and density model based on the rock physics study

The dynamic parameters of the fluid simulation were calculated in the last chapter. The result

of the rock physics study can help us to translate the gas saturation and pressure in each cell to the

seismic related physical parameters (velocity and density). The density was estimated by a linear

function, and s-wave velocity is a secondary parameter that fluctuates with the density change.

The gas injection in the reservoir can decrease the density up to 2.9% so that the shear wave will

increase up to 1.5% in the gas saturation equal to 50%. The shear wave velocity and density change

in the plume (after 1 to 5-year injection) were demonstrated in Figure 6-27. For this purpose, we

used a linear converter as Figure 6-26 (for the density and Vs) on the matrices from fluid simulation

procedure that shows the gas saturation in 2D dimension.

176
 The P-wave velocity has a different calculation method that introduced in section 6.16.

For each kind of fluid mix and average method, a P-wave velocity model was calculated (Figure

6-28 for the fine mixed (Reuss average), Figure 6-29 and Figure 6-30 for the semi-patchy mix

(Brie and VRH average) and finally Figure 6-31 for the fully patchy mix (Voigt average)). As

mentioned in the last section, the type of fluid mixture (can be patchy or fine mixed) has a huge

effect on the P-wave velocity. The maximum range of P-wave velocity change by the CO2 injection

in the Basal Belly River sandstone is up to 15%, in the fine mixed type (fine mixed) for the 20%

gas saturation. This condition for the fully patchy condition is about 2% variation in Vp for a gas

saturation equal to 20%. As motioned in the last section, Lumley (2010) demonstrated a semi

patchy condition with the CO2 injection in Nagaoka field (Figure 6-21), so the field result shows

that Vp is near to VRH and Brie’s average. However, in this research for an accurate estimation,

all mixed conditions will be tested in the next chapter for time-lapse seismic modeling and

imaging.

177
 Year 1

Year 2

Year 3

Year 4

185 m
Year 5

Figure 6-27: The density, and shear wave velocity change during the gas injection by
the constant bottom hole pressure (49.4 bar for five-year). The shear modulus remains
constant after injection, but decreasing in the density can make a small increase in the Vs
value.

178
Figure 6-28. The P-wave velocity model based on the Reuss average method that shows a fine
mixed fluid type after 1, 3 and five years’ injection. This model shows a uniform velocity
change in the reservoir volume.

179
 185 m

Figure 6-29. The P-wave velocity model based on the Brie’s average method that shows
a semi-patchy mixed fluid type after 1, 3 and five years’ injection.

180
Figure 6-30. The P-wave velocity model based on the VRH average method that shows
a semi-patchy mixed fluid type after 1, 3 and five years’ injection.

181
 184 m

Figure 6-31. The P-wave velocity model based on the Voigt average method that shows
a fully-patchy mixed fluid type after 1, 3 and five years’ injection.

182
Chapter 7. SEISMIC IMAGING

7.1 Introduction

The goal of the seismic studies in the last century was the exploration and imaging for

predicting promising structures to drill production wells in the low-risk location. Currently, the

reservoirs mostly have been explored, and they are in the production stage. Now the seismic studies

can help to characterize the reservoir parameters and time-lapse variation of the reservoir’s

dynamic parameters during production/injection.

We now focus on the interpretability of seismic study with considering dynamic parameters

of the reservoir and plume size and geometry. Also, we examine the influence of the acquisition

configuration including surface seismic, VSP and Cross Well surveys. In the dissertation, I used

advanced methods for the seismic modeling and imaging included acoustic forward modeling and

Reverse Time Migration (RTM). For this purpose, I improved finite difference Matlab codes for

modeling and RTM, and made it possible to have flexible source and receiver locations. The code

makes it possible to control and check the influence of the acquisition geometry on the seismic

response of a reservoir for a successful time-lapse program. Also, code was developed to import

geomodel data from Petrel and simulation data from ECLIPSE. By this code, the velocity (P-wave

velocity by Gassmann’s equation) and density were calculated and located in the corresponding

cells. We used synthetic velocity models to make seismic model and to image and compare seismic

responses of a reservoir with different CO2 saturations, pressure, and the plume size for various

saturation types and for a different kinds of acquisition geometry.

The content and research topics in this chapter will include:

183
 1- Introduction to the forward modeling and RTM.

2- The seismic response for the different plume size and saturation in a reservoir.

3- The acquisition configuration influence in a time lapse study.

4- The seismic time-lapse models for the FRS project with various types of mixing of the

reservoir fluids.

5- The reservoir dynamic parameters for time lapse studies and seismic models.

6- The elastic response of the reservoir.

7.2 The research method

In this chapter, the forward seismic modeling and RTM methods, the problems and noise

associated with the RTM algorithem and some methods for noise reduction are described. The

seismic modeling and analysis of the reservoir were assessed by seismic finite difference time

domain (FDTD) modeling based on an acoustic velocity-stress staggered leapfrog scheme. The

FDTD is second order in time and fourth order in space within a Central Finite Difference (CFD)

framework. The boundary conditions are stablished at all edges of the input model except at the

surface, based on a Perfectly Matched Layers (PML) approach.

Based on the synthetic models, there is an amplitude change in the reservoir and a time delay

in the deeper levels because of velocity changes that result from CO2 injection. The effect of the

time delay is removed after depth migration with an accurate velocity model. As mentioned, the

seismic models include surface seismic, VSP and cross well surveys. The well seismic surveys

show high amplitudes due to gas injection, and because of lower noise content in these methods,

we expect to map the reservoir properties in the early injection step (even in the patchy mixed

condition) by well seismic acquisition. The surface seismic models show lower amplitudes than
184
the well seismic methods after injection. The source of the main noises as traffic, wind,

electricity lines are on the surface.So the monitoring by surface seismic methods may not give a

proper result in the first years of injection (for a low-velocity variation in the reservoir due to CO2

injection) when saturation and plume size are small, but we will demonstrate that the surface

seismic can generate better images of a reservoir compare to well seismic.

7.3 Acoustic forward modeling strategy

The 2D acoustic wave equation can be expressed by Euler’s equation and the equation of

continuity (e.g., Brekhovskikh, 1960 and Zakaria et al., 2000). A system of first-order differential

equations regarding the particle velocities and stresses is given by Eq. 7-1.

v
x 1

p
,
t  x 
v
 Euler
p
z 1 ,
t  z 
p 2  v x v
z 
   vP   , Continuity
t  x z 
Eq. 7-1

Where p is the pressure, vx and vz are particle velocities in lateral x and vertical z

directions respectively. The parameters  and v p are density and the P-wave velocity and t is

the time. The numerical solution is based on the FDTD of the staggered grid in a leapfrog scheme.

The FDTD is 2nd order in time and fourth order in space on Central Finite Difference (CFD). The

185
Perfectly Matched Layers (PML) boundary condition of Zhou (2003) is used for all edge of

the model except the surface.

7.4 RTM migration strategy

The RTM include three simultaneous imaging conditions given by Eq. 7-2:

Tmax
I u ( x )   Su (t , x ) Ru (t , x ) dt ,
0

Tmax
I v ( x )   S v (t , x ) Rv (t , x ) dt ,
0

Tmax
I p ( x )   S p (t , x ) R p (t , x ) dt ,
0
Eq. 7-2

Where I( x ) is the migrated image in the subsurface coordinate x  ( x, y,z ) , T is

max

maximum recorded time, S( t , x ) is the forward propagated source and R( t, x ) is the backward

propagated receivers. The subscripts p , u and v correspond to three images for pressure and

displacements obtained by the imaging conditions. Note that here, the Einstein summation

convention is not used for repeated indices. The imaging condition of the RTM algorithm is a

cross-correlation of forward propagation sources and backward propagating receivers (Whitmore

et al., 2012):

186
 1
I(X )   S ( X , t ) R( X , T  t )dt
A( X )
X  ( x, y , z )
Eq. 7-3

The RTM method is a robust migration method for imaging complex geology conditions

(see section 7.17).

In this chapter, I demonstrate the seismic model results with abbreviation SM. For example, the

seismic model of the baseline and after two-years do injection will be defined as SM(base) and

SM(m2). Also, the migrated data by RTM method will be demonstrated with RTM(SM(base)).

Figure 7-1 shows the abbreviations that used for the seismic modeling and RTM.

Figure 7-1. Description of the abbreviations.

7.5 Boundary Conditions

Boundary conditions are crucial in a synthetic modeling code that solves PDE numerically.

The ordinary boundary can reflect the energy inside the model, and the result will be noisy data

for the seismic processing stage. For the purpose of noise reduction because of the recursive

wave, the Perfectly Matched Layer (PML) were include for the three internal boundaries of the

187
model. Moreover, the upper boundary of the medium is free surface bound in the modeling

code (see Figure 7-2).

Figure 7-2.The Boundary condition in the seismic model. The orange rectangles show
the internal boundaries.

7.5.1 The Perfectly Matched Layer (PML)

The first study of the PML boundary type dates back to (Berenger, 1994) in electromagnetics

computation. PML boundary condition is referred to an absorbing layer or boundary for different

kinds of the wave equation. It can make a mathematically infinite space for the wave to avoid the

wave radiating back inside the model. Thus, it can have a simulated medium condition as open

internal boundaries without any recursive waves. In the current modeling code, the absorption

procedure begins at 20 cells from the boundary. It is an optimum size to make a near zero amplitude

content for the wave in the internal boundaries.

The importance of usage PML boundary condition in the seismic modeling is related to the

decreasing noise in the migration step. The RTM migration code can not support and eliminate the

noise of the boundaries, so for an ideal result, the noise should be removed from seismic data

before migration.

188
 The formulation of split PML method (SPML) was used for this chapter. In this approach,

the wave field was split into two components (Carcione et al., 2002) as (Eq. 7-4).

p  px  pz

px v
 v2 x
t x
pz v
 v2 z
t z
v x p

t x
v z p

t z
Eq. 7-4
by adding the decaying coefficient d(u)=(d(x), d(z)) the Eq. 7-4 can be reworked as Eq. 7-5

(Collino and Tsogka, 2001):

p  px  pz

px v
 d ( x) px  v 2 x
t x
pz v
 d ( z ) pz  v 2 z
t z
v x p
 d ( x )v 
x x
t
v z p
t
 d ( z )v
z  z
Eq. 7-5

189
 Where d(x) and d(z) can define Absorbing Boundary Condition (ABC) coefficients as

shown in Figure 7-3.The ABC coefficients can be represented by Eq. 7-6 (Collino and Tsogka,

2001):

u
d (u )  d 0 ( ) 2
L
3v
d0   ln( R)
2L
u  x, z
Eq. 7-6

L indicates the thickness of PML boundary and R is usually chosen between 10-3 to 10-6.

SPML is applicable also for the elastic medium as well (Collino and Tsogka, 2001). A sample of

SPML for the elastic medium and the P-wave source was tested in the MATLAB, and the result

is demonstrated in Figure 7-4.

190
Figure 7-3. The model boundary is shown by ABCD. AB, CD, and AD have a SPML
boundary condition, and BC is a free surface boundary.

191
Figure 7-4. The different components of the wave at the PML absorption boundaries for a P-
wave source in an elastic homogenous medium.

7.6 Low-frequency noise due to RTM procedure

The RTM algorithm can generate various type of noise (Khalil et al., 2014). Low-frequency

noise is one of artifacts that can be recognized in high-velocity contrast zones. After time reversal

of the receiver wavefield, the artifacts of the RTM occur where the two wavefields are traveling

in the same direction; the inverse scattering imaging condition attenuates low wavenumber noise

(Whitmore et al., 2012). Also, the Laplacian filter is used for low-wavenumber noise reduction

(Liu et al., 2010) and (Martinez, 2016).

In this research, we use subtraction of monitor seismic model and the baseline with the first

order derivative for the noise reduction. The work procedure is:

1- Construct the seismic model including the reservoir of interest.

192
 2- Construct a baseline without any fluid substitution, and a monitor model after fluid

substitution.

3- Make a seismic model and RTM image of both velocity models

4- The difference between the seismic models as RTM(SM(m))-RTM(SM(baseline)) can

decrease the noise. This method is used in the current research for the noise reduction without

any effect on the seismic data quality.

5- A derivation of the RTM result can decrease the low-frequency noise effect. This method can

reduce the amplitude of the seismic result. This function is available in the MATLAB

software. Figure 7-14 and Figure 7-15 later in this chapter show the result of (diff) function

that used on the RTM result. The (diff) can be explained as (from MATLAB help):

diff(X), for a vector X, is [X (2)-X (1), X (3)-X (2), ... X (n)-X (n-1)].

diff(X), for a matrix X, is the matrix of row differences, [X (2: n, :) – X (1: n-1, :)].

7.7 Condition for successful 4D study

Seismic inversion can provide us four acoustic attributes including: Vp, Vs, density and Q

(Mavko, 2010). For the reservoir study, one needs to have an ideal estimate for converting acoustic

attributes to the reservoir’s static and dynamic parameters. In a time-lapse study, an interpretation

is possible by calculating the difference of seismic images during production. The first step is

acquiring seismic data before any injection called the baseline survey. The repeated seismic

acquisition and difference of the data should be interpretable. For interpretability, we need to

address two parameters in the seismic data, the amplitude change reflectivity, and time delay

193
because of the velocity change in the reservoir area. The parameters change the acoustic

properties are plume size, pressure, and saturation.

A time-lapse (4D) study needs a 3D repeatable acquisition, so for this purpose the receivers

and source points should be exactly in the same place. It means a successful 4D study needs

specific CMP points for baseline acquisition. The FRS 4D seismic design is described in chapter

3.

The factors for a successful 4D study are (Johnston, 2013):

1- Integrate reservoir data with the 4D seismic interpretation.

2- Understanding of the rock physics behind the production or injection procedures.

3- Low-noise and repeatable seismic data.

4- Accurate reservoir characterization.

5- Optimal timing of repeat surveys.

7.8 Interpretability of an event

For quantitative seismic interpretation, the amplitude and coherency of data is important for

an interpreter. An seismic event such as a reservoir or a structural shape can be interpretable by

the phase change and visible amplitude variation, and a 4D seismic data change can be represented

by:

4D seismic change = (Monitor data + ERRORm + Noisem) - (Base data + ERRORb + Noiseb)

194
 Acquisition footprint and noise can mask a weak seismic response of changes in a

reservoir in a time-lapse study. In real field acquisition, amplitude can be affected by source,

receivers, physical properties of the earth and environmental noises including:

1- The source and the recording system (arrays, type, and coupling)

2- All kinds of noise in the field (road and traffic noise, electricity line, wind)

3- Reflection coefficients of the formations and layers

4- Absorption of the formations

5- Multiples and ghosts.

In this chapter, I consider a noise-free data set for a primarily technical research about the

seismic responses of the reservoir. In the next section, the influence of the plume size and

formation saturation is tested by seismic forward modeling.

7.9 Plume size and velocity variation

The reservoir imbibition/drainage always cause a change in the fluid content and pore pressure

of the formation. The secondary effects of the fluid substitution are the velocity and density

variations. The halo of the velocity change in a reservoir is a function of the plume size and the

CO2 saturation (and the plume size is a function of the porosity and permeability). So, the velocity

change is a function of such parameters as:

195
 𝛥𝐾 = 𝑓(𝜑 , Δ𝜌, 𝛥𝐾𝑓 )
Eq. 7-7
𝛥𝑉 = 𝑔(𝛥𝐾, 𝛥𝜌)

Eq. 7-8

Plume size = h(k, Q)

Eq. 7-9

Where:

k: The permeability

ΔK: The bulk modulus variation in the formation due to fluid substitution

𝛥𝐾𝑓 : the difference of bulk modulus between initial and the secondary fluid in the reservoir

Φ: The porosity

Δ𝜌: The density change in the reservoir by the fluid substitution, this parameter is a function

of the phase saturation and porosity.

Q: A term for the production/injection strategy

7.10 A diffusive and solid velocity model

In the reservoir, through injection/production (fluid substitution), the saturation of CO2

changes spatially. During fluid substitution procedure in a reservoir model the velocity and density

variation of the cells in the reservoir network are a function of phase saturation. For this test, I

considered a typical model of a reservoir. As the result of the previous chapter (Chapter 6), a

velocity change in a reservoir can possibly result in two different versions of P-wave velocity

anomalies. The “solid” velocity model is equivalent to the fine mixed fluid saturation for which
196
the P-wave velocity is calculated by the Reuss average (non-linear). The “diffusive” velocity

model is a result of patchy or semi patchy mixed type saturation that is calculated by a linear

conversion of the saturation to the P-wave velocity (as Voigt or VRH average). I generated two

different velocity models to test the seismic response, as shown in Figure 7-5. I considered that the

fluid can diffuse semi-homogeneously in x and y-directions (the permeability in the y direction is

seen to be lower than x direction) in the model. The gravity effect was not considered in the fluid

injection in the diffusion model.

As mentioned above, the “solid” velocity pattern (Figure 7-5, right diagram) is made by a non-

linear function as the Reuss average (or fine mixed fluid type) because the velocity of reservoir

cells drop immediately after a low gas injection volume. The left diagram in Figure 7-5 is a

“diffusive” velocity model that as mentioned, velocity decreases from the center to the ellipsoid

boundary in a linear gradient.

In this section, the objective of all tests are for the diffusive and solid velocity models with

different size (plume size) and velocity/density variations.

Figure 7-5. The internal structure of diffusive (an injective or productive point in the
middle of the ellipsoid) and solid velocity models. The reduction is linear from the center to
the outer bound in the diffusive model. The unit of velocity is m/s. The dimension of the
ellipsoid will define with the big and small diameter (a,b).

197
7.11 Seismic response of a solid and diffusive velocity model

The first seismic model (Figure 7-6) compares the responses of a diffusive and solid velocity

models in a homogeneous medium. For a simple model, even small alteration of velocity in the

seismic resolution size is detectable, so this test will show the amplitude change by the solid and

diffusive velocity without considering the absolute magnitude of the amplitude change.

Figure 7-6 shows a medium with V=2500 m/s and density=2200 kg/m3 and 1000*620 m

dimension. The maximum velocity change in the center of the diffusive ellipsoid and solid shape

is 7% equal to 175 m/s. The seismic modeling code can generate pressure, horizontal and vertical

components. Figure 7-7 shows the Uz component of the seismic response; in the left diagram the

surface effects (direct and surface waves) were not included.

Figure 7-6.The velocity model with two diffusive and solid velocity changes. The velocity
change in the center of diffusive model or whole solid shape is -7% equal -175 m/s (2325 m/s).
The size is ellipsoids are (200,40m).

198
Figure 7-7.The Uz component of the seismic response of diffusive and solid models in a
homogeneous medium. The left is the SM(m), and the right is SM(m)-SM(baseline).

Figure 7-8. RTM(SM(m)) for Uz component and RTM(SM(m))-RTM(SM(baseline))

Figure 7-9. RTM(SM(m)) for pressure component and RTM(SM(m))-

RTM(SM(baseline))

199
 The results (Figure 7-7, Figure 7-8 and Figure 7-9) show that a simple velocity anomaly

with a small change in a homogeneous and isotropic medium is detectable by the seismic method.

Although impossible in the real world, the absence of noise and other reflectors in the medium is

an advantage for the detectability of the event in our model (a reservoir). The seismic response and

RTM results show a high amplitude with a clear image for the solid velocity boundary compared

to the diffusive velocity distribution.

The second experiment was for a model with solid and diffusive velocity anomalies in a simple

three-layer medium (as shown in Figure 7-10). The seismic responses are demonstrated in Figure

7-11 for pressure, Figure 7-12 for Uz and Figure 7-13 for Ux components. The processed RTM

results for the pressure and Ux components are shown in Figure 7-14 and Figure 7-15 respectively

and show that low-frequency RTM noise was eliminated. Figure 7-16 and Figure 7-17 show the

same results without removing the noise. The diffusive velocity anomaly showed a seismic

response that is weaker than the solid velocity anomaly and the amplitude in the seismic model

and migrated section is less than for the solid form. However, both shapes caused a time delay

effect under the ellipsoids in the seismic models.

200
Figure 7-10.The velocity model for a. Three-layer model as baseline b. Model a with diffusive
and solid velocity models as monitored model c. Subtracted result (Monitored-Baseline
model)

201
Figure 7-11.The pressure component of the seismic acoustic model for: a. 3-layer baseline
b. Baseline plus diffusive and solid velocity ellipsoids c. The difference

Figure 7-12. The Uz component (a.SM(base,Uz), b.SM(monitor,UZ), c.SM(base,Uz)-

SM(monitor,Uz)). It is similar to the pressure component but with the lower amplitudes.

202
Figure 7-13.The Ux component for (a.) 3-layer baseline (b.) Baseline plus diffusive and solid
velocity ellipsoids; (c.) The difference

Figure 7-14. a. diff (RTM (SM (baseline, Pressure))), b. diff (RTM (SM (m, Pressure))),
c. diff (RTM (SM (baseline, Pressure)- RTM (SM (m, Pressure)))
203
Figure 7-15. a. diff (RTM (SM (baseline, Ux))), b. diff (RTM (SM (m, Ux))),
c. diff (RTM (SM (baseline, Ux)- RTM (SM (m, Ux)))

Figure 7-16. a. (RTM (SM (baseline, P))), b. (RTM (SM (m, P))),
c. (RTM (SM (baseline, P)- RTM (SM (m, P)))

204
 Figure 7-17. (RTM (SM (baseline, Ux))), (RTM (SM (m, Ux))),
(RTM (SM (baseline, Ux)- RTM (SM (m, Ux)))

The solid velocity anomaly generates a clear amplitude change in the reservoir (or velocity)

border, and in the migrated section, the location of the velocity anomaly matches tothe real

location. For the diffusive velocity anomaly, the seismic response does not show any amplitude

change in the boundary, and after migration, only a shadow of the central point of the shape is

visible. This test shows that the dimension of the solid velocity shape is measurable, but for a

diffusive velocity, the seismic can not show the exact velocity change geometry or reservoir size.

7.12 Acquisition geometry for 4D seismic surveys

In this part, I check the acquisition geometry and its relationship to 4D seismic data quality.

The acquisition geometry can change the reservoir, imaging condition and in the real world, impact

the noise level related to surface activity. In the field acquisition, well seismic acquisition methods
205
generally help lower noise than surface acquisition. I tested three different acquisition surveys

with dense receivers:

1- surface 2D configuration

2- vertical Seismic Profile (VSP)

3- cross-well for a simple model

The velocity and density models are shown in Figure 7-18, with dimensions of 1000x620 m.

The acquisition patterns are listed in Table 7-1. The results demonstrate that for gas detection, the

well seismic methods are much reliable because the amplitudes from the reservoir will be within

the threshold range. For a better imaging condition, the shots and receivers should be out of the

gas plume. The surface seismic acquisition has a better migration aperture, and so the image of the

reservoir can be better, but the amplitude due to injection is less than from well seismic results,

due to distance from the reservoir.

Figure 7-21 shows the result of cross-well acquisition. In the results of tests, and after

reduction the RTM algorithm low-frequency noises, the cross well seismic acquisition with the

200 offset between shot and receiver wells shows a consistent image of the reservoir (Figure

7-21.d). The result of acquisition with surface seismic and VSP pattern are shown in Figure 7-19

and Figure 7-20, respectively.

206
 Table 7-1: Acquisition parameters and patterns
Acquisition Receivers Geophone Spreads Spreads Shot Record
type Spread lenght interval start point end point point lenght
Surface 2D 1000 1 (0,0) (1000,0) (500,0) 0.5 s
VSP 600 1 (600,0) (600,600) (400,0) 0.5 s
Cross Well 600 1 (600,0) (600,600) (400,295) 0.5 s

Figure 7-18. The diffusive velocity and density models for a 7% and 3% change in the
ellipsoid shape. The ellipsoid dimensions are180m wide and 10m in thickness

Figure 7-19. The seismic model (a) and migrated section (b) for the surface survey.

207
 Figure 7-20. A seismic record and migrated section for the VSP acquisition

Figure 7-21. a. The seismic response of cross well acquisition pattern of the model in Figure
7-18. b. After eliminating the surface and shot effects. c. The migrated data (from a). d. The
noise reduced migrated section.

Conclusions: The acquisition geometry can have a significant effect on the seismic response.

The surface acquisition has a better imaging condition, and the boundary of the reservoir can be

recognized properly. However, the acquired amplitude level of reflections from the reservoir in
208
the surface seismic is lower than for the well seismic methods. Thus, the surface seismic

method can be a reliable method for the large production/sequestration fields with the significant

change in the saturation in the reservoir. For the small fields and the reservoir activities with small

saturation change, the well seismic methods are a better choice for the reservoir characterization.

Of course, the low level of the noise content in the well seismic acquisition can help to detect a

lower saturation level and velocity change in the reservoir.

7.13 Seismic interpretability of a diffusive velocity model

The saturation value and effective pressure (difference between confining pressure and pore

pressure) are two parameters that play the leading role in the velocity change in the reservoir. As

mentioned in the previous chapters, the pressure change in the FRS project is not large(<2MPa),

so the pressure effect on the velocity was included in the velocity modeling.

Other parameters with an effect on the seismic response are plume size. In this section, I

investigate the saturation (or velocity) offsets and the plume size influence on the seismic modeling

results.

Figure 7-23 (column a) shows two diffusive velocity ellipsoids with same central velocity

change and different size. Columns b and c show the seismic response and migrated results of the

plumes with various dimensions. As we expected, the bigger plume shows the greater response

with the same amplitude and a larger anomaly has more chance to be detectable.

Conclusions: In a homogenous media, a small change in the velocity or density may be

detectable by the seismic reflection method (Figure 7-22.a1, b1, c1). When we work with the real

earth, we deal with a non-homogeneous and anisotropic earth and the imaging is more challenging.

209
 Figure 7-22: The seismic response (column b) and RTM result (column c) for a model
(100 * 20 m) with a different velocity anomalies (column a). Higher velocity difference
causes greater amplitudes for the surface acquisition survey.

210
 Figure 7-23: The seismic response (column b) and RTM result (column c) for two
models with a different plume size and 5% velocity change in the center of ellipsoids in
column a. a1: 50*20 m and a2:200*20 m.

7.14 Seismic imaging for FRS project

From the rock physics study and based on the reservoir simulation results, we modeled each

cell in the seismic model for different fluid saturation types. The velocity geomodel (generated by

a well data with fine grid size) that was introduced in Section 4.15 is the baseline model for the

seismic time-lapse study of the FRS project. The editing of the baseline geomodel based on the

reservoir simulation data (section 6.17) it is possible to calculate an accurate velocity and density

models for the particular time of the injection (Figure 7-26). For the P-wave velocity change, we

defined four different models based on the upper (Voigt) to the lower (Reuss) boundaries and VRH

and Brie’s average for the Vp change by the injection (Figure 6-28 to Figure 6-31).

In the previous section, we compared the solid and diffusive velocity anomalies and the

seismic response of them. The different kind of velocity calculations from the rock physics
211
methods will generate different velocity shapes in the reservoir as either diffusive or solid

anomaly shapes. The Reuss average (fine mixed fluid type) has a large velocity decrease for low

CO2 saturation levels and this average will make a solid velocity anomaly in the reservoir.

The surface and well seismic acquisition (VSP and cross well), have been modeled and data

were processes through to final images. For a realistic image for a single shot VSP, we need to

have precise survey design parameters to correctly image the plume size and geometry after

migration. It means a wide distance between the shot to receivers is needed to obtain enough data

and CMP from the entire reservoir. Also, it helps to have enough migration aperture for the

migration process. A short interval between the shot to receivers may generate a pour imaging

condition for the modeling and migration. Figure 7-29 is a sample for a weak design pattern for

the VSP with a shot that does not clearly image the real plume geometry.

212
 a

b c

Figure 7-24. The baseline P (a) and S-wave (b) velocity (m/s) and density (c) (gr/cc) models.
The reservoir saturation effect on the velocity and density after injection is included in these
models.

The main comparison in a time-lapse study is between the seismic monitor surveys (generated

with the specified model in defined time) with the baseline survey. For the first step, the velocity

and density model (Figure 7-24) and the migrated section for the baseline data (Figure 7-25) were

generated.

In the second phase, Figure 7-26 shows how we include the velocity and density variations to

the baseline models. The seismic response of the VRH model for different acquisition

configurations as shown in Figure 7-27 for a multi-shot surface seismic, Figure 7-29 for a single

shot VSP and Figure 7-31 for a single shot cross well experiment. For the well seismic experiences,

213
the receiver are from the surface to 600 m depth, and the offset between shot and the receivers

in the well is 200m.

The imaging condition is better for the surface configuration, but the well seismic data show

a higher amplitude as we expected. The well seismic geometry has a better result for tomography

and velocity estimation especially with the patchy fluid mixed condition or very low gas saturation.

The surface experiment shows an excellent image of the geometry of the plume as we see in Figure

7-27.f.

Figure 7-28 compares the amplitude change due to injection for different injection years.

Figure 7-25. The migrated acoustic Uz component for the baseline velocity and density
model (surface seismic, five shots and 996 receivers).

214
 Figure 7-26: The velocity and density models before and after five years’ injection with
a BHP=49.4 bar in the gas phase for the VRH average. The original physical properties
oriented by the seismic interpretation result. a. The base model before injection. b. The
perturbation model base on the saturation results. c. The physical properties after
injection. d. The magnified figures on the reservoir zone.

215
 Figure 7-27: The seismic model generated by the velocity and density patterns
introduced in Figure 7-26 (VRH average) for a surface seismic experience with one shot in
x=500 m and receivers with 1 m interval and from 0 to 1000 m. a. Baseline seismic model.
b. Baseline RTM result. c. Monitor seismic model. d. Monitor RTM result. e. The
difference between monitored and baseline seismic models (amplitude ten times magnified).
f. The difference between RTM results (amplitude ten times multipled).

216
Figure 7-28.The amplitude of the seismic acoustic seismic modeling (Figure 7-27.A, section
AA’ on the red line) for the baseline, after one, three and five-year injection (VRH average
method).

217
 Figure 7-29: The seismic model generated by the velocity and density patterns
introduced in Figure 7-26 (VRH model) for a VSP survey with one shot at x=400 m and
receivers with 1 m interval at x= 600 and extending from 0 to 600 m depth. a. Baseline
seismic model. b. Baseline RTM result. c. Monitor seismic model. d. Monitor RTM result.
e. The difference between monitored and baseline seismic models. f. The difference
between RTM results.

218
Figure 7-30. A VSP seismic model and image for a wide source to receivers aperture (400 m
distance). a and b show the seismic model and image for the baseline and c and d are for the
five-year injection calculated by Reuss average, d and e are the difference of the 5-year
injected model and baseline (wavelet: 55 Hz Ricker).

219
Figure 7-31. The seismic model generated by the velocity and density patterns introduced in
Figure 7-26 for a Cross-Well survey with one shot at x=400 m and 295 m depth and receivers
with 1 m interval at x= 600 and extending from 0 to 600 m depth. a. Baseline seismic model.
b. Baseline RTM result. c. Monitor seismic model. d. Monitor RTM result. e. The difference
between monitored and baseline seismic models. f. The difference between RTM results.

220
 Figure 7-32: The time lapse seismic models: a. The seismic model for one-year
injection. b. The difference between the baseline and (a). c. The difference between seismic
models after five and one year of injection. d. Migrated section of (a). e. The difference of
migration sections between the baseline and one year of injection data. f. The difference of
migrated data between five years and one-year of injection.

For the Reuss, Voigt and Brie average methods, the seismic models and migrated sections

were generated for the first and fifth year of injection (Figure 7-33 and Figure 7-34). The

comparison of the migrated sections shows the difference in the seismic response for the different

average type (mixing form). The Reuss average presents a fine mixed fluid of brine, and CO2 will

be detectable after one year of injection. The Voigt average yields a weak seismic response, and

the RTM result can not show a detectable amplitude after one year of injection (in Figure 7-32 the

amplitude for the Voigt average was magnified ten times). The other average methods (Brie and

VRH) show verysimilar results, and the seismic response of them is interpretable if we consider

noise free data without any acquisition footprint.

221
 Figure 7-33. The migrated seismic data from the reservoir’s response by different kinds
of average related to the mixed fluid condition after a year injection. The left figures are
the model made by the rock physics models after one year injection for the Reuss(a), Brie
(b), VRH (c) and Voigt (d) averages.

222
 Figure 7-34. The migrated seismic data from the reservoir’s response by different kinds
of average related to the mixed fluid condition after 5-year injection. The left figures are
the model made by the rock physics models after five years injection for the Reuss(a), Brie
(b), VRH (c) and Voigt (d) averages.

223
 After five-years do injection, the plume diameter is around 185 m and the seismic

responses for three average methods (Reuss, Brie, and VRH) are recognizable. The amplitude for

the Reuss average is higher than other methods because the reservoir made a solid velocity

anomaly.

The thickness of the seismic response shows a thicker event, because of wavelet shape (Ricker

wavelet- 45 Hz) but the horizontal size of the seismic response is equal to the reservoir size (as

Figure 7-35). The size of the reservoir in the real field data may be smaller than real reservoir size

seismic due to limits of seismic resolution. The velocity variation of the difference of the migrated

seismic response after five-years of injection is shown in Figure 7-35. The size of the difference

migrated anomaly shows a close match with the plume size for the surface acquisition survey.

Figure 7-35.The variation of velocity due to injection after 5-year injection (calculated
by Reuss average) and the time-lapse seismic migrated response of it (RTM(SM(5,R)-
RTM(SM(base)))
224
 Conclusion: The CO2 saturation in the reservoir simulation is limited to being less than

50% because of trapping efficiency. The velocity and density changes for patchy or semi-patchy

mixed type are 5 to 12% with a diffusive velocity anomaly. The statistical distribution for a patchy

average shows a normal Gaussian distribution form with a high variance, and it can yield a weaker

seismic response compared to a solid velocity ellipsoid created by a fine mixed fluid with a small

variance (Reuss average) (Figure 7-36 and Figure 7-37).

Figure 7-36.The statistical distribution of the velocity change in the reservoir cells in the
patchy mixed (Voigt average).

225
Figure 7-37. The statistical distribution of the velocity change in the reservoir’s cells in the fine
mixed (Reuss average).

7.15 FRS reservoir seismic time-lapse results

In this section I attempt to check the validity of seismic inversion to predict the CO2 saturation

condition in the reservoir. The seismic time-lapse results are generated by subtracting the baseline

seismic from the seismic data. I try to make a reservoir time-lapse result based on saturation, and

we will make synthetic seismic model and RTM image. Finally, the result of seismic data for both

cases was compared.

Figure 7-38 shows the work flow for the saturation modelling. The test are done by a surface

2D survey with one shot at the well position and receivers over a 1-kilometer spread with a 1m

interval. Figure 7-39 shows the difference of velocity in the reservoir between the injection years

in the VRH average model.

The seismic response in this test is a function of:

a- The gas plume growth speed

226
 b- The average methods used for the velocity calculation and fluids mixing types.

The seismic model based on the reservoir changes over time is very sensitive to the average

mixing method. The current study uses the VRH average method for the velocity estimation, and

because it is a semi-linear function (Figure 6-25), the difference in the seismic images are

negligible over the natural amplitude range (Figure 7-40, Figure 7-41, Figure 7-42 and Figure

7-43). The entirely linear function for the velocity change (Voigt average) shows a high

compatibility with the seismic time-lapse results. The fine mixed fluid type (Reuss average

method) can show a significant difference between seismic time-lapse model and seismic made by

the reservoir changes.

227
 A

Baseline
Base Model of the
Seismic
Reservoir model
Model Seismic
Time
Lapse
The Reservoir The Seismic
Model after one- Model after
year Injection one year Seismic
Time
The Reservoir The Seismic Lapse
Model after n- year Model after n
Injection year

The Reservoir
Model after 1- year
Injection Reservoir
Seismic
Time
Model
Lapse
The Reservoir
Model after 2- year
Injection
Reservoir
Seismic
Time
Model
The Reservoir Lapse
Model after n- year
Injection

Figure 7-38: The research routine to compare results of the seismic and reservoirs time lapse
surveys. A. shows the direct seismic time-lapse, B. seismic time-lapse based on the reservoir
time lapse

228
Figure 7-39. The difference model (time lapse) for the p wave velocity (by VRH average) in the
reservoir between different years of injection. The result calculated according to the CO2
saturation content and Gassmann’s equation for a semi-patchy mixed condition.

229
 Figure 7-40. Left: SM (R (5-year injection))-SM (R(1-year injection)) and right: SM (R ((5
year) - (1-year injection))). As mentioned previously, SM stand for seismic model (Acoustic),
and R is calculated Vp based on reservoir simulation result

Figure 7-41. The difference between two model in Figure 7-40, the left figure shows same
amplitude scale and the right one is 100 times magnified amplitude

Figure 7-42. The RTM results for the seismic models in Figure 7-40

230
 Figure 7-43. The difference of RTM images in Figure 7-42. The left shows the difference in
natural amplitude and the right figure shows 100 times magnified.

Figure 7-44. The saturation and P-wave velocity change distribution after a year stopping the
injection.

Conclusion: the reservoir’s dynamic parameters are convertible to the seismic response and

vice versa if there is a linear function between the saturation and velocity change. The patchy and
231
semi-patchy mixed type are good examples for this linear or semi-linear conversions. Figure

7-44 shows an influence of a converter in the new population distribution and variance.

7.16 The elastic medium and the seismic response of the reservoir

The previous results were for acoustic models, in which we ignored the shear modulus and S-

wave propagation. The wave equation with S-waves can be written as (Eq. 7-10):

v  
x  1 ( xx  xz ), 
t  x x

v  

z  1 ( zx  zz ) 
t  z z 
 v v
xx  (   2  ) x   z


t x z 
 v v
zz   x  (   2  ) z  Eq. 7-10
t x z 
 v
xz   ( x  z )
v 
t x z


The MATLAB seismic code was extended to model elastic waves with FDTD method, and in

this section, the seismic models of the CO2 injected are compared with the baseline in the FRS

project. Initially, a simple three-layer model was tested to compare acoustic and elastic seismic

modeling (Figure 7-45). The source is a P-wave at 4 m depth that by reflecting at the surface layer,

creates a converted S-wave. The response of PS and SS waves are zero at zero-offset due to no

conversion (by solving Zoeppritz’s equation, the reflection coefficient of PS-wave in the zero

offset is always zero). However, the amplitude of SS-wave is considerably high at far offsets.

232
 Figure 7-46 is a seismic model for a shot in an acoustic medium and the reservoir event

is a PP-wave at the <0.4 s. Figure 7-47 demonstrates the elastic medium response of the reservoir

for a shot position 250 m from the well site (incidence angle with the reservoir body is equal 40

degree) to observe clearly the response of PS and SS-waves due to the CO2 injection (see Figure

7-47). As it can be seen in Figure 7-47, the SS-wave response has a strong amplitude compared to

the PP and PS-wave response.

Figure 7-45. The acoustic (left) and elastic seismic response (right) for a three-layer
model (top).

233
 Figure 7-46. The P-wave seismic response for the acoustic wave propagation. a. shows the
seismic response for the baseline model. b. the seismic model after five-year injection by Brie’s
model. c. the difference section shows a PP response of the reservoir.

234
 Figure 7-47. The seismic response for an elastic model. a. baseline. b. after five-year
injection by Brie’s model. c. The difference section and PP, PS and SS seismic response of
the reservoir. As demonstrated in Figure 7-45, the SS-wave amplitude is considerable at far
offsets.
235
7.17 Seismic response of CO2 injection in a complex geological setting

The geomodel and velocity model in the FRS project is simple with the flat formation

condition. In this section, there is an attempt to fix a CO2 reservoir in a layer in the Marmousi

model (the original Marmousi model was shown in Figure 7-48 and with a reservoir in Figure

7-49) to find out the seismic response. This reservoir is placed at 550 to 800 m depth, and the

properties are matched with a CO2 injection with 40-50% gas saturation. The primary reservoir

fluid is considered as a brine with 8000 ppm salinity (same as the FRS). The acoustic seismic data

were migrated (as Figure 7-50 and Figure 7-51). The difference section between the CO2 injected

model and baseline is shown in Figure 7-52. The reservoir location is recognizable with acceptable

geometry. The complex geology area can be adequately mapped by an advanced migration method

(RTM) with a high-quality acquisition, and the change in a reservoir activity with higher than 10%

change in the reflection coefficient is recognizable by the seismic surface method.

Conclusion: An accurate seismic model and image in the complex geology set can be

generated by the seismic forward modeling and RTM migration method. The main goal of the

reservoir simulation study is to make a predictable velocity model by the rock physics roles and

purpose of the seismic study is to create a velocity model in the seismic resolution range to

calculate the geometry of plume and migration and saturation.

236
 0 5500

5000
500

4500

1000
4000

1500 3500

3000
2000

2500

2500
2000

3000 1500
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Figure 7-48.The original Marmousi model (P-wave velocity)

0 5500

5000
500

4500

1000
4000

1500 3500

3000
2000

2500

2500
2000

3000 1500
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Figure 7-49. A new physical property (Vp) defined as a CO2 injected reservoir pointed by the
red rectangle.

237
 0 50

40
500
30

20
1000
10

1500 0

-10
2000
-20

-30
2500
-40

3000 -50
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Figure 7-50.The seismic imaging result on the original Marmousi model. The acoustic
wave forward modeling and RTM migration method was used.

Figure 7-51.The seismic imaging result for Marmousi model and the implemented reservoir.

238
 0 100

80
500
60

40
1000
20

1500 0

-20
2000
-40

-60
2500
-80

3000 -100
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Figure 7-52.The subtract of monitor seismic model of the baseline model. The red
rectangle shows the location of the reservoir.

239
Chapter 8. Conclusions and the recommendations for future work

The project is a full geological, geophysical and engineering studies about the CO2 injection

in the shallow reservoir with focus on the CO2 plume migration and leakage detection by the

available methods. The project area is 20 km southwest of Brooks in Alberta. The first injection

target is the Basal Belly River Sandstone (BBRS) in 300 m depth and P=3 MPa and T=13 oC and

brine salinity S= 8000 ppm. The homogeneity in the BBRS layer around the injection well and

simple structural geometry with no fault and fracture in the project area can help researcher to

establish geophysical procedures to explain the plume migration and possible leakage by the

empirical examination and mathematical formulation.

In the first step the seismic design for seismic time-lapse research was evaluated by the

attributes study of the acquisition parameters. The 3D-3C seismic baseline data were acquired and

processed data shows a high resolution seismic image in the reservoir level. The baseline seismic

interpretation shows horizontal layering for the reservoir strata and adjacent formations. The

seismic attribute studies identified the absence of major fracture and faults in the BBRS. From a

very detailed interpretation a reliable geometric frame for the geomodel was constructed. Well log

data (10-22) is the main information source for the geomodel. This well is main object for injection.

According to the depth of the injection zone and low temperature and pressure. The black-oil

simulation was not appropriate for the study, so the compositional method was used for the fluid

simulation. The potential for a gas to fluid CO2 phase change point is another limitation for a

compositional simulation, so the gas injection form was selected for the program. Based on the

240
simulation, the injected CO2 plume reaches a diameter of 185 m for BHP=4.9 MPa after a

five-year injection period.

For a long-term simulation with a compositional simulator, a long processing time will be

needed. A simple general method was introduced for a long-term plume size estimation based on

the short-term simulation. The example of simulation and suggested equation output were

compared with the acceptable difference in the plume size estimation.

In the rock physics study, the P-wave velocity variation by CO2 injection was controlled by

the lab study test results, field measurements and calculated by fluid substitution equation. The

mixed type of the fluid has a very important role in the velocity change. Uniform mixed type

saturation shows the largest velocity drop of -16% for CO2 saturation<15% but the uniform mixed

type is not possible for CO2 in the gas phase and brine. The best match for the fluid mix in the FRS

reservoir can be explained as a semi patchy mixed saturation and that velocity can be determined

by Brie or VRH averages. All possible models were introduced in the study and the velocity

models were generated by each fluid mix type.

The uniform velocity made by the Reuss average and uniform mixed type results a clear

amplitude change in the reservoir (or velocity) boundaries, and in the migrated section, the location

of the anomaly is matched with the real location. In the diffusive velocity test, the seismic response

can not show clear amplitude changes at the boundary, and after migration, a shadow of the central

part of the anomaly is visible. This test shows that the dimension of the solid velocity shape is

measurable, but for a diffusive velocity, the seismic can not show the velocity change geometry or

reservoir size.

241
 The acquisition configuration can have a significant effect on the seismic response.

Surface acquisition has a better imaging condition, and the boundary of the reservoir can be

recognized properly. However, the acquired amplitude level of the reservoir in the surface seismic

is lower than for well based seismic methods. Thus, the surface seismic method can be a reliable

method for large sequestration fields with a large change in the saturation in the reservoir. For the

small fields and the reservoir activities with small saturation change, the well seismic methods

(VSP and cross-well surveys) are a better choice for the reservoir characterization. The low level

of the noise content in the well seismic acquisition can also help to detect a lower saturation and

the velocity change in the formation.

8.1 The research trend in the future

The main goal of the research was evaluation of the BBRS reservoir by the integration of the

different disciplines with focus on the seismic method. The main parameters that can be solved

are:

Plume size

Velocity of the reservoir cells (by FWI method)

Estimation of the fluids mixed model in the reservoir

The plume growth rate (large scale Permeability)

Saturation

Porosity

The plume size and influence of the mixed model was discussed and the base science for the

fluid mixed model and plume growth checked by synthetic models.

242
 The next step recommended for the reservoir study is the generation if an accurate

velocity model (for the reservoir) by the seismic method. The most powerful method for the

velocity model generation is Full Waveform Inversion (FWI). Displacement vectors in Eq. 7-1

show that to characterize the acoustic wavefield, multicomponent acquisition and imaging are

useful. Table 8-1 lists the specifications for a possible FWI study.

Figure 8-1. The concept of FWI (Martinez, 2016). The FWI method is a suitable way
for correcting the velocity model according to the initial model and seismic acquired data.
It can be a revolutionary approach to explaining velocity change (that can be translated to
the saturation) in a reservoir by seismic 4D data in the seismic resolution range.

243
 Figure 8-1 shows the steps for updating the velocity model with the seismic data and

Table 8-1 describes the approach for an FWI study. FWI is a robust method for constructing the

velocity model, and at this point, the advantage of it is estimating the velocity of a reservoir.

Table 8-1: The specification of FWI study.

Characteristics Full Waveform Inversion (FWI)
Pre-stack modeling, typically by two-way
• Role of wave equation: wave equation.
•Acquisition requirements & “Transmission & reflection tomography”:
data preparation •Requires a good initial model.
•Long offsets and low frequencies.
•Accurate kinematics and dynamics.
•Requires a large amount of computer
resources
•Resolution of final model: •Medium to high.

Finally, for the seismic analysis, a three-component time-lapse data from the project with the

simulation result will be a great oppurtunity for the reservoir, rock physics and seismic to put

onemore step forward in the reservoir estimation and evaluation.

244
 Appendices

Appendix A. Porosity and Permeability determined by NMR logging

Porosity: In clean, water-filled formations, NMR effective porosity (MPHI) should

approximately equal neutron-density cross plot porosity. In shaly sands, MPHI should

approximately equal density porosity, calculated with the correct grain density; however, the

MPHI may not equal effective porosity because of the effects of HI and long T1 components:

𝑇
− 𝑊
𝑀𝑃𝐻𝐼 = 𝜑𝑒 . 𝐻𝐼. [1 − 𝑒 𝑇1 ]

where

MPHI is measured by the NMR tool;

ϕe is effective porosity of the formation;

HI is related to the amount of fluid in the effective porosity system;

TW is polarization time used during logging;

T1 is longitudinal relaxation time of the fluid in the effective porosity system.

MPHI is almost always less than NMR total porosity (MSIG):

MSIG=MPHI+MCBW

where MSIG is measured by NMR total-porosity logging, and clay-bound-water porosity

(MCBW) is measured by the NMR tool with partial-polarization acquisition. In very clean

formations, however, NMR MCBW is virtually zero, and then MPHI equals MSIG

Permeability: For permeability, two approaches are using:

a- The free fluid (Timur-Coates) model

245
 b- The Schlumberger-Doll-Research (SDR) model

In the simplest form of the free-fluid model, permeability, kCoates, is expressed as follows

𝜑 𝑀𝐹𝐹𝐼 2
𝑘𝐶𝑜𝑎𝑡𝑒𝑠 = [( )2 ]
𝐶 𝑀𝐵𝑉𝐼

where ϕ is MSIG (a total porosity by using both a short TE (0.6 ms) with partial polarization

and long TE (1.2 ms)) , MBVI is obtained through the CBVI (BVI is estimated by summing the

MRIL T2 distribution up to the time T2cutoff ) or SBVI (BVI obtained by the MRIL spectral

method. This BVI estimate is determined from a

model that assigns a percent of the porosity in each spectral bin to bound water.

Various models are available for use with this method) method, MFFI (The free fluid index) is the

difference between MSIG and MBVI (assuming that there is no clay-bound water), and C is a

formation-dependent variable. The free-fluid model is very flexible and has been calibrated using

core data for successful use in different formations.

Using the SDR model, permeability is expressed as

2
𝑘𝑆𝐷𝑅 = 𝐶 𝑇2𝑔𝑚 𝜑4

Where

ϕ: NMR effective porosity (MPHI),

T2gm: the geometric mean of the T2 distribution

C: a formation-dependent variable.

The SDR model is works properly in water-saturated zones.

246
 source:

Coates, G. R., Xiao ,L., Prammer, M. G., 1999, NMR logging principles & applications,

Haliburton Energy Services, Houston

Petrowiki.org

247
Appendix B. Some relations about the reservoir simulation

Formation volume factor for gas (CO2)

It is the ratio of the gas volume in the reservoir to the standard condition that is p=1 bar and

T= 15oC. The real gas equation is the base for the formulation of the formation volume factor for

gas:

VR znRT psc zTpsc

Bg   
Vsc p zscnRTsc pTsc

Formation volume factor for brine

For the brine, the following formula is an estimation for formation volume factor with

considering brine volume and density change:

Vrc  sc
Bw 
Vsc  rc
where:

Vrc = volume occupied by a unit mass of water at reservoir conditions (weight of gas

dissolved in water at reservoir or standard conditions is negligible), ft3,

Vsc= volume occupied by a unit mass of water at standard conditions, ft3,

ρsc = density of water at standard conditions, lbm/ ft3,

ρrc= density of water at reservoir conditions, lbm/ ft3.

Another alternative for the FVF calculation for the brine was explained by McCain

(1990,1991) as:

248
 Bw  1  Vwp  1  VwT 
Where:

Vwp  1.0001102  1.33391104 T  5.50654 107 T 2

VwT  1.95301109 pT  1.72834 1013 p 2T  3.58922 107 p  2.253411010 p 2

Where p = pressure in psia, and T = temperature in °F. According to McCain, this correlation

agrees with a limited set of published experimental data to within 2%. The correlation is

considered valid for temperatures to 260°F, and pressures to 5,000 psia. An increase in dissolved

solids causes a slight increase in ΔVwT and a small decrease in ΔVwp, which offset each other to

within 1%.

Solution gas/oil ratio (GOR)

It is the amount of gas dissolved in the reservoir’s fluid in the different pressure. It shows the

amount of volatile part in liquid.

Rs=Volume of gas evolved from liquid/Volume of produced liquid following gas evolution

For the Black Oil Simulator, the GOR is relevant data that should define for the simulator,

but in compositional simulation, it is calculated by EoS in the simulator.

249
 References

Alemo, B. L., Aker, E., Soldal, M., Johnsen, O., Aagaard, P., 2011, Influence of CO2 on rock
physics properties in typical reservoir rock: A CO2 flooding experiment of brine saturated
sandstone in a CT-scanner, Energy Procedia 4 , 4379–4386
Ansari, A., Brossier, R., Garambois, S., Audebert, F., Thore, P., Virieux, J., Time-lapse imaging
using regularized FWI: a rubustness study, SEG Lass Vegas, Annual Meeting, 2012
Bachu, S., 2013, Drainage and Imbibition CO2 /Brine Relative Permeability Curves at in Situ
Conditions for Sandstone Formations in Western Canada, Energy Procedia, P. 4428-4436
Batzle, M., and Wang, Z., 1992, Seismic properties of pore fluids: Geophysics, 57, 1396–1408.
Baysal, E., D. D. Kosloff, and J. W. Sherwood, 1983, Reverse time migration: Geo-physics,48,
1514–1524.
Bennion, B., Bachu, S., 2005, Relative permeability characteristics for supercritical CO2
displacing water in a variety of potential sequestration zones in the western Canada
sedimentary basin, Paper SPE 95547 at SPE Annual Technical Conference and Exhibition,
Dallas, TX, p. 9–12
Berenger, J. E., 1994, A perfectly matched layer for the absorption of electromagnetic waves,
Comput. Phys. 114.185-200
Berryman, J. G., 1999, Origin of Gassmann’s equations: Geophysics, 64, 1627–1629.
Berryman, J. G., and Milton, G. W., 1991, Exact results for generalized Gassmann’s equation in
composite porous media with two constituents: Geophysics, 56, 1950–1960.
Brekhovskikh, L., 1960, Waves in layered media, Academic Press
Burton, M., Kumar, N., Bryant, S. L., CO2 injectivity into brine aquifers: Why relative
permeability matters as much as absolute permeability, Proceedings of the 9th International
Conference on Greenhouse Gas Control Technologies (GHGT-9), 16–20 November 2008,
Washington DC, USA, Energy Procedia, V.1,Feb.2009,P 3091-3098,ELSEVIER
Burton, M., Bryant, S. L., Surface dissolution: minimizing groundwater impact and leakage risk
simultaneously, Energy Procedia 1 (2009) 3707–3714, ELSEVIER
Carcione, J. M., Herman, G. C., and ten Kroode A. P. E., 2002, Seismic modeling, GEOPHYSICS,
VOL. 67, NO. 4 (JULY-AUGUST 2002); P. 1304–1325
Chadwick et.al. 2009

250
Collino, F., and C. Tsogka, 2001, Application of the perfectly matched absorbing layer model
to the linear elastodynamic problem in anisotropic heterogeneous media: Institut National
de Recherche en Informatique et en Automatique.
Connolly, P., 1999, Elastic impedance: The Leading Edge, 18, 438–452,
Cordsen, A., Galbraith, M., and Peirce, J., 2000, Planning land 3-D seismic surveys: Soc. Expl.
Geophys., Tulsa, Oklahoma.
Danesh, A., 1998, PVT and phase behavior of petroleum reservoir fluids: Elsevier.

Davies R.J., Cartwright J.A., Stewart S.A., Lappin M., Underhill J.R., 2004,3D seismic
technology: Application to the exploration of sedimentary basins, The Geological
Society.

Eberhart-Phillips, D., Han, D-H., and Zoback, M. D., 1989, Empirical relationships among
seismic velocity, effective pressure, porosity, and clay content in sandstone: Geophysics,
54, 82–89.

Eisinger, C.L., Jensen, J.L., 2009, Geology and Geomodelling, Wabamun area CO2
sequestration project (WASP), University of Calgary

Elenius, M.L., Tchelepi, H.A. , Johannsen, K., 2010, CO2 trapping in sloping aquifers: high
resolution numerical simulations, XVIII International Conference on Water Resources,
Barcelona

Endres, A. L., and Knight, R., 1989, The effect of microscopic fluid distribution on elastic wave
velocities: The Log Analyst, Nov-Dec, 437–444.

Evans, B. J., 1997, A handbook for seismic data acquisition: Soc. Expl. Geophys., Tulsa,
Oklahoma.

Fanchi, J.R., 2006, Principle of applied reservoir simulation, Elsevier

Garcia, J.E., 2001. Density of aqueous solutions of CO2. LBNL Report 49023, Lawrence
Berkeley National Laboratory, Berkeley, CA

Gassmann, F., 1951, Über die Elastizität poröser Medien: Vierteljahress-chrift der
Naturforschenden Gesellschaft in Zürich,96, 1–23.

251
Gerritsma P., 2005, Advance seismic data processing, Course note, Tehran.
Gregory, A. R., 1976, Fluid saturation effects on dynamic elastic properties of sedimentary rocks:
Geophysics, 41, 895–921.
Han, D.H., Batzle, M. L., 2004, Gassmann’s equation and fluid-saturation effects on seismic
velocities, Geophysics, March-April
Han, D.H., Nur, A., and Morgan, D., 1986, Effects of porosity and clay content on wave velocities
in sandstones: Geophysics, 51, 2093– 2107.
Hassanzadeh, H., Pooladi-Darvish, M., Elsharkawy A. M., Keith, W. K.,2008, Predicting PVT
data for CO2-brine mixtures for black-oil simulation of CO2 geological storage,
International Journal of Greenhouse Gas Control 2 , P65-77
Hashin, Z., and Shtrikman, S., 1962, A variational approach to the elastic behavior of multiphase
materials: J. Mech. Phys. Solids, 11,127–140.
Hays, J.D, Imbrie, J., and Shackleton, N.J., Variations in the Earth's Orbit: Pacemaker of the Ice
Ages, Science, 194, no. 4270 (1976), 1121-1132.
Hill, R., 1963, Elastic properties of reinforced solids: Some theoretical principles: J. Mech. Phys.
Solids, 11, 357–372.
Hill R., 1952, The elastic behaviour of a crystalline aggregate: Proceedings of the Physical Society,
Section A, 65, 349–354
Journel, A. G., Huijbregts, C. J., Mining Geostatistics, Academic Press, 1991
Katahara, K.W. 1996. Clay Mineral Elastic Properties. Presented at the 1996 SEG Annual
Meeting, Denver, 10-15 November. Paper No. 1996-1691.
Kearey P., Brooks M., Hill I., 2002, An Introduction to Geophysical Exploration, Blackwell
Science
Landrø M., Geophysics, (May-June 2001), Discrimination between pressure and fluid saturation
changes from time-lapse seismic data, Vol. 66, NO. 3, P. 836–844,
Li et al., (2006)
Liner, C. L., and Gobeli, R., 1997, 3-D seismic survey design and linear V(z): 67th Ann. Internat.
Mtg., Soc. Expl. Geophys, Expanded Abstracts, 43-46.
Liner, C. L., Underwood, W. D., 1999, 3-D seismic survey design for linear V(z) media:
Geophysics, 64, 486-493.
Liner C., 2004, Elements of 3-D Seismology, PennWell Books

252
Lumley, D., 4D seismic monitoring of CO2 sequestration, ASEG Extended Abstracts 2010:
21st Geophysical Conference
Margrave, G. F., 1997, Seismic acquisition parameter considerations for a linear velocity medium:
67th Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 47-50.
Martinez, 2016
Mavko, G., Mukerji, T., and Dvorkin, J., 1998, The Rock Physics Handbook: Tools for Seismic
Analysis in porous media: Cambridge Univ.Press.
Mavko, G., Rock Physics, Course was held by CSEG, 2014, Calgary
McCain, W.D. Jr.: McCain, W.D. Jr. 1990. The Properties of Petroleum Fluids, second edition.
Tulsa, Oklahoma: PennWell Books.
McCain Jr., W.D. 1991. Reservoir-Fluid Property Correlations-State of the Art (includes
associated papers 23583 and 23594). SPE Res Eng 6 (2): 266-272. SPE-18571-PA.
Murphy, W., Reischer A., and Hsu, K., 1993, Modulus decomposition of compressional and shear
velocities in sand bodies: Geophysics, 58, 227–239.
Nowroozi D., Lawton D.C., and Khaniani H., Seismic modeling and imaging for CO2 injection
at the FRS project, SEG 2016, Dallas
Nowroozi D., Lawton D.C., and Khaniani H., FWI framework for Seismic modeling and imaging
for CO2 injection at the FRS project, Geoconvention 2016, CREWES annual report, 2015
Nowroozi, D., Lawton, D.C., Rock physics study of the Nisku aquifer from results of reservoir
simulation, Geoconvention 2015, CREWES annual report, 2014
Nowroozi, D., Lawton, D.C., Seismic parameter design for reservoir monitoring, Brooks, Alberta,
Geoconvention 2015 and CREWES annual report, 2014
Nowroozi, D., Lawton, D.C., Reservoir simulation for a CO2 sequestration project, CREWES
annual report, 2013
Nur, A., Mavko, G., Dvorkin, J., and Gal, D., 1995, Critical porosity: The key to relating physical
properties to porosity in rocks: 65th Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded
Abstracts, 878.
Nygaard, R.,2010, Geomechanical analysis, Wabamun area CO2 sequestration project (WASP),
University of Calgary
O’Connell, R. J., 1984, A viscoelastic model of anelasticity of fluid saturated porous rocks, in
Johnson, D. L., and Sen, P.N., Eds., Physics and chemistry of porous media: Amer. Inst.
Physics, 166–175.

253
Packwood, J. L., and Mavko, G., 1995, Seismic signatures of multiphase reservoir fluid
distribution: Application to reservoir monitoring: 65th Ann. Internat. Mtg., Soc. Expl.
Geophys., Expanded Abstracts, 910–913.
Pagani et al. 2005
Pegah, E., Feiz Aghaei, B., Javaherian, A. R., and Nowroozi, D., Determination of Bin size and
Migration aperture in 3-D seismic survey design for AHWAZ oil field with using linear
velocity model (LVZ): EAGE, Expanded Abstract, First International Petroleum
Conference & Exhibition, Shiraz, Iran, 4-6 May 2009.
Perkins, E., Fundamental geochemical processes between CO2, water and minerals, Alberta
Innovates – Technology Futures
Petit et al., 1999
Pooladi-Darvish, M., 2009, Geological storage of CO2, Part 2: Reservoir engineering aspects,
Technical video series, Fekete
Prasad, M., Kopycinska, M., Rabe, U. et al. 2002. Measurement of Young's modulus of clay
minerals using atomic force acoustic microscopy. Geophys. Res. Lett. 29 (8): 1172.
Ramamoorthy, R., and Murphy, W. F., 1998, Fluid identification through dynamic modulus
decomposition in carbonate reservoirs: Trans. Soc. Prof. Well Log Analysts 39th Ann.
Logging Symp., paper Q.
Raymer, L. L, Hunt, E. R., and Gardner, J. S., 1980, An improved sonic transit time-to-porosity
transform: Trans. Soc. Prof. Well Log Analysts 21st Ann. Logging Symp.
Reuss, 1929
Sbar, M. L., 2000, Exploration risk reduction: An AVO analysis in the offshore middle Miocene,
central Gulf of Mexico: The Leading Edge, 19, 21–27.
Schlumberger, 2009, ECLIPSE Black-Oil reservoir simulation-Training and exercise guide V2
Scotese et al. 2002
Smith T.M., Sondergeldz C.H., Rai C.S., (March-April 2003), Gassmann fluid substitutions: A
tutorial, 2003, Geophysics, Vol. 68, NO. 2 P. 430–440,
Sondergeld, C. H., and Rai, C. S., 1993, A new exploration tool: Quantitative core characterization:
Pageoph, 141, 249–268.
Span, R., and Wagner, W., 1996, A New Equation of State for Carbon Dioxide Covering the Fluid
Region from the Triple-Point Temperature to 1100 K at Pressure up to 800 Mpa, J. Phys.
Chem. Ref. Data, Vol.25, No. 6

254
Spencer, J.W., Cates M. E., and Thompson, D. D., 1994, Frame moduli of unconsolidated
sands and sandstones: Geophysics, 59, 1352–1361.
Stone, D.G., Designing Seismic Surveys in Two and Three Dimensions, 1994, SEG, Tulsa,
Oklahoma.
Sun, S., Han, D., Batzle, M.,2009, CO2 Velocity Measurement and Models for Temperatures down
to -10 and up to 200 oC and Pressure up to 100 MPa, SEG 2009 International Exposition
and Annual Meeting, Houston
Telford, W. M., Geldart, L. P., and Sheriff, R. E., 1990, Applied geophysics, 2nd ed: Cambridge
Univ.Press.
TOF, 2009, TABLE of formation, Alberta
Veeken P.C.H., 2007, Seismic stratigraphy, basin analysis and reservoir characterization, Elsevier
Vernik, L., 1998, Acoustic velocity and porosity systematics in siliciclastics: The Log Analyst,
July-Aug., 27–35.
Vermeer, G. J. O., 2002, 3-D seismic survey design, Soc. Expl. Geophys., Tulsa, Oklahoma.
Wang, Z.Z., Wang, H., and Cates, M.E. 1998. Elastic Properties of Solid Clays. Presented at the
1998 SEG Annual Meeting, New Orleans, 13-18 September. Paper No. 1998-1045.
Wang, A., 2000, Velocity-density relationships in sedimentary rocks, in Wang, Z., and Nur, A.,
Eds., Seismic and acoustic velocities in reservoir rocks, 3: Recent developments: Soc. Expl.
Geophys, 8–23.
Wang, Z., 2001, Fundamentals of seismic rock physics: Geophysics, 66, 398–412.
Wang, Z., Wang, H., and Cates, M. E., 2001, Effective elastic properties of clays: Geophysics, 66,
428–440.
Whitmore, N.D. and Crawley, S., 2012, Applications of RTM inverse scattering imaging
conditions: SEG Expanded Abstracts 82nd Annual Meeting, Las Vegas, Nevada
Wyllie, M.R. J., Gregory, A. R., and Gardner, L.W., 1956, Elastic wave velocities in
heterogeneous and porous media: Geophysics, 23, 41– 70.
Yam, Helen, CO2 rock physics: a laboratory study, 2011, MSc thesis, Department of Physics,
University of Alberta
Zakaria, A., J. Penrose, F. Thomas, and X. Wang, 2000, The two-dimensional numerical modeling
of acoustic wave propagation in shallow water: Acoustics.
Zhou, M., 2003, A well-posed PML absorbing boundary condition for 2D acoustic wave equation:
UTAM Annual Report.
255