NUMERICAL MODELING OF FRACTURE PERMEABILITY CHANGE IN

NATURALLY FRACTURED RESERVOIRS USING A FULLY COUPLED

DISPLACEMENT DISCONTINUITY METHOD

A Dissertation

by

QINGFENG TAO

Submitted to the Office of Graduate Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

May 2010

Major Subject: Petroleum Engineering

 NUMERICAL MODELING OF FRACTURE PERMEABILITY CHANGE IN

NATURALLY FRACTURED RESERVOIRS USING A FULLY COUPLED

DISPLACEMENT DISCONTINUITY METHOD

A Dissertation

by

QINGFENG TAO

Submitted to the Office of Graduate Studies of

Texas A&M University
in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Approved by:

Co-Chairs of Committee, Christine A. Ehlig-Economides

Ahmad Ghassemi
Committee Members, A. Daniel Hill
Ding Zhu
Yalchin Efendiev
Head of Department, Stephen A. Holditch

May 2010

Major Subject: Petroleum Engineering

 iii

ABSTRACT

Numerical Modeling of Fracture Permeability Change in Naturally Fractured Reservoirs

Using a Fully Coupled Displacement Discontinuity Method. (May 2010)

Qingfeng Tao, B.S., China University of Geosciences;

M.S., University of North Dakota

Co-Chairs of Advisory Committee: Dr. Christine A. Ehlig-Economides

Dr. Ahmad Ghassemi

Fractures are the main flow channels in naturally fractured reservoirs. Therefore

the fracture permeability is a critical parameter to production optimization and reservoir

management. Fluid pressure reduction caused by production induces an increase in

effective stress in naturally fractured reservoirs. The change of effective stress induces

fracture deformation and changes fracture aperture and permeability, which in turn

influences the production. Coupled interactions exist in the fractured reservoir: (i) fluid

pressure change induces matrix deformation and stress change; (ii) matrix deformation

induces fluid volume change and fluid pressure change; (iii) fracture deformation

induces the change of pore pressure and stress in the whole field (the influence

disappears at infinity); (iv) the change of pore pressure and stress at any point has an

influence on the fracture and induces fracture deformation. To model accurately the

influence of pressure reduction on the fracture permeability change in naturally fractured

reservoirs, all of these coupled processes need to be considered. Therefore, in this

dissertation a fully coupled approach is developed to model the influence of production

 iv

on fracture aperture and permeability by combining a finite difference method to solve

the fluid flow in fractures, a fully coupled displacement discontinuity method to build

the global relation of fracture deformation, and the Barton-Bandis model of fracture

deformation to build the local relation of fracture deformation.

The fully coupled approach is applied to simulate the fracture permeability

change in naturally fracture reservoir under isotropic in situ stress conditions and high

anisotropic in situ stress conditions, respectively. Under isotropic stress conditions, the

fracture aperture and permeability decrease with pressure reduction caused by

production, and the magnitude of the decrease is dependent on the initial effective in situ

stress. Under highly anisotropic stress, the fracture permeability can be enhanced by

production because of shear dilation. The enhancement of fracture permeability will

benefit to the production of oil and gas.

 v

ACKNOWLEDGEMENTS

I would like to thank one of my co-advisors, Dr. Ehlig-Economides, for her

support and guidance. She provided me the opportunity to study the naturally fractured

reservoirs and supported me to choose the topic of the dissertation which I am very

interested in. I will forever be indebted to my other co-advisor, Dr. Ghassemi, who led

me into the field of geomechanics. The idea of using the displacement discontinuity

method in this study is from the discussions with Dr. Ghassemi two years ago. He

provided me lots of help during the past six years and is always available whenever I

need help in my study and work. I would like also to thank the other committee

members, Dr. Hill, Dr. Zhu and Dr. Efendiev, for their help. The discussions on this

topic with Dr. Zeng of University of North Dakota and Dr. Wu of Colorado School of

Mines are greatly appreciated. I am grateful to my parents for their continuous support

and to my wife for her understanding and love. Finally I would like to thank all

professors I took classes from and all my friends.

 vi

TABLE OF CONTENTS

Page

ABSTRACT…… ............................................................................................................. iii

ACKNOWLEDGEMENTS ............................................................................................... v

TABLE OF CONTENTS .................................................................................................. vi

LIST OF FIGURES ........................................................................................................... ix

LIST OF TABLES .......................................................................................................... xiv

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

1.1 Problem and objective ...................................................................................... 2

1.2 Numerical methods for deformable fractured reservoirs ................................. 7
1.2.1 Continuum methods .......................................................................... 8
1.2.2 Discrete element method (DEM) .................................................... 11
1.2.3 Displacement discontinuity method (DDM) ................................... 13
1.3 Summary of the dissertation ........................................................................... 14

CHAPTER II DISPLACEMENT DISCONTINUITY METHOD .............................. 17

2.1 Elastic DDM ................................................................................................... 18

2.2 Fully coupled DDM for porous media saturated with a compressible
single-phase fluid ........................................................................................... 26
2.2.1 Constitutive equations of a porous medium saturated with a
compressible single-phase fluid ...................................................... 28
2.2.2 Pressure diffusion in a porous medium ........................................... 31
2.2.3 Fundamental solutions for a single fracture segment in an
infinite two-dimensional porous medium ....................................... 33
2.2.4 Solutions for multiple fracture segments in an infinite two-
dimensional porous medium ........................................................... 36
2.2.5 Determination of the fracture discontinuous displacement ............. 42
2.3 Model verification .......................................................................................... 46
2.4 Chapter conclusions ....................................................................................... 54

CHAPTER III NONLINEAR DEFORMATION OF A SINGLE ROUGH

FRACTURE UNDER STRESS ............................................................ 55
 vii

Page

3.1 Surface characteristics of a fracture ............................................................... 56

3.1.1 Joint compressive strength (JCS) .................................................... 56
3.1.2 Basic friction angle (φb) and residual friction angle (φr) ................. 58
3.1.3 Joint roughness coefficient (JRC) ................................................... 60
3.2 Normal deformation ....................................................................................... 62
3.3 Shear deformation and dilation ...................................................................... 69
3.4 Fracture aperture and permeability ................................................................ 73
3.5 Chapter conclusions ....................................................................................... 75

CHAPTER IV MODELING OF THE FRACTURE APERTURE AND

PERMEABILITY CHANGE IN FRACTURED RESERVOIRS ........ 77

4.1 Fluid flow in the fracture network.................................................................. 78

4.2. Mechanical coupling of fracture deformation ............................................... 81
4.2.1 Local relation between stress and displacement to fracture
deformation ..................................................................................... 82
4.2.2 Global relation between stress and displacement to fracture
deformation ..................................................................................... 83
4.3 Uncoupled and coupled solution methods ..................................................... 84
4.3.1 Uncoupled method .......................................................................... 85
4.3.2 Coupled method .............................................................................. 89
4.4 Chapter conclusions ....................................................................................... 93

CHAPTER V MODEL APPLICATIONS ................................................................... 94

5.1 Fracture aperture and permeability change under isotropic conditions ......... 94
5.1.1 Parameters and assumptions............................................................ 95
5.1.2 Results for isotropic stress conditions ............................................. 98
5.1.3 Effect of initial effective normal stress ......................................... 108
5.1.4 Effect of ratio of hydraulic fracture aperture to mechanical
fracture aperture (wef /wf) .............................................................. 110
5.2 Fracture aperture and permeability change under high anisotropic in situ
stress conditions ........................................................................................... 114
5.3 Chapter conclusions ..................................................................................... 121

CHAPTER VI CONCLUSIONS AND RECOMMENDATIONS ............................. 123

6.1 Conclusions .................................................................................................. 123

6.2 Recommendations ........................................................................................ 125

NOMENCLATURE ....................................................................................................... 126

REFERENCES… ........................................................................................................... 132

 viii

Page

APPENDIX A FUNDAMENTAL SOLUTIONS FOR FLUID SOURCE ................. 137

APPENDIX B FUNDAMENTAL SOLUTION FOR DISPLACEMENT

DISCONTINUITIES SOURCE .......................................................... 139
ij
APPENDIX C COEFFICIENT MATRIX, C p , FOR FLUID DIFFUSION IN A
REGULAR FRACTURE NETWORK ............................................... 143

APPENDIX D COMBINED SET OF LINEAR EQUATIONS IN MATRIX

FORM FOR A REGULAR FRACTURE NETWORK ...................... 151

VITA………….. ............................................................................................................ 154

 ix

LIST OF FIGURES

Page

Figure 1-1. Dual-porosity model (Warren and Root, 1963). ............................................. 2

Figure 1-2. Illustration of the fluid flow in the interconnected pores in a porous
matrix and the induced deformation of the porous matrix (influence of
fluid flow on the matrix deformation). The dashed red line represents
the boundary of the porous matrix before fluid injection/production. ........... 5

Figure 1-3. Illustration of the compression of a continuum porous matrix and the
induced pore pressure change and fluid flow in the interconnected pores
(influence of matrix deformation on the fluid flow and pore pressure
change). The dashed red line represents the boundary of the porous
matrix before deformation. ............................................................................. 5

Figure 1-4. Illustration of a fracture in a fluid-saturated porous media. ........................... 6

Figure 1-5. Schematic representation of double porosity model (Lewis and

Ghafouri, 1997). ............................................................................................. 9

Figure 1-6. A real rough fracture in porous media and its idealized smooth fracture
in two dimensions (Sun, 1994). .................................................................... 11

Figure 1-7. Geometry of fracture system in the DFN model (Min et al., 2004). ............ 13

Figure 2-1. A small discontinuous segment in an infinite two-dimensional

nonporous medium (after Crouch and Starfield, 1983). ............................... 20

Figure 2-2. A curvy fracture discretized into 5 segments in an infinite two-

dimensional nonporous medium. ................................................................. 21

Figure 2-3. Local co-ordinate for the jth fracture segment in an elastic nonporous
medium. ........................................................................................................ 23

Figure 2-4. Influence of jth fracture segment on the ith fracture segment in an
elastic nonporous medium. ........................................................................... 25

Figure 2-5. A thin line fracture in an infinite two-dimensional elastic porous

medium, and the line fracture starts from (-a,0) and ends at (a,0). .............. 34
 x

Page

Figure 2-6. A curvy fracture discretized into 5 segments in an infinite two-

dimensional porous medium saturated with a single-phase fluid................. 38

Figure 2-7. Local co-ordinate for the jth fracture segment in a porous medium. ........... 39

Figure 2-8. Influence of the jth fracture segment on the ith fracture segment in an
elastic porous medium. ................................................................................. 41

Figure 2-9. Time marching scheme, χ represents Dn, Ds or qint. ..................................... 44

Figure 2-10. A line crack with constant pressure loading. ............................................... 47

Figure 2-11. Comparison of the modeled crack width using elastic DD with the
analytical solution......................................................................................... 48

Figure 2-12. Mode I loading: the crack opens as a function of time at x=0.2 in. ............ 50

Figure 2-13. Comparison of the modeled crack openings at short time and long time
with the analytical solutions for Mode I loading. ......................................... 51

Figure 2-14. Mode II loading: the crack closes as a function of time at x=0.2 in............ 51

Figure 2-15. The crack closing at t=1.91×105 hrs for Mode II loading. .......................... 52

Figure 2-16. The crack width for Mode I +II loading at x=0.2 in.................................... 52

Figure 2-17. Comparison of the modeled crack openings at short time and long time
with the analytical solutions for Mode I+II loading. .................................... 53

Figure 3-1. Tilt test on fractured sample. ........................................................................ 58

Figure 3-2. Typical JRC values for joint samples of different roughness (Barton and
Choubey, 1977). ........................................................................................... 61

Figure 3-3. Measurements of the closure under normal stress of an artificially-

induced tensile fracture in a rock core (Goodman, 1976). ........................... 63

Figure 3-4. Normal stress (σn) vs closure curves for a range of fresh fractures in
different rock types, under repeated loading cycles (Bandis et al., 1983). .. 64

Figure 3-5. Normal stress (σn) vs closure curves for a range of weathered fractures
in different rock types under repeated loading cycles (Bandis et al.,
1983). ............................................................................................................ 65
 xi

Page

Figure 3-6. Linear plots of Dn/σn vs Dn for different fracture types, indicating good
hyperbolic fit irrespective of the stress history and the loading mode
(Bandis et al. 1983). ..................................................................................... 68

Figure 3-7. Shear stress – shear displacement for joints with different normal stress
and JRC (Barton et al., 1985). ...................................................................... 71

Figure 3-8. Cumulative mean shear stress---shear displacement (a) and dilation (b)
curves (Bandis et al., 1981). ......................................................................... 72

Figure 3-9. Laminar flow through a pair of smooth parallel plates. ............................... 73

Figure 3-10. Comparison of mechanical aperture and hydraulic aperture (Barton et

al., 1985; Olson and Barton, 2001). ............................................................. 75

Figure 4-1. Fluid flow through a rough fracture. ............................................................ 79

Figure 4-2. Fluid flow through an artificial fracture represented using average
fracture aperture. .......................................................................................... 80

Figure 4-3. Local relation of fracture deformation.......................................................... 83

Figure 4-4. Interface flow rate between fracture and matrix........................................... 87

Figure 4-5. Illustration of effective stress change on fracture......................................... 88

Figure 5-1. Well located at the center of a fractured field, which is surrounded by
matrix rock of effectively infinite extent. ..................................................... 96

Figure 5-2. Nonlinear fracture normal deformation. ....................................................... 97

Figure 5-3. Pore pressure distribution after 360 days production. .................................. 99

Figure 5-4. Fracture aperture declines with time. ......................................................... 100

Figure 5-5. Effective normal stress and fracture aperture change with time for the
fracture intersected with well. .................................................................... 100

Figure 5-6. Fracture permeability declines with time. .................................................. 101

Figure 5-7. Comparison of transient pressure behavior at bottom hole with constant
production rate between fixed fracture permeability and stress-
dependent fracture permeability case. ........................................................ 101
 xii

Page

Figure 5-8. Comparison of transient pressure behavior at bottom hole with constant
production rate between fixed fracture permeability and stress-
dependent fracture permeability case for a long time production to show
the flow behavior in the surrounded matrix rock. ...................................... 102

Figure 5-9. Comparison of transient pressure behavior at bottom hole with constant
production rate between the case with a well connected with a fracture
network and the case without any fracture. ................................................ 103

Figure 5-10. A well is intersected with a fracture in a non-fractured reservoir. ............ 104

Figure 5-11. Comparison of transient pressure behavior at bottom hole with constant
production rate between the case with a well connected with a fracture
network and the case with only one fracture in the reservoir. .................... 104

Figure 5-12. A well is located at the center of a matrix in a fractured network

surrounded by matrix rock. ........................................................................ 106

Figure 5-13. Comparison of transient pressure behavior at bottom hole with constant
production rate between the case with a well connected with a fracture
network and the case with a well at the center of a matrix in a fracture
network. ...................................................................................................... 107

Figure 5-14. Pressure derivative curves for successive build ups. ................................. 108

Figure 5-15. Effect of initial effective in situ stress on the fracture permeability
change. ........................................................................................................ 109

Figure 5-16. The ratio wef/wf as a function of wf. .......................................................... 111

Figure 5-17. Bottom hole pressure declines with time for three cases: wef=wf,
JRC=10.2 and JRC=12. .............................................................................. 112

Figure 5-18. Log-log plot of the pressure derivatives for three cases: wef=wf,
JRC=10.2 and JRC=12. .............................................................................. 112

Figure 5-19. The mechanical aperture of fracture intersected with well changes with
time for three cases: wef=wf, JRC=10.2 and JRC=12................................. 113

Figure 5-20. The effective hydraulic aperture of fracture intersected with well
changes with time for three cases: wef=wf, JRC=10.2 and JRC=12........... 113

Figure 5-21. The permeability of fracture intersected with well changes with time
for three cases: wef=wf, JRC=10.2 and JRC=12. ........................................ 114
 xiii

Page

Figure 5-22. The relation of shear stress and shear displacement used in the
modeling. .................................................................................................... 115

Figure 5-23. Well located at the center of a fractured field under anisotropic stress
field and the fractured network is surrounded by matrix rock. .................. 117

Figure 5-24. Shear displacement distribution after 360 days production for the case
fractures are already yielded before production. The arrow represents
the shear direction. ..................................................................................... 118

Figure 5-25. Fracture permeability distribution after 360 days production for the
case fractures are already yielded before production. ................................ 118

Figure 5-26. Distribution of fracture permeability and shear displacement (shown

with arrows) after 360 days production for the case fractures are already
yielded before production. .......................................................................... 119

Figure 5-27. Fracture permeability increases with production for the case the
fracture are already yielded before production........................................... 119

Figure 5-28. Log-log plot of pressure drop and pressure derivative for the case in
which the fracture permeability of most fractures are enhanced by
production. .................................................................................................. 120

Figure 5-29. Distribution of fracture permeability and shear displacement (shown

with arrows) after 360 days production for the case with symmetric
fracture network and production wells. ...................................................... 121

Figure C-1. A regular fracture network. ........................................................................ 148

Figure C-2. Discretization of type (a) fracture segment. ............................................... 149

Figure C-3. Discretization of type (b) fracture segment. .............................................. 150

Figure D-1. A regular fracture network with indices of fracture segments. ................. 151
 xiv

LIST OF TABLES

Page

Table 2-1. Parameters in the modeling of pressurized crack. ......................................... 50

Table 3-1. Typical JCS values (ISRM, 1978). ................................................................ 57

Table 3-2. Basic friction angles of various unweathered rocks obtained from flat
and residual surfaces (Barton and Choubey, 1977)........................................ 59

Table 5-1. Rock and fracture parameters in the modeling. ............................................. 97

Table 5-2. Production rate history. ................................................................................ 108

 1

CHAPTER I

INTRODUCTION

A large percent of oil and gas around the world are produced from naturally

fractured reservoirs. Naturally fractured reservoirs are found in essentially all types of

lithologies including sand stones, carbonates, shales, cherts, siltstones, etc. (Aguilera,

1995). A natural fracture is “a naturally occurring macroscopic planar discontinuity in

rock due to deformation or physical diagenesis” (Nelson, 1985). Generally fractures are

the main flow channels, and the matrix provides the main storage capacity. Some

reservoirs, e.g. tight gas reservoirs, are not possible to produce without the existence of

natural fractures (microfractures). Therefore the fracture permeability is critical to the

hydrocarbon production. This chapter will start with an explanation of the problem and

the objective of this research. Next will be a review of the previous numerical methods

on the modeling of deformable fractured reservoirs. Finally there will appear a summary

of the dissertation.

__________________
This dissertation follows the style of SPE Reservoir Evaluation and Engineering.
 2

1.1 Problem and objective

Warren and Root (1963) presented a dual-porosity model to represent naturally

fractured reservoirs (Figure I-1). The highly heterogeneous system was treated as a

homogeneous system with two media – fractures and matrix. Both the matrix and the

fractures were characterized by two parameters – porosity and permeability.

Pseudosteady state flow was assumed in the matrix, as well as an interporosity flow

parameter for flow between matrix and fractures. Later a dual-porosity model with

transient flow in matrix for low permeability reservoirs was presented by De Swaan

(1976), Najurieta(1980), Cinco and Samaniego(1982). Both fracture permeability and

matrix permeability were treated as constant during production and independent of stress

and pressure. In all of these cases, flow to the well was only via fractures.

Figure I-1. Dual-porosity model (Warren and Root, 1963).

 3

In reality, reservoir pressure decreases with production for most cases, and the

effective stress in the reservoir increases, and both fractures and matrix can deform with

the increase of effective stress. For hard rocks, the deformation due to normal stress

change is small and can be neglected. However, the deformation for weak rocks or

fractured rocks can be large enough to change the reservoir properties and influence the

production. The dependence of formation permeability on pressure for a single porosity

system has been investigated by Gray et al. (1963), Vairogs et al. (1971), Thomas and

Ward (1972), Raghavan et al. (1972), Vairogs and Rhoades (1973), Samaniego et al.

(1976, 1977), Jones and Owens (1980), Samaniego and Cinco-Ley (1989), Buchsteiner

et al. (1993), Chin et al. (1998), and Davies and Davies (1999). The pressure

dependence of matrix permeability occurs as the porosity and connectivity of pores

decrease with increase in effective stress. But the permeability change in tight gas

reservoirs mainly results from the closure of microcracks with the increase of effective

stress (Ostensen, 1986).

Generally fractures are more deformable than the matrix in a naturally fractured

reservoir, and the permeability of fractures, not the matrix, dominates the flow behavior.

Furthermore, fractures are more sensitive to pressure and stress change than the matrix,

and the fracture deformation mechanism is much more complicated than matrix

deformation. The effect of stress on the aperture and permeability of a single fracture

has been well investigated in laboratory by Iwai (1976), Goodman (1976), Bandis et al.

(1983), and Barton et al. (1985). Experimental data show a nonlinear relation between
 4

normal stress and fracture closure. Bandis et al. (1983) presented a hyperbolic formula

to represent the normal stress–fracture closure relation. For shear deformation

experimental data show an approximately linear relation between shear stress and shear

displacement before yielding, and then shows a complicated relation after yielding.

Shear deformation can also induce fracture opening as the opposed asperities of a

fracture slide over each other and cause an increase in aperture. Chapter III will

elaborate on these mechanisms.

In naturally fractured reservoirs, there are coupled interactions between porous

matrix and fluid, as well as between fractures. Biot (1941, 1956) developed a theory of

poroelasticity for porous media saturated with incompressible fluid to account for the

coupled diffusion–deformation mechanism. Rice and Cleary (1976) extended the theory

for porous media saturated with compressible fluid. Biot’s theory of poroelasticity is a

continuum theory for a porous medium consisting of an elastic matrix containing

interconnected fluid-saturated pores. The fluid diffusion in porous media induces porous

matrix deformation (Figure I-2) and stress redistribution, and porous matrix deformation

also induces fluid flow (Figure I-3) and fluid pressure redistribution. If there is a

discontinuous surface (fracture) in the continuum porous media shown in Figure I-4, the

deformation of the fracture (opening or closing) will induce the deformation of the

porous matrix and also pore pressure change and fluid flow, which will be elaborated in

Chapter II.
 5

inflow
expansion

Matrix
fluid

Figure I-2. Illustration of the fluid flow in the interconnected pores in a porous matrix and the
induced deformation of the porous matrix (influence of fluid flow on the matrix deformation). The
dashed red line represents the boundary of the porous matrix before fluid injection/production.

outflow
compression

Matrix
fluid

Figure I-3. Illustration of the compression of a continuum porous matrix and the induced pore
pressure change and fluid flow in the interconnected pores (influence of matrix deformation on the
fluid flow and pore pressure change). The dashed red line represents the boundary of the porous
matrix before deformation.
 6

matrix fluid

fracture

Figure I-4. Illustration of a fracture in a fluid-saturated porous media.

In a naturally fractured reservoir, there are many fractures in the porous medium.

In addition to the interactions of fluid, porous matrix and fracture, there are interactions

between fractures including mechanical deformation and fluid flow, which will be

elaborated in Chapter II. One fracture deformation will cause stress change in the field

and induce deformation of other fractures (Crouch and Starfield, 1983, Curren and

Carvalho, 1987, Cheng and Predeleanu, 1987, Carvalho, 1990). The fluid injection or

production from one fracture can also induce fluid pressure change in other fractures, as

well as mechanical deformation. Crouch and Starfield (1983) developed a displacement

discontinuity method (DDM) to model the interactions between fractures and also the
 7

influence of fracture deformation on the stress redistribution in elastic nonporous media.

Curren and Carvalho (1987), Cheng and Predeleanu (1987) and Carvalho (1990)

developed a poroelastic DDM for fluid-saturated porous media with many discontinuous

surfaces (fractures) in it. The poroelastic DDM can be applied to model the coupled

interactions of fractures, porous matrix and fluid in porous media with fractures. This

method has been applied to simulate the hydraulic fracturing in continuum porous media

(Vandamme and Roegiers, 1990). But the poroelastic DDM has not been applied to

model the interactions of fracture, porous matrix and fluid in fractured porous media.

The oil and gas production from naturally fractured reservoirs will induce the

change in fracture aperture and permeability, thereby changing reservoir properties and

influencing production. The objective of this study is to develop an approach to

investigate the change in fracture aperture and permeability in naturally fractured

reservoirs. This approach will consider the coupled interactions of porous matrix, fluid

and fractures and the real mechanism of fracture deformation.

1.2 Numerical methods for deformable fractured reservoirs

Many researchers have investigated the deformation of fractures in fluid-

saturated fractured porous media using numerical methods (Asgian, 1988, 1989; Sun,

1994; Chen and Teufel, 1997; Gutierrez and Makurat, 1997; Lewis and Ghasouri, 1997;

Meng, 1998; Shu, 1999; Min et al., 2004; and Bagheri, 2006). The numerical methods

can be classified as continuum methods, the discrete element method, and the

displacement discontinuity method. The continuum methods treat the fractured media as
 8

an equivalent continuum media for fluid flow model, or mechanical deformation, or both.

The stress and pore pressure in the equivalent continuum media are solved by using a

finite difference method (FDM) or a finite element method (FEM). The discrete

element method (DEM) treats matrix elements divided by fractures as discrete, and

calculates the contact and deformation of the matrix element boundaries. The

displacement discontinuity method (DDM) is an indirect boundary element method. The

DDM gives an analytical solution for the influence of a fracture in a continuum media

and then sums the influences of all fractures for a fractured media by the superposition

method (refer to Chapter II for details).

1.2.1 Continuum methods

Lewis and Ghafouri (1997) developed a finite element dual porosity model.

They modeled fluid flow using a dual porosity model – the fracture and matrix were

treated as overlapping continuum media (Figure I-5). Flow properties (fracture

permeability, matrix permeability, etc.) were assumed to be independent of pore pressure

and stress. The fluid pressure change was uncoupled with the mechanical deformation

of the fractured media. The fractured media were treated as continuum elastic media.

Fluid pressure change caused by production was solved separately from porous matrix

deformation. The effective stress change resulting from pore pressure change was

derived using Terzaghi’s effective stress law (effective stress = total stress – fluid

pressure). Finally the deformation of fractured media was modeled according to the

effective stress change.

 9

Figure I-5. Schematic representation of double porosity model (Lewis and Ghafouri,
1997).

Chen and Teufel (1997) presented a partially coupled method for deformable

fractured media. For fluid flow, the fractured media was assumed as a dual porosity

model – fracture and matrix are two overlapping continuum media. For geomechanics,

the fractured media was assumed as continuum poroelastic media and the coupling

between porous matrix and fluid was based on Biot’s theory of poroelasticity. The

fracture and matrix were virtually combined into one media with one combined porosity

and compressibility in the mechanical model. Therefore the fracture deformation was

oversimplified as matrix deformation. The fracture aperture and permeability was

independent of pressure and stress. Meng (1998) and Shu (1999) used similar models

to model the coupled processes considering two fluid phases.

Sun (1994) used a discrete fracture element approach to model the deformable

fractured porous media. Fluid flow was modeled in both the porous medium and a

discrete fracture network. The transient flow rate between fracture network and porous

media was determined by the pressure gradient. For the mechanical model, each
 10

fracture was treated as equivalent elastic medium having the same stress-displacement

relation as the fracture deformation. The shear displacement and dilation of fracture was

neglected, and only normal deformation was considered. The coupling of fluid flow and

mechanical deformation was based on Biot’s theory of poroelasticity. The stress

dependent fracture permeability was calculated according to the fracture aperture, which

was idealized as a smooth fracture approximating the real rough fracture (Figure I-6).

Bagheri (2006), and Bagheri and Setteri (2008) developed an equivalent

continuum media for fractured porous media considering both fluid flow and a

mechanical model. For the fluid flow model, an element of fractured porous media was

transformed to an element of equivalent continuum media with a tensor permeability to

make the continuum media element have similar flow properties to the fractured medium

element. For the deformation model, the fractured medium was transformed to an

equivalent continuum poroelastic medium with the same deformation characteristics as

the fractured medium. Only normal deformation of fractures was considered. And only

small fracture deformation was allowed in the model. The fracture permeability and

porosity was dependent on pressure and stress.

 11

Figure I-6. A real rough fracture in porous media and its idealized smooth fracture in two
dimensions (Sun, 1994).

1.2.2 Discrete element method (DEM)

Gutierrez and Makurat (1997) combined a thermal reservoir fluid flow simulator

code PROFHET (Propagation of fluid and heat) and a discrete element code UDEC

(Universal distinct element code) to analyze the hydro-thermo-mechanical behavior of

fractured hydrocarbon reservoirs. Fluid flow was modeled in both the discrete fracture

network and the porous matrix, and the interface flow rate was determined by pressure

gradient between fracture and matrix. The stress change induced by fluid flow was input

into UDEC to calculate the fracture deformation. The Barton-Bandis model of fracture

deformation was applied. The results from UDEC were not used to recalculate the fluid
 12

flow. The matrix in UDEC was defined as impermeable, which reduced the coupling of

porous matrix and fluid described by Biot’s theory of poroelasticity.

Min et al. (2004) used UDEC to model the effect of stress on fracture

permeability for a fractured media (Figure I-7). The matrix was assumed as

impermeable and the fluid flow was only in the fracture network. The fracture aperture

changed with different stress loading according to the fracture deformation model. They

modeled the fracture aperture changes at various stress conditions including isotropic

stress loadings of different magnitudes and anisotropic loadings of different magnitudes

and ratios of the maximum principal stress to the minimum principal stress. Then they

modeled the flow rate through the fracture network with a fluid pressure loading. After

comparing with the Darcy’s flaw, the permeability for the fracture network was

determined and the effect of stress on the permeability of fracture network was evaluated.
 13

Figure I-7. Geometry of fracture system in the DFN model (Min et al., 2004).

1.2.3 Displacement discontinuity method (DDM)

Asgian (1988, 1989) investigated the coupled fluid and porous matrix

deformation in fractured media using an elastic DDM. The elastic DDM (an indirect

boundary element method) was developed (Crouch and Starfield, 1983) to model the

deformation of elastic nonporous media containing discontinuous surfaces (fractures).

The matrix was assumed as impermeable and fluid flow was only in fractures. The fluid

flow in fractures was coupled with the fracture deformation. The fracture permeability

was also dependent on the fracture aperture according to the cubic law and the fracture
 14

aperture varied with the change of fluid pressure and effective stress. This method

allowed the fracture to deform in normal and shear with a large displacement. But the

matrix was assumed as impermeable in the elastic DDM, which limits its application in

fractured hydrocarbon reservoirs where the matrix is the main storage.

1.3 Summary of the dissertation

Chapter I describes the problem to be solved and the objective of the study, and

also critically reviews previously published methods. Pressure reduction caused by

production in naturally fractured reservoirs induces the effective stress change. The

effective stress change affects the reservoir properties, which in turn influences the

production. Up to now the effect of production on reservoir properties including fully

coupled interactions of porous matrix, fluid and fractures in naturally fractured

reservoirs, especially fracture permeability change, has not really been addressed. The

objective of this study is to develop an approach to investigate the change in fracture

aperture and permeability in naturally fractured reservoirs.

Chapter II describes the DDM including elastic DDM and fully coupled

poroelastic DDM. The elastic DDM gives the analytical solutions of induced stress and

displacement at any point in a continuum elastic nonporous medium by a small thin

discontinuous surface (fracture) with finite length and then sums the influences of all

discontinuous surfaces (fractures) at any point using superposition. The fully coupled

DDM is based on Biot’s theory of poroelasticity. The fundamental solutions for stress

and pore pressure at any point induced either by a small discontinuous surface (fracture)
 15

with finite length or by constant rate fluid injection/production to a line source (fracture

at a well) are derived analytically. At any point, the influences by all fractures due to

displacement discontinuities or fluid injection/production can be obtained by the

superposition method. The fully coupled poroelastic DDM is verified using the classic

pressurized crack problem. Provided the stress and pore pressure change in the fractures

in a fractured porous media, the fracture aperture change can be determined using the

fully coupled poroelastic DDM.

Chapter III describes the characteristics of fracture surfaces, the nonlinear

Barton-Bandis model of fracture deformation, and the relation of fracture permeability to

fracture aperture in rough fractures. In the nonlinear fracture deformation model, the

relation of normal stress and fracture closure is represented by a hyperbolic formula.

The relation of shear stress and shear displacement is linear before yielding and too

complicated to represent using simple functions after yielding. The model also includes

shear dilation which is the fracture opening caused by shear displacement. The fracture

conductivity has a cubic relation to the effective hydraulic aperture but not the average

mechanical aperture. The effective hydraulic aperture is related with the average

mechanical aperture using the parameter for the surface roughness of fracture.

Chapter IV presents a new numerical method to determine the fluid pressure,

fracture aperture change and stress change simultaneously by combining a finite

difference method (FDM) for solving the diffusivity equation for fluid flow in fractures,

a fully coupled displacement discontinuity method (DDM) to build the global relation of

fracture deformation, and the nonlinear Barton-Bandis model of fracture deformation to

 16

build the local relation of fracture deformation. The fracture permeability changes with

the fracture aperture change.

Chapter V illustrates applications of the method described in Chapter IV under

both isotropic in situ stress conditions and highly anisotropic in situ stress conditions.

The increase of the compression stress induced by pressure depletion in naturally

fracture reservoirs tends to reduce fracture aperture and permeability, but fracture slip

caused by shear stress can increase fracture aperture and permeability due to shear

dilation.

Chapter VI gives conclusions of the dissertation and recommendations for future

work.
 17

CHAPTER II

DISPLACEMENT DISCONTINUITY METHOD

The displacement discontinuity method (DDM) is an indirect boundary element

method of solving linear elastic problems given the boundary conditions and assuming

continuous stress and discontinuous displacement at the boundaries. Crouch and

Starfield (1983) developed an elastic DDM for elastic nonporous media and applied the

elastic DDM to model the joint deformation and slip due to mining jointed rock. In the

fluid-saturated porous media, there are coupled processes between the porous matrix and

fluid. Both porous matrix deformation and fluid pressure change can cause

redistribution of stress and fluid flux. Curran and Carvalho (1987), Cheng and

Predeleanu (1987), and Carvalho (1990) presented a coupled DDM for fluid-saturated

porous media and provided the fundamental solutions of stress, displacement and pore

pressure induced by constant displacement discontinuities or continuous fluid

injection/production along a line fracture in an infinite continuum porous medium

saturated with a compressible singe-phase fluid. The induced stress and pore pressure

by the displacement discontinuities or fluid injection/production from all fractures in a

porous medium are the sum of the fundamental solutions using superposition. All

fractures in an infinite fluid-saturated porous medium are treated as boundaries. If the

change of stress and pore pressure in all fractures in the fluid-saturated system are

provided as boundary conditions, the displacement discontinuities and fluid

injection/production rate in all fractures can be determined by solving a set of linear

 18

equations established from the fully coupled DDM, and vice versa. Therefore, the fully

coupled DDM can be applied to investigate the change of fracture aperture and the

interface flow rate between fracture and matrix (similar to the fluid injection/production

rate from a fracture into the surrounded matrix) if the stress and pore pressure in all

fractures in fractured porous media is provided. A pressurized crack problem is

provided as a case to verify the fully coupled DDM and show the coupled interactions

between the fluid and porous matrix.

This chapter will describe the elastic DDM in Section 2.1. Section 2.2 will

provide the fully coupled poroelastic DDM. Section 2.3 will verify the fully coupled

poroelastic DDM with an analytical solution, and Section 2.4 will provide conclusions of

this chapter.

2.1 Elastic DDM

The elastic DDM is an indirect boundary element method to cope those problems

involving pure elastic nonporous media containing thin fractures. The elastic DDM is

based on an analytical solution for the constant discontinuity of a displacement (e.g., a

finite fracture segment) in an infinite elastic nonporous medium. For an infinite elastic

nonporous medium containing multiple fractures, the fractures are divided into N

elemental segments with the displacement in each segment assumed to have a constant

discontinuity. At any point, the influence of displacement discontinuities from all

fractures in the system can be obtained by summing the effects of all N elements using

the fundamental analytical solutions.

 19

Crouch and Starfield (1983) developed the fundamental solutions (Eq. (2-1)) of

induced stresses at any point (x, y) for an infinite two-dimensional homogeneous and

isotropic elastic nonporous medium containing a finite small thin fracture with constant

normal and shear displacement discontinuities (Figure II-1). The fracture length is 2a (a

is the half length of fracture segment) and its center is located at (0, 0).

The stress components, σ xx , σ yy , σ xy at the field point (x, y) induced by the

normal displacement discontinuity, Dn ,and Shear displacement discontinuity, Ds, are

given by

 ∂2 f ∂3 f   ∂2 f ∂3 f 
σ xx = 2GDn  + y  + 2GDs  2 +y 
 ∂y
2
∂y 3   ∂x∂y ∂x∂y 2 

 ∂2 f ∂3 f  ∂3 f
σ yy = 2GDn  − y  − 2GDs y (2-1)
 ∂y
2
∂y 3  ∂x∂y 2

∂3 f  ∂2 f ∂3 f 
σ xy = −2GDn y + 2GD 
s + y 
∂x∂y 2  ∂y
2
∂y 3 

where G is the shear modulus, and f is a function of the position (x, y) of the field point

relative to the center of the fracture and the half length of the fracture segment a given

by:

  y 
f ( x, y ) = −
1 y
 y arctan − arctan 
4π (1 − υ )   x−a x+a , (2-2)
− (x − a )ln (x − a ) + y + (x + a )ln (x + a ) + y 
2 2 2 2


with Poisson’s ratio, υ. Note that in this dissertation SI units are used in all equations

except for the specified equations, but oilfield units are shown in the results.
 20

+Ds +Dn

2a

Figure II-1. A small discontinuous segment in an infinite two-dimensional nonporous medium (after
Crouch and Starfield, 1983).

Following is an illustration of the elastic DDM method:

The curvy fracture in a two-dimensional infinite nonporous medium shown in

Figure II-2 is discretized into 5 segments and the influence of displacement

discontinuities on an arbitrary field point (x, y) from the curvy fracture can be

approximated by summing the influences from the 5 fracture segments on the point (x, y).
 21

(x,y)

2 a5
(x5,y5)
2 a4
(x4,y4)
y 2 a3
(x3,y3)
2 a2
2 a1 (x2,y2)
(x1,y1)

Figure II-2. A curvy fracture discretized into 5 segments in an infinite two-dimensional nonporous
medium.

The fundamental solutions (Eq. (2-1)) are for a fracture segment parallel to the x-

axis and the center of the fracture segment located at (0, 0). To apply the fundamental

solutions, the field point (x, y) shown in Figure II-2 must be transformed into a local co-

ordinate system for the jth fracture segment with an angle βj with x-axis, as in Figure II-3.

The x -axis of the local co-ordinate system is parallel to the orientation of the jth fracture

segment. The field point (x, y) is transformed to the local x , y co-ordinate.

x = (x − x j )cos β j + ( y − y j )sin β j
(2-3)
y = −(x − x j )sin β j + ( y − y j )cos β j
 22

where (xj, yj) is the midpoint of the jth fracture segment. The induced stresses on the

field point in the local x , y co-ordinate system by the normal and shear displacement

discontinuities of the jth fracture are:

j j  ∂2 f ∂3 f  j 
∂2 f ∂3 f 
σ xx = 2G D n  + y  + 2G D s  2 +y 
 ∂y
2
∂y 3   ∂x ∂y ∂x ∂y 2 

j  ∂2 f
j
∂3 f  j
∂3 f
σ yy = 2G D n  2 − y 3  − 2G D s y (2-4)
 ∂y ∂y  ∂x ∂y 2

j  2 
j
∂3 f
j
∂ f ∂3 f
σ xy = −2G D n y + 2G D s  2 + y 3 
∂x ∂y 2  ∂y ∂y 

where

  y 
f (x , y ) = −
1 y
 y  arctan − arctan
4π (1 − υ )   x − aj x + a j  (2-5)
− (x − a j )ln (x − a ) + y 2 + (x + a j )ln (x + a ) + y2 
2 2
j j 

The induced stresses on the field point (x, y) in the x, y co-ordinate system by the jth

fracture segment are obtained by transforming the Eq. (2-4) from the local x , y co-

ordinate system to the x, y co-ordinate system using the transformation formula (Crouch

and Starfield, 1983).

j j j j
σ xx = σ xx cos 2 β j − σ xy sin 2β j + σ yy sin 2 β j

j j j j
σ yy = σ xx sin 2 β j + σ xy sin 2β j + σ yy cos 2 β j (2-6)

σ xy =  σ xx − σ yy  sin β j cos β j + σ xy (cos 2 β j − sin 2 β j )

j
 j j
 j

 
 23

The induced stresses on an arbitrary point (x, y) by the displacement discontinuities of

the curvy fracture are approximated by superposition as the sum of the influences from

all 5 fracture segments.

5 j
σ xx = ∑ σ xx
j =1

5 j
σ yy = ∑ σ yy (2-7)
j =1

5 j
σ xy = ∑ σ xy
j =1

(x,y)

y x
y βj
2 aj
(xj,yj)

Figure II-3. Local co-ordinate for the jth fracture segment in an elastic nonporous medium.
 24

If the field point (x, y) coincides with the midpoint (xi, yi) of the ith fracture

segment, the Eq. (2-4) are the induced stresses on the ith fracture segment by the normal

and shear displacement discontinuities of the jth fracture segment in the local x , y co-

ordinate system (Figure II-4). The induced stresses on the ith fracture segment by the jth

fracture segment can be transformed into normal and shear stresses to the ith fracture

segment using the following formula.

ij j ij j j ij
σ n = σ xx cos 2 γ + σ xy sin 2γ '+ σ yy sin 2 γ
(2-8)
ij
 j j
 ij ij j
 ij ij

σ s =  σ xx − σ yy  sin γ cos γ − σ xy  cos 2 γ − sin 2 γ 
   

ij
π ij ij
where γ = + β i − β j , and σ n , σ s are the induced normal and shear stresses on the ith
2

fracture segment by the jth fracture segment. Combining Eqs. (2-4) and (2-8) yields:
ij ij j ij j
σ n = A Dn + B Ds
(2-9)
ij ij j ij j
σ s = E Dn + F Ds

where

ij
 ∂2 f  ij ij
 ∂3 f ij
∂3 f 
A = 2G  2 +  cos 2 γ − sin 2 γ  y 3 − sin 2 γ y
 ∂y   ∂y ∂x ∂y 2 

ij  ij
 ∂2 f ∂3 f  ij
∂3 f ij
 ∂2 f ∂ 3 f 
B = 2G cos 2 γ  2 +y  − sin 2
γ y + sin 2 γ  + y 
  ∂x ∂y ∂x ∂y 2  ∂x ∂y 2  ∂y 2
 ∂y 3 
(2-10)
ij
 ∂ f 
ij 3 ij ij
 ∂3 f 
E = 2Gy sin 2 γ 3 +  cos 2 γ − sin 2 γ  2
 ∂y   ∂x ∂y 
 25

ij  ij
 ∂2 f ∂ 3 f   2 ij 2  ∂ f
ij 2
∂ 3 f 
F = 2G sin 2 γ  +y  −  cos γ − sin γ 
 2 + y 
  ∂x ∂y ∂x ∂y 2    ∂y ∂y 3 

The induced normal and shear stresses on the ith fracture segment by the normal and

shear displacement discontinuities of all fracture segments are obtained by summing the

solutions in Eq. (2-9).

i 5 ij j 5 ij j
σ n = ∑ A Dn + ∑ B Ds
j =1 j =1

(2-11)
i 5 ij j 5 ij j
σ s = ∑ E Dn + ∑ F Ds
j =1 j =1

γ
2 ai
y x (xi,yi)

y β
2 aj
(xj,yj)

Figure II-4. Influence of jth fracture segment on the ith fracture segment in an elastic nonporous
medium.
 26

If the displacement discontinuities of the curvy fracture in the example shown in

Figure II-2 are unknown variables, but the normal and shear stresses in the curvy

fracture are known, the induced displacement discontinuities of the curvy fracture for the

stresses on the fracture can be obtained by simultaneously solving the following 10 sets

of linear equations (shown in matrix form) built from Eq. (2-11).

 1 
 11 B A B A B A B A B  D n   σ n 
11 12 12 13 13 14 14 15 15 1
A
 11 11 12 12 13 13 14 14 15 15  1  1 
E F E F E F E F E F  D s   σ s 
 21 21 22 22 23 23 24 24 25 25  2  2 
A B A B A B A B A B  Dn  σ n 
 21   2 
E   σ s 
21 22 22 23 23 24 24 25 25 2
F E F E F E F E F Ds
    
 31 31 32 32 33 33 34 34 35 35
 3   3 
A B A B A B A B A B  D n   σ n 
= 3 (2-12)
 31 31 32 32 33 33 34 34 35 35  3   
E F E F E F E F E F  D s   σ s 
 41 41 42 42 43 43 44 44 45 45  4  4 
A B A B A B A B A B  D n   σ n 
 41 41 42 42 43 43 44 44 45 45  4  4 
E F E F E F E F E F  D s   σ s 
 51  5   5 
B A B A B A B A B  D n   σ n 
51 52 52 53 53 54 54 55 55
A
    
E F E F E F E F E F  D s   σ s 
51 51 52 52 53 53 54 54 55 55 5 5

This method is for elastic nonporous media. Next section will give the DDM in

porous media saturated with compressible single-phase fluid.

2.2 Fully coupled DDM for porous media saturated with a compressible single-phase

fluid

The interaction of fluid and porous matrix plays a key role in the matrix

deformation and fluid flow in the fluid-saturated porous media. The porous matrix
 27

deformation causes pore pressure change, thereby causing pressure diffusion. A

disturbance in the pore pressure also causes deformation of the solid matrix. Biot (1941)

developed a theory of poroelasticity for a porous medium saturated with an

incompressible fluid. The theory of poroelasticity was extended to the porous media

saturated with compressible fluid by Rice and Cleary (1976). Based on the theory of

poroelasticity, Carvalho (1990) gave the fundamental solutions of induced stress and

pore pressure for a finite thin fracture segment with a fluid injection/production source in

an infinite two-dimensional homogeneous and isotropic porous medium saturated with a

compressible single-phase fluid. The induced stress and pore pressure by a single long

fracture or many fractures with fluid injection/production can be obtained by discretizing

the fracture or fractures into N fracture segments and summing the influences of all N

fracture segments. If the induced stress and pore pressure in fractures are known, the

normal and shear deformation of fractures and fluid injection/production rate (interface

flow rate) in these fractures can be obtained by numerically solving a set of linear

equations built from the fundamental solutions.

This section will start from the constitutive equations of a porous medium

saturated with a compressible single-phase fluid in subsection 2.2.1. The constitutive

equations give the relations of stress, pore pressure, strain and fluid volume changes.

Then the pressure diffusion equation for flow in the porous medium will be given in

subsection 2.2.2. Based on the coupled constitutive equations and the pressure diffusion

equation, the fundamental solutions of induced stress and pore pressure for a single finite

thin fracture segment under constant displacement discontinuities or constant rate fluid
 28

injection/production in an infinite two-dimensional homogeneous and isotropic porous

medium saturated with a compressible single-phase fluid will be given in subsection

2.2.3. Subsection 2.2.4 will describe how superposition of the fundamental solutions

enables consideration of a long fracture or many fractures, and the subsection 2.2.5 will

give a method for determining the normal and shear fracture deformation and fluid

injection/production rate given the time dependent stress and pore pressure in fractures.

2.2.1 Constitutive equations of a porous medium saturated with a compressible single-

phase fluid

The relation of stress to strain and pore pressure for a linear isotropic poroelastic

medium is given by Biot’s theory of poroelasticity (Biot, 1941):

 υ 
σ xx = 2G exx + ekk  − α p
 1 − 2υ 
 υ 
σ yy = 2G e yy + ekk  − α p
 1 − 2υ 
 υ  (2-13)
σ zz = 2G ezz + ekk  − α p
 1 − 2υ 
σ xy = 2Gexy
σ xz = 2Gexz
σ yz = 2Geyz

where σxx, σyy, σzz, σxy, σxz, and σyz are stress components and exx, eyy, ezz, exy, exz, and

eyz are strain components of the porous medium, ekk is the volumetric strain

( ekk = exx + eyy + ezz ), p is the pore pressure, α is Biot’s poroelastic coefficient. Tensile
 29

stress and strain are treated as positive in this dissertation. The strain is defined

according to the displacement:

∂u x
exx =
∂x
∂u y
e yy =
∂y
∂u
ezz = z
∂z
1  ∂u ∂u  (2-14)
exy =  x + y 
2  ∂y ∂x 
1  ∂u ∂u 
exz =  x + z 
2  ∂z ∂x 
1  ∂u ∂u 
e yz =  y + z 
2  ∂z ∂y 

where ux, uy and uz are the components of displacement of the porous medium along x, y

and z direction, respectively. The static solid is subject to the following force balance

(Biot, 1941):

∂σ xx ∂σ xy ∂σ xz
+ + =0
∂x ∂y ∂z
∂σ xy ∂σ yy ∂σ yz
+ + =0 (2-15)
∂x ∂y ∂z
∂σ xz ∂σ yz ∂σ zz
+ + =0
∂x ∂y ∂z

Combining Eqs. (2-13), (2-14) and (2-15) yields the Navier equation of poroelasticity:
 30

 ∂ 2u x ∂ 2u x ∂ 2 u x   ∂ 2u x ∂ 2 u y ∂ 2 u z 
G 2 + 2 + 2  +
G  + +  − α ∂p = 0
∂z  1 − 2ν  ∂x 2 ∂y∂x ∂z∂x 
 ∂x ∂y   ∂x
 ∂ 2u y ∂ 2u y ∂ 2u y   ∂ 2u x ∂ 2u y ∂ 2u z 
G 2 + 2 + 2 +
G  + 2 +  − α ∂p = 0 (2-16)
 ∂x ∂y  
∂z  1 − 2ν  ∂x∂y ∂y ∂z∂y  ∂y

 ∂ 2u ∂ 2u ∂ 2u  G  ∂ 2u x ∂ u y ∂ 2u z  ∂p
2

G 2z + 2z + 2z  + + + 2 −α =0
 ∂x ∂y ∂z  1 − 2ν  ∂x∂z ∂y∂z ∂z  ∂z

The total volumetric deformation ( ekk ) of the porous medium consists of the

pore space change ( ς p ) and the deformation of the solid porous matrix ( ς s ). The

deformation of the solid porous matrix is due to the fluid pressure and effective stress

loading:

(i) the effect of fluid pressure (the compression stress or strain is negative):

ς s1 = −
p
(1 − φ ) (2-17)
Ks

(ii) the effect of effective stress loading

σ kk'
ς s2 = (2-18)
3K s

where K s is the bulk modulus of the solid and φ is the porosity. The average effective

stress ( σ kk' / 3 ) has the following relation with the volumetric strain and pore pressure

(Carvalho, 1990):

σ kk' σ xx' + σ yy' + σ zz' Km

= = K mekk + p (2-19)
3 3 Ks
 31

where K m (Km < Ks) is the bulk modulus of the porous matrix. Combining Eqs. (2-17)

and (2-18), and substituting Eq. (2-19) result in the deformation of the solid porous

matrix:

p  Km 
− (1 − φ )
Km
ςs = ekk +  (2-20)
Ks Ks  Ks 

The pore space change is obtained by subtracting the deformation of the solid porous

matrix from the total volumetric strain and using the definition of Biot’s coefficient, α,

( α = 1 − K m / K s ):

ς p = αekk + (α − φ )
p
(2-21)
Ks

2.2.2 Pressure diffusion in a porous medium

The fluid mass balance equation in a porous medium (matrix) gives that the fluid

flowing into/out is equal to the sum of the increase of fluid mass in the pore space and

injected/produced fluid:

∂ (ρ f qx ) ∂ (ρ f q y ) ∂ (ρ f qz ) ∂ (ρ f V f )
− − − = − ρ f qs (2-22)
∂x ∂y ∂z ∂t

where ρ f is the fluid density, qx, qy, qz are the fluid flow rate components in x, y, z

direction, respectively, V f is pore space, qs is the injection/production rate and t is time.

The fluid is compressible and the fluid density is pressure dependent:

∂ρ f
= co ρ f (2-23)
∂p
 32

where co is the fluid (for example oil) compressibility.

In a unit volume porous media, the pore volume is φ, and the pore volume

change is ς p , and the RHS of Eq. (2-22) is rewritten as:

∂ (ρ f qx ) ∂ (ρ f q y ) ∂ (ρ f qz ) ∂ (ρ f ) ∂ (ς p )
− − − =φ + ρf − ρ f qs (2-24)
∂x ∂y ∂z ∂t ∂t

Assuming Darcy’s Law for fluid flow,

kAx ∂p
qx = −
µ ∂x
kAy ∂p
qy = − (2-25)
µ ∂y
kAz ∂p
qz = −
µ ∂z

where k is the matrix permeability and assumed as homogeneous and isotropic, Ax, Ay,

Az are the cross section areas in x, y, and z direction, respectively, and µ is the fluid

viscosity.

Substituting Eqs.(2-21), (2-23) and (2-25) into Eq. (2-24), neglecting the term with

∂p 2
( ) (Lee et al., 2003) and assuming small change in the pore volume (noting that the
∂xi

cross section area for a unit volume is 1) yield:

k  ∂ 2 p ∂ 2 p ∂ 2 p  1 ∂p ∂e
 2 + 2 + 2  = + α kk − qs (2-26)
µ  ∂x ∂y ∂z  M ∂t ∂t

1 α −φ
where = φco + and M is the Biot modulus. In Eq. (2-26), the left side is the net
M Ks

1 ∂p
flow rate into the unit porous medium from the boundaries, the first right term ( )
M ∂t
 33

is the fluid volume change due to the pore pressure change, the second right term

∂ekk
(α ) is the fluid volume change due to the effective stress change, and the final term
∂t

( qs ) is a source term.

2.2.3 Fundamental solutions for a single fracture segment in an infinite two-dimensional

porous medium

The fundamental solutions of poroelastic DDM include induced stress,

displacement and pore pressure from both the pressure/flow rate disturbance and the

displacement discontinuities. For a plane strain condition (three-dimensions are

simplified to two-dimensions), there is a constant discontinuity in the media and also

constant flow (injection or production) along a thin fracture with a length of 2a from t=0

(Figure II-5). The initial conditions are defined in Eq. (2-27) and the inner and outer

boundary conditions are defined in Eqs. (2-28) and (2-29). Since only the induced

solutions for changes in stress, displacement and pore pressure are needed, the initial

values of stress, displacement and pore pressure are set as zero.

(i) The initial conditions are given by

p = 0

at t = 0, ∀x, y u x = u y = 0 (2-27)

σ xx = σ yy = σ xy = 0
 34

(ii) Boundary conditions are given by

Inner boundary:

( ) ( )
 u x x,0− − u x x,0+ = Ds

( ) (
at y = 0, x ≤ a u y x,0− − u y x,0+ = Dn) (2-28)

qs = −2a q0

where q0 is a unit flow rate (q0=1 m3/sec).

Outer boundary:

u x = u y = 0
at x 2 + y 2 → ∞  (2-29)
σ xx = σ yy = σ xy = p = 0

Dn+
Ds+
qs
x

Figure II-5. A thin line fracture in an infinite two-dimensional elastic porous medium, and the line
fracture starts from (-a,0) and ends at (a,0).
 35

Using the initial and boundary conditions (Eqs. (2-27) – (2-29)), Eqs. (2-16) and

(2-26) can be solved for separate inner boundary conditions – constant volume flow rate

injection/production (ux(x, 0-)- ux(x, 0+)=0, uy(x, 0-)- uy(x, 0+)=0, qs =-2aq0) at the inner

boundary and constant displacement discontinuity (DD) (ux(x, 0-)- ux(x, 0+)=Ds, uy(x, 0-)-

uy(x, 0+)= Dn, qs =0) at the inner boundary (Carvalho, 1990). The induced displacement,

pore pressure and stress at any point (x, y) and time t by the constant volume

injection/production rate and by the displacement discontinuities including normal and

shear displacement discontinuities through the fracture segment are given in the

Appendix A and Appendix B, respectively (Carvalho, 1990). The final fundamental

solutions for poroelastic DDM are obtained by combining the solutions of the constant

volume rate fluid injection/production and the constant displacement discontinuities in

the fracture segment.

Induced pore pressure:

p ( x, y, t ) = p dn ( x, y, t ) Dn + p ds ( x, y, t ) Ds + p q ( x, y, t ) qint (2-30)

Induced displacement:

u x ( x, y, t ) = u xdn ( x, y, t ) Dn + u xds ( x, y, t ) Ds + u xq ( x, y, t ) qint

(2-31)
u y ( x, y, t ) = u ydn ( x, y, t ) Dn + u yds ( x, y, t ) Ds + u yq ( x, y, t ) qint

Induced stress:
 36

σ xx ( x, y, t ) = σ xxdn ( x, y, t ) Dn + σ xxds ( x, y, t ) Ds + σ xxq ( x, y, t ) qint

σ yy ( x, y, t ) = σ yydn ( x, y, t ) Dn + σ yyds ( x, y, t ) Ds + σ yyq ( x, y, t ) qint (2-32)

σ xy ( x, y, t ) = σ xydn ( x, y, t ) Dn + σ xyds ( x, y, t ) Ds + σ xyq ( x, y, t ) qint

where Dn and Ds are the normal and shear displacement discontinuity sources, and qint is

the fluid source term in a fracture (interface flow rate between fracture and matrix), and

the superscripts dn, ds and q denote normal displacement discontinuity source, shear

displacement discontinuity source and fluid source, respectively. The induced pore

pressure, p q , displacement in x direction, u xq and in y direction, u yq , stress components,

σ xxq , σ yyq and σ xyq by the constant rate fluid injection/production from a fracture segment

are listed in Appendix A. The induced pore pressure, p dn and p ds , displacement in x

direction, u xdn and u xds , and in y direction, u ydn and u yds , stress components, σ xxdn , σ yydn ,

σ xydn , σ xxds , σ yyds and σ xyds by the constant normal and shear discontinuous displacement of a

fracture segment are listed in Appendix B.

2.2.4 Solutions for multiple fracture segments in an infinite two-dimensional porous

medium

For a long fracture or many fractures in a porous medium saturated with a

compressible single-phase fluid, the induced stresses and pore pressure can be

approximated by summing the fundamental solutions for a system of fracture segments.

Figure II-6 shows a porous medium containing a curvy fracture like the one in section
 37

2.1 that was in a nonporous medium. The curvy fracture is discretized into 5 fracture

segments shown in Figure II-6. To apply the fundamental solutions to the jth fracture

segment, it is necessary to transform the x, y co-ordinates of the segment into the local x ,

y co-ordinate system using the transformation formula in Eq. (2-3). The pore pressure

and stresses induced by the normal and shear displacement discontinuities and the fluid

injection/production of the jth fracture segment in the local x , y co-ordinate system

(Figure II-7) are given in Eqs. (2-33) and (2-34), respectively.

Induced pore pressure:

j j j j j j j
p ( x , y , t ) = p ( x , y , t ) D n + p ( x , y , t ) D s + p q ( x , y , t ) q int
dn ds
(2-33)

Induced stress:

j j j j j j j
σ xx ( x , y , t ) = σ xdnx ( x , y , t ) D n + σ xdsx ( x , y , t ) D s + σ xqx ( x , y , t ) q int

j j j j j j j
σ yy ( x , y , t ) = σ ydny ( x , y , t ) D n + σ ydsy ( x , y , t ) D s + σ yqy ( x , y , t ) q int (2-34)

j j j j j j j
σ xy ( x , y , t ) = σ xdny ( x , y , t ) D n + σ xdsy ( x , y , t ) D s + σ xqy ( x , y , t ) q int
 38

(x,y)

2 a5
(x5,y5)
2 a4
(x4,y4)

y 2 a3
(x3,y3)
2 a2
2 a1 (x2,y2)
(x1,y1)

Figure II-6. A curvy fracture discretized into 5 segments in an infinite two-dimensional porous
medium saturated with a single-phase fluid.

The stresses induced by the jth fracture segment in the local x , y co-ordinate system

can be transformed to the x, y co-ordinate system using the transformation formula in Eq.

(2-6). Now the induced stresses from all 5 fracture segments can be obtained by

superposition. (Eq. (2-7)). As the pore pressure is a scalar, it is independent of the

orientation of the co-ordinate system. The induced pore pressure by the jth fracture in

the x, y co-ordinate system is the same as that in the local x , y co-ordinate system.

j j
p ( x, y , t ) = p ( x , y , t ) , (2-35)

And the induced pore pressure by the curvy fracture can be obtained by summing the

induced pore pressure from all 5 fracture segments.

 39

5 j
p ( x, y , t ) = ∑ p ( x, y , t ) (2-36)
j =1

(x,y)

y x
y βj
2 aj
(xj,yj)

Figure II-7. Local co-ordinate for the jth fracture segment in a porous medium.

The normal and shear stresses induced on the ith fracture segment by the jth

fracture segment shown in Figure II-8 are obtained by projecting the stresses in Eq. (2-

34) to the plane of the ith fracture using the formula in Eq. (2-8). The normal and shear

stresses and pore pressure induced on the ith fracture segment by the constant rate fluid

injection/production and the constant normal and shear displacement discontinuities of

the jth fracture are:

 40

ij ij j ij j ij j
σ n = A D n + B D s + C q int

ij ij j ij j ij j
σ s = E D n + F D s + K q int (2-37)

ij ij j ij j ij j
p = L D n + H D s + N q int

where

   ij ij   ij ij   ij ij 
ij ij ij ij j ij j ij j
A x, y, t  = cos 2 γ σ xdnx  x, y, t  + sin 2 γ σ xdny  x, y, t  + sin 2 γ σ ydny  x, y, t 
       
   ij ij   ij ij   ij ij 
ij ij ij ij j ij j ij j
B x, y, t  = cos 2 γ σ xdsx  x, y, t + sin 2 γ σ xdsy  x, y, t  + sin 2 γ σ ydsy  x, y, t 
       
 ij ij   ij ij   ij ij   ij ij 
ij ij j ij j ij j
C  x, y, t  = cos 2 γ σ xqx  x, y, t  + sin 2 γ σ xqy  x, y, t  + sin 2 γ σ yqy  x, y, t 
       
ij  j
   ij ij   ij ij      ij ij 
ij ij ij ij j ij ij j
E  x, y, t  = sin γ cos γ  σ xdnx  x, y, t  − σ ydny  x, y, t  −  cos 2 γ − sin 2 γ  σ xdny  x, y, t 
          
ij  j
   ij ij   ij ij     ds  ij ij 
ij ij ij ij j ij ij j
F  x, y, t  = sin γ cos γ  σ xdsx  x, y, t − σ ydsy  x, y, t  −  cos 2 γ − sin 2 γ  σ x y  x, y , t  (2-38)
          
ij  j
   ij ij   ij ij      ij ij 
ij ij ij ij j ij ij j
K  x, y, t  = sin γ cos γ  σ xqx  x, y, t  − σ yqy  x, y, t  −  cos 2 γ − sin 2 γ σ xqy  x, y, t 
          
   ij ij 
ij ij ij j
L x, y, t  = p dn  x, y, t 
   
 ij ij   ij ij 
ij j
H  x, y, t  = p ds  x, y, t 
   
 ij ij   ij ij 
ij j
N  x, y , t  = p q  x, y , t 
   
where

x = (xi − x j )cos β j + ( yi − y j )sin β j

ij

(2-39)

y = −(xi − x j )sin β j + ( yi − y j )cos β j

ij
 41

2 ai βi
y x (xi,yi)
y βj
2 aj
(xj,yj)

Figure II-8. Influence of the jth fracture segment on the ith fracture segment in an elastic porous
medium.

The normal and shear stresses and pore pressure induced on the ith fracture segment by

the constant rate fluid injection/production and constant normal and shear displacement

discontinuities of all fracture segments are obtained by summing the solutions in Eq. (2-

37).

i 5 ij j 5 ij j 5 ij j
σ n = ∑ A D n + ∑ B D s + ∑ C q int
j =1 j =1 j =1

i 5 ij j 5 ij j 5 ij j
σ s = ∑ E D n + ∑ F D s + ∑ K q int (2-40)
j =1 j =1 j =1

i 5 ij j 5 ij j 5 ij j
p = ∑ L D n + ∑ H D s + ∑ N q int
j =1 j =1 j =1
 42

2.2.5 Determination of the fracture discontinuous displacement

Up to now we have determined normal and shear stresses and pressure given

discontinuous displacements in the fractures and fluid injection/production sources.

However, the practical application may require determination of the fracture

discontinuous displacement given stress and fluid pressure in fractures. Because the

stress and pore pressure changes induced by the constant rate fluid injection/production

and displacement discontinuities of fractures are a function of time, it is necessary to

account for the time dependent changes. For time dependent normal displacement

discontinuity, Dn, shear displacement discontinuity, Ds, or injection/production flow rate

(interface flow rate between fracture and matrix), qint, a time marching scheme like that

shown in Figure II-9 is used to discretize the time dependent quantity into N constant

steps and use superposition to account for each step change at the time it occurs. The

constant step source except for the first one does not start at the time zero (t=0). Thus a

time shift is needed to apply the fundamental solution and the influence coefficients. For

example, at time τξ , if constant ∆Dn(xj,yj,τξ), ∆Ds(xj,yj,τξ) and ∆qint(xj,yj,τξ) of the jth

fracture segment are added, the induced stresses and pore pressure on the ith fracture

segment at time t by the added sources will be :

 43

jξ jξ jξ
σ n = ∑ A(t − τ ξ ) ∆ D n + ∑ B (t − τ ξ ) ∆ D s + ∑ C (t − τ ξ ) ∆ q int
i 5 ij 5 ij 5 ij

j =1 j =1 j =1

jξ jξ jξ
σ s = ∑ E (t − τ ξ ) ∆ D n + ∑ F (t − τ ξ ) ∆ D s + ∑ K (t − τ ξ ) ∆ q int
i 5 ij 5 ij 5 ij
(2-41)
j =1 j =1 j =1

jξ jξ jξ
p = ∑ L(t − τ ξ ) ∆ D n + ∑ H (t − τ ξ ) ∆ D s + ∑ N (t − τ ξ ) ∆ q int
i 5 ij 5 ij 5 ij

j =1 j =1 j =1

jξ jξ jξ
where ∆ D n , ∆ D s and ∆ q int denote the increments of normal displacement

discontinuity, shear displacement discontinuity and injection/production flow rate

A(t − τ ξ ) , B (t − τ ξ ) ,
ij ij
(interface flow rate) of the jth fracture segment at time τξ.

C (t − τ ξ ) , E (t − τ ξ ) , F (t − τ ξ ) , K (t − τ ξ ) , L(t − τ ξ ) , H (t − τ ξ ) , and N (t − τ ξ ) are the

ij ij ij ij ij ij ij

influence coefficients of jth fracture segment on the ith fracture element at time step ξ

and defined in the Eq. (2-38). The total induced stresses and pore pressure on the ith

fracture segment at time t are obtained by summing the influences from all time steps.

ξ 5 ij jh ξ 5 ij jh ξ 5 ij jh
σ n (t ) = ∑ ∑ A(t − τ h ) ∆ D n + ∑ ∑ B(t − τ h ) ∆ D s + ∑ ∑ C (t − τ h ) ∆ qint
i

h=0 j =1 h=0 j =1 h=0 j =1

ξ 5 ij jh ξ 5 ij jh ξ 5 ij jh
σ s (t ) = ∑ ∑ E (t − τ h ) ∆ D n + ∑ ∑ F (t − τ h ) ∆ D s + ∑ ∑ K (t − τ h ) ∆ qint
i
(2-42)
h=0 j =1 h=0 j =1 h=0 j =1

ξ 5 ij jh ξ 5 ij jh ξ 5 ij jh
p (t ) = ∑ ∑ L(t − τ h ) ∆ D n + ∑ ∑ H (t − τ h ) ∆ D s + ∑ ∑ N (t − τ h ) ∆ qint
i

h=0 j =1 h=0 j =1 h=0 j =1

where h is the time step index.

 44

χ(xj,yj,τ)
∆χ(xj,yj,τξ)

∆χ(xj,yj,τ1)
∆χ(xj,yj,τ0)
τ0 τ1 τξ t τ

Figure II-9. Time marching scheme, χ represents Dn, Ds or qint.

If the induced stresses and pore pressure at all fracture segments shown in Figure

II-6 are known, the step change of normal and shear displacement discontinuities and

injection/production flow rate can be solved from τ0 to τξ. Firstly, at time τ0 (τ0=0), the

induced stresses and pore pressure on the ith fracture segment from τ0 to τ1 are known,

Eq. (2-42) is rewritten as Eq. (2-43) (note that there is only one time step).

5 ij j0 5 ij j0 5 ij j0
σ n (τ 1 ) = ∑ A(τ 1 − τ 0 ) ∆ D n + ∑ B(τ 1 − τ 0 ) ∆ D s + ∑ C (τ 1 − τ 0 ) ∆ q int
i

j =1 j =1 j =1

5 ij j0 5 ij j0 5 ij j0
σ s (τ 1 ) = ∑ E (τ 1 − τ 0 ) ∆ D n + ∑ F (τ 1 − τ 0 ) ∆ D s + ∑ K (τ 1 − τ 0 ) ∆ q int
i
(2-43)
j =1 j =1 j =1

5 ij j0 5 ij j0 5 ij j0
p(τ 1 ) = ∑ L(τ 1 − τ 0 ) ∆ D n + ∑ H (τ 1 − τ 0 ) ∆ D s + ∑ N (τ 1 − τ 0 ) ∆ q int
i

j =1 j =1 j =1

One set of linear equations can be built from Eq. (2-43), and the increment of normal

displacement discontinuity, shear displacement discontinuity and injection/production

flow rate (interface flow rate) for all fracture segments at time τ0 can be solved from the

set of linear equations. By the similar way, the step sources at other time steps can be
 45

solved. For the last time step, the induced stresses and pore pressure at time t are known

and the step sources before the step τξ are already solved, and only the last step sources

are not known and need to be solved (2-44).

jξ jξ jξ

∑ A(t − τ ξ ) ∆ D n + ∑ B(t − τ ξ ) ∆ D s + ∑ C (t − τ ξ ) ∆ q int =

5 ij 5 ij 5 ij

j =1 j =1 j =1
ξ −1 5 ij jh ξ −1 5 ij jh ξ −1 5 ij jh
σ n (t ) − ∑ ∑ A(t − τ h ) ∆ D n − ∑ ∑ B(t − τ h ) ∆ D s − ∑ ∑ C (t − τ h ) ∆ q int
i

h =0 j =1 h =0 j =1 h =0 j =1

jξ jξ jξ

∑ E (t − τ ξ ) ∆ D n + ∑ F (t − τ ξ ) ∆ D s + ∑ K (t − τ ξ ) ∆ q int =
5 ij 5 ij 5 ij

j =1 j =1 j =1
ξ −1 5 ξ −1 5 ξ −1 5
(2-44)
ij jh ij jh ij jh
σ s (t ) − ∑ ∑ E (t − τ ) ∆ D − ∑ ∑ F (t − τ ) ∆ D − ∑ ∑ K (t − τ ) ∆ q
i
h n h s h int
h =0 j =1 h =0 j =1 h =0 j =1

jξ jξ jξ

∑ L(t − τ ξ ) ∆ D + ∑ H (t − τ ξ ) ∆ D + ∑ N (t − τ ξ ) ∆ q
5 ij 5 ij 5 ij
n s int =
j =1 j =1 j =1
ξ −1 5 ij jh ξ −1 5 ij jh ξ −1 5 ij jh
p(t ) − ∑ ∑ L(t − τ h ) ∆ D n − ∑ ∑ H (t − τ h ) ∆ D s − ∑ ∑ N (t − τ h ) ∆ q int
i

h =0 j =1 h =0 j =1 h =0 j =1

Another set of linear equations can be built from Eq. (2-44) and the increment of normal

displacement discontinuity, shear displacement discontinuity and injection/production

flow rate /interface flow rate for all fracture segments at time τξ can be solved from the

set of linear equations. The final normal and shear displacement discontinuities and

injection/production flow rate (interface flow rate) of every fracture segment at time t

can be obtained by summing all of these step increments.

 46

j ξ jh
Dn = ∑ ∆ Dn
h =0

j ξ jh
Ds = ∑ ∆ Ds (2-45)
h =0

j ξ jh
q int = ∑ ∆ q int
h =0

2.3 Model verification

It is difficult to find analytical solutions for most real problems. Numerical

methods have the advantage to solve the real problems sometimes with very complicated

boundary conditions. A few special problems with simple boundary conditions can be

solved analytically, and these analytical solutions are very helpful to check and verify

the numerical solution by the DDM. Here, the DDM is applied to a line crack in an

infinite medium.

An infinitely thin line crack with a length of ∆L in an infinite elastic medium is

subject to a constant pressure (tensile stress) p along the crack surfaces (Figure II-10).

The normal relative displacement of the two crack surfaces (opening), wf , was solved by

Sneddon (1951).

wf =
(1 − v ) p ∆L 1−
x2
(2-46)
G (∆L / 2)2
where − ∆L / 2 ≤ x ≤ ∆L / 2 .
 47

Figure II-10. A line crack with constant pressure loading.

This problem can be solved using the DDM. The line crack is separated into N

segments, each of which represents an elemental displacement discontinuity. And the

displacement of every segment can be solved by applying the boundary conditions

(constant pressure along the crack surfaces). For a elastic nonporous medium with a

shear modulus of 9.06×105 psi and a Poisson’s ratio of 0.2, there is an infinite thin line

crack with a length of 39.37 in, and a constant injection pressure of 145 psi above the

reservoir pressure (∆p = pinj – p0 = 145 psi) applied to the crack surfaces. The original

crack aperture is assumed as zero and the effective stress is zero. Figure II-11 shows that

the crack width modeled using the DDM is consistent with the analytical solution.
 48

6.E-03

5.E-03

4.E-03
Crack width (in)

Elastic DD
Analytic
3.E-03

2.E-03

1.E-03

0.E+00
-20 -15 -10 -5 0 5 10 15 20
x(in)

Figure II-11. Comparison of the modeled crack width using elastic DD with the analytical solution.

If the elastic porous medium is saturated with fluid, a constant fluid pressure

applied to the crack surfaces will cause a transient crack opening. In addition to the

stress applied to the crack surfaces, there is also a fluid pressure applied to the pore

pressure field in the porous medium. It is common to separate the pressure application

into two loading processes (Detournay and Cheng, 1988): (i) Mode I loading – normal

stress loading; (ii) Mode II loading – pore pressure loading. Mode I loading tends to

open the crack. But the opening of crack will cause a compression on the porous

material around the crack. At very early time stage, the fluid in the pores cannot move

out and the porous material shows undrained behavior, and the pore pressure around the

crack has an instant increase. The induced pore pressure dissipates and decreases with
 49

time until it reaches a drained stage with no pore pressure gradient. The crack width

increases when the poroelastic material changes from undrained stage at early time to the

drained stage at long time as the pore pressure dissipates and the material around the

crack becomes more “soft”. Mode II loading tends to reduce the crack opening as the

fluid flows into the porous material around the crack and increase the pore pressure

which tends to cause an expansion of the porous material around the crack.

Considering the Mode I and Mode II loading processes for the same crack and

loading as before and poroelastic and fluid parameters listed in Table II-1, the crack

shows a transient opening. If only Mode I loading, the crack width increases with time

and reaches a stable state at long time as in Figure II-12. At short time, the crack opens

as the crack in an elastic material with a Poisson’s ratio the same value as the undrained

Poisson’s ratio in Figure II-13. At long time, it evolves to the drained stage with the

opening as the crack in an elastic material with a Poisson’s ration the same value as the

drained Poisson’s ratio in Figure II-13. If only Mode II loading, the crack closes with

time (Figure II-14) as the fluid flows from the crack into the adjacent formation. The

crack closure approaches its maximum values at infinite time when the pore pressure

around the crack approaches the fluid pressure in the crack. Figure II-15 shows the

fracture closing at 1.91×105 hours, which is smaller than the opening induced by Mode I

loading. The crack still opens with the fluid injection with a constant pressure modeled

by combining Mode I and Mode II loading. The crack has an instant opening, and then

the width reduces with time. But the crack is still open at long time (Figure II-16). The

crack shows the same opening as the analytical solution for the undrained case (Figure
 50

II-17). But crack width reduces with time due to Mode II loading, and approaches a

smaller opening at long time instead of approaching the analytical solution for drained

stage (Figure II-17).

Table II-1. Parameters in the modeling of pressurized crack.

Shear modulus G (psi) 8.6×105

Possoin’s ratio υ 0.16
Undrained Possoin’s ratio υu 0.31
Matrix permeability (md) 0.8
Matrix porosity φ 0.2
Biot’s coefficient α 0.83
Fluid viscosity µ (cp) 1
Fluid compressibility (/psi) 2.9×10-6

5.8E-03

5.6E-03
Crack opening (in)

5.4E-03

5.2E-03

5.0E-03

4.8E-03

4.6E-03
1.E-08 1.E-06 1.E-04 1.E-02 1.E+00 1.E+02 1.E+04 1.E+06

Time (hr)

Figure II-12. Mode I loading: the crack opens as a function of time at x=0.2 in.
 51

6.E-03

5.E-03
Crack opening (in)

4.E-03

3.E-03 Analytical, Undrained

Analytical, Drained
2.E-03 Numerical, Short time
Numerical, Long time
1.E-03

0.E+00
-20 -15 -10 -5 0 5 10 15 20
x (in)

Figure II-13. Comparison of the modeled crack openings at short time and long time with the
analytical solutions for Mode I loading.

0.0E+00

-5.0E-04
Crack closing (in)

-1.0E-03

-1.5E-03

-2.0E-03
1.E-09 1.E-06 1.E-03 1.E+00 1.E+03 1.E+06

Time (hr)

Figure II-14. Mode II loading: the crack closes as a function of time at x=0.2 in.
 52

0.0E+00

-5.0E-04
Crack closing (m)

-1.0E-03

-1.5E-03

-2.0E-03
-20 -15 -10 -5 0 5 10 15 20
x (m)

Figure II-15. The crack closing at t=1.91×105 hrs for Mode II loading.

4.8E-03

4.6E-03
Crack opening (in)

4.4E-03

4.2E-03

4.0E-03

3.8E-03
1.E-09 1.E-06 1.E-03 1.E+00 1.E+03 1.E+06

Time (hr)

Figure II-16. The crack width for Mode I +II loading at x=0.2 in.
 53

6.0E-03

5.0E-03
Crack opening (in)

4.0E-03

3.0E-03
Analytical, Undrained
Analytical, Drained
2.0E-03 Numerical, Short time
Numerical, Long time
1.0E-03

0.0E+00
-20 -15 -10 -5 0 5 10 15 20
x (m)

Figure II-17. Comparison of the modeled crack openings at short time and long time with the
analytical solutions for Mode I+II loading.
 54

2.4 Chapter conclusions

This chapter described the DDM including elastic DDM for nonporous media

and fully coupled poroelastic DDM for porous media saturated with a compressible

single-phase fluid. The fully coupled DDM is based on Biot’s theory of poroelasticity.

For an infinite elastic porous medium containing fractures, if the change of stress and

pore pressure in these fractures are known, the fracture aperture change can be

determined by using the fully coupled DDM. In real situations, neither the change of

stress in fracture nor fracture aperture change is known in the reservoir. But many

investigations have shown that there is a relation between the stress change and the

fracture aperture change in fractures. Chapter III will give the surface characteristics of

fractures with rough surfaces and the relation of stress and fracture deformation. The

pore pressure change in the fractures is not known directly either. Usually only the flow

rate or fluid pressure in the well is known while producing from a fractured reservoir,

the required fluid pressure change in fractures is determined using a numerical finite

difference method (FDM) described in Chapter IV. Finally the fracture aperture change

due to production can be determined by combining the DDM, the constitutive model of

fracture deformation and an FDM to determine the fluid pressure change in fractures.
 55

CHAPTER III

NONLINEAR DEFORMATION OF A SINGLE

ROUGH FRACTURE UNDER STRESS

The fracture is also termed a joint in geology publications. In this dissertation

both fracture and joint describe two contacting rough surfaces with voids that are

completely connected in three-dimensional space. The rough fracture under stress will

deform with the change of stress. There are three types of deformation – normal

deformation, shear deformation and dilation. The deformation for a single rough

fracture has been studied by testing the stress–displacement relationship of natural or

artificially fractures in laboratories (Goodman, 1976, Bandis et al., 1981, Bandis et al.,

1983, Sun et al., 1985, Boulon et al., 1993, Huang et al., 2002, Lee and Cho, 2002). The

constitutive model (Barton-Bandis model) for fracture deformation was presented based

on the experimental results by Bandis et al. (1983) and Barton et al. (1985). The

empirical model only needs some basic fracture characteristic parameters, e.g. the joint

roughness coefficient (JRC), the joint compressive strength (JCS) etc., which can be

measured in laboratory. The fracture deformation usually causes the fracture opening or

closure, and changes the fracture aperture. The “cubic law” which is derived from the

fluid flow between two smooth plates is also applicable to calculate the hydraulic

conductivity or permeability for closed rough fractures with a correction coefficient

(Witherspoon et al., 1980). Barton et al. (1985) presented a method to correlate the

effective hydraulic aperture to the average mechanical aperture and the “cubic law” is
 56

applicable using the correlated effective hydraulic aperture. Consequently, the fracture

permeability change caused by stress change also can be derived and analyzed.

The chapter will start with the fracture surface characteristics in Section 3.1.

Then Section 3.2 will give the relation between normal stress and normal deformation.

Section 3.3 will show the mechanism of shear deformation and dilation, and also the

relation between shear stress and shear displacement. Section 3.4 will give the

definitions for the effective hydraulic aperture and the average mechanical aperture, and

how they are related to permeability. Finally, the conclusions of this chapter will be

given in section 3.5.

3.1 Surface characteristics of a fracture

The fracture deformation depends on the fracture surface characteristics. The

constitutive models need values for surface characteristics, such as JRC, JCS,

unconfined compression strength (rock adjacent to the wall) (σc), residual friction angle

(φr), etc. JRC, JCS and φr are three key parameters in the Barton-Bandis joint model.

Barton and Choubey (1977), and Barton (1982) developed methods to quantify these

parameters for fractures.

3.1.1 Joint compressive strength (JCS)

The measurement of JCS is fundamentally important because it is largely the thin

layers of rock adjacent to joint walls that control the strength and deformation properties

of the rock mass as a whole (Barton and Choubey, 1977). Usually for natural fractures,
 57

JCS is much smaller than the strength of intact rock as the fracture surface is weakened

by weathering (e.g. mechanical disintegration, chemical decomposition). JCS can be

measured by Schmidt Hammer Index test (Barton and Choubey, 1977). Typical JCS

values are listed in Table III-1.

Table III-1. Typical JCS values (ISRM, 1978).

Grade Description Field identification JCS (MPa)

S1 Very soft clay Easily penetrated several inches by fist <0.025
S2 Soft clay Easily penetrated several inches by thumb 0.025-0.05
S3 Firm clay Can be penetrated several inches by thumb 0.05-0.10
with moderate effort
S4 Stiff clay Readily indented by thumb but penetrated 0.10-0.25
only with great effort
S5 Very stiff clay Readily indented by thumbnail 0.25-0.50
S6 Hard clay Indented with difficulty by thumbnail >0.50
R0 Extremely weak Indented by thumbnail 0.25-1.0
rock
R1 Very weak rock Crumbles under firm blows with point of 1.0-5.0
geological hammer, can be peeled by a
pocket knife
R2 Weak rock Can be peeled by a pocket knife with 5.0-25
difficulty, shallow indentations made by
firm blow with point of Geological hammer
R3 Medium strong Cannot be scraped or peeled rock with a 25-50
pocket knife, specimen can be fractured
with single firm blow of geological hammer
R4 Strong rock Specimen requires more than one blow of 50-100
geological hammer to fracture it
R5 Very strong rock Specimen requires many blows of 100-250
geological hammer to fracture
R6 Extremely strong Specimen can only be chipped with >250
rock geological hammer

Note: Grades S1 to S6 apply to cohesive soils, for example clays, silty clays, and combinations
of silts and clays with sand, generally slow draining. Discontinuity wall strength will generally
be characterized by grades R0-R6 (rock) while S1-S6 (day) will generally apply to filled
discontinuities.
 58

3.1.2 Basic friction angle (φb) and residual friction angle (φr)

φb is the friction angle for unweathered fracture and φr is for weathered fracture

angle. The friction angle is defined as arctan (τpeak/σn), where τpeak is the shear stress

required to initiate the fracture to slide under a normal stress σn. The friction angle

between two rough surfaces (unweathered or weathered) can be measured by the tilt test

shown in Figure III-1. The sample is tilted till the upper surface starts to slide. The angle

between the initial sliding surface and the horizontal surface is the friction angle. The

friction angle is an important parameter to predict the shear strength, thereby predicting

the shear displacement, shear dilation, etc. Friction angle values for most unweathered

rocks lie between 25° to 35° and are listed in Table III-2 (Barton and Choubey, 1977).

Under a high level of normal stress the rock beneath the weathered surface comes into

effect and the residual friction angle φr approaches the basic friction angle φb. However,

under a low level of normal stress the residual friction angle φr is much lower than the

basic friction angle φb.

Friction angle

Figure III-1. Tilt test on fractured sample.

 59

Table III-2. Basic friction angles of various unweathered rocks obtained from flat and
residual surfaces (Barton and Choubey, 1977).

Rock type Moisture condition Basic friction Reference*

angle
Sandstone Dry 26--35 Patton, 1966
Sandstone Wet 25--33 Patton, 1966
Sandstone Wet 29 Ripley & Lee,
1962
Sandstone Dry 31--33 Krsmanovid, 1967
Sandstone Dry 32--34 Coulson, 1972
Sedimentary Sandstone Wet 31--34 Coulson, 1972
Rocks Sandstone Wet 33 Richards, 1975
Shale Wet 27 Ripley & Lee,
1962
Siltstone Wet 31 Ripley & Lee,
1962
Siltstone Dry 31--33 Coulson, 1972
Siltstone Wet 27--31 Coulson, 1972
Conglomerate Dry 35 Krsmanovid, 1967
Chalk Wet 30 Hutchinson, 1972
Limestone Dry 31--37 Coulson, 1972
Limestone Wet 27--35 Coulson, 1972
Basalt Dry 35--38 Coulson, 1972
Basalt Wet 31--36 Coulson, 1972
Fine-grained Dry 31--35 Coulson, 1972
granite
Fine-grained Wet 29--31 Coulson, 1972
Igneous granite
Rocks Coarse-grained Dry 31--35 Coulson, 1972
granite
Coarse-grained Wet 31--33 Coulson, 1972
granite
Porphyry Dry 31 Barton, 1971b
Porphyry Wet 31 Barton, 1971b
Dolerite Dry 36 Richards, 1975
Dolerite Wet 32 Richards, 1975
Amphibolite Dry 32 Wallace et al.,
1970
Gneiss Dry 26--29 Coulson, 1972
Metamorphic Gneiss Wet 23--26 Coulson, 1972
Rocks Slate Dry 25--30 Barton, 1971b
Slate Dry 30 Richards, 1975
Slate Wet 21 Richards, 1975
* Refer to Barton and Choubey (1977) for specific references.
 60

3.1.3 Joint roughness coefficient (JRC)

In general the joint surface roughness can be characterized by waviness (large

scale undulations which, if interlocked and in contact, cause dilation during shear

displacement since they are too large to be sheared off) and by unevenness (small scale

roughness that tends to be damaged during shear displacement unless the discontinuity

wails are of high strength or the stress levels are low, so that dilation can also occur on

these small scale features) (IRSM, 1978). Barton and Choubey presented a method to

describe the JRC and also presented a formula (Eq.3-1) to calculate the peak shear

strength τpeak according to the JRC index.

  JCS  
τ peak = σ n' tan  JRC log10  
' 
+ φ  (3-1)
 σn 
r
 

where σn′ is the effective normal stress and φr is the friction angle for weathered fracture.

The JRC index can be measured by a tilt test or estimated by comparing with the profiles

measured on other joints shown in Figure III-2 (Barton and Choubey, 1977).
 61

Figure III-2. Typical JRC values for joint samples of different roughness (Barton and Choubey,
1977).
 62

3.2 Normal deformation

The two rough surfaces of a fracture are weaker and more deformable than intact

rock. The normal deformation of the two rough surfaces in response to the normal stress

change across the fracture or fluid pressure change in the void space of the fracture has a

direct important influence on the fracture aperture and fracture permeability. The

normal deformation of a fracture can be characterized by the relationship between the

effective stress across the fracture and the fracture closure (the change of the average

aperture of the fracture).

Goodman (1976) measured the fracture closure as a function of normal stress on

artificially induced tensile fractures in rock cores. He measured the axial displacement

of an intact rock core under axial stress and axial displacement of a rock core of the

same size and an artificially induced tensile fracture perpendicular to the axis under the

same axial stress. The difference of the two displacements is the fracture closure.

Fracture closure measurements were made for both mated fractures, for which the two

surfaces of fracture were placed the same relative positions that they occupied before

fracturing the core, and non-mated fractures, for which the two surfaces of fracture were

rotated from their original positions relative to one another (Figure III-3). The stress-

closure curves show high non-linearity, and the non-mated fracture has greater closure.
 63

Figure III-3. Measurements of the closure under normal stress of an artificially-induced tensile
fracture in a rock core (Goodman, 1976).

Bandis et al. (1983) have measured closure curves for a fracture under normal

stress for a variety of natural and unfilled fractures with different degrees of weathering

and roughness in slate, dolerite, limestone, siltstone and sandstone (Figure III-4 and

Figure III-5). They used the same method as Goodman used to determine fracture

closure for natural fractures. As expected, the fracture closures for weathered fractures

(Figure III-5) were much greater than for fresh fractures (Figure III-4) under the same

stress condition. With the increase of normal stress (σn), the stress–closure curves

became gradually steeper and developed into virtually straight lines where the fractures

have reached their fully closed state. There was permanent deformation observed during

the loading–unloading cycle. Therefore the deformation characteristics of fractures also

depend on the stress history of the fractures.

 64

Figure III-4. Normal stress (σn) vs closure curves for a range of fresh fractures in different rock
types, under repeated loading cycles (Bandis et al., 1983).
 65

Figure III-5. Normal stress (σn) vs closure curves for a range of weathered fractures in different
rock types under repeated loading cycles (Bandis et al., 1983).
 66

Based on the experimental results Bandis et al. (1983) presented a hyperbolic

function (Eq. (3-2)) to represent the normal stress–closure relationship.

Dn
σn = (3-2)
aa − b Dn

where Dn is the fracture closure, aa and b are constants. Eq. (3-2) was rearranged into a

linear form:

Dn
= aa − b Dn (3-3)
σn

aa and b can be obtained by using Eq. (3-3) to fit the measured normal stress–closure

data, and Figure III-6) shows that Eq.(3-3) fits well with measured data. When σn

approaches infinity, the fracture closure approaches the maximum fracture closure Dnmax

and Dnmax is equal to aa/b according to Eq. (3-3). For extremely small normal stress

(σn→0), the fracture closure will be small (Dn →0), and hence the initial normal fracture

stiffness for σn→0 is defined:

σn 1
K ni = σ n →0 = (3-4)
Dn aa

Therefore Eq. (3-2) can be rewritten by substituting the two parameters initial normal

fracture stiffness (Kni) and maximum fracture closure (Dnmax) for aa and b:

K ni Dn
σn = (3-5)
1 − Dn / Dn max
 67

The normal stiffness (Kn) is then derived from Eq. (3-5) as a function of Dn or σn:

∂σ n K ni
Kn = = (3-6)
∂Dn (1 − Dn / Dn max )2

or

∂σ n K ni
Kn = = (3-7)
∂Dn [1 − σ n / (K ni Dn max + σ n )]2

Bandis et al. (1983) also derived the empirical formulae for Dnmax (Eq. (3-8)) and Kni (Eq.

(3-9)) in terms of JCS, JRC index and average fracture aperture (wf):

D
 JCS 
Dn max = A1 + B1( JRC ) + C1  (3-8)
 w 
 f 

JCS
K ni = A2 + B 2 ( JRC ) + C 2 (3-9)
wf

where A1, A2, B1, B2, C1, C2 and D are coefficients determined by fitting experimental

data.
 68

Figure III-6. Linear plots of Dn/σn vs Dn for different fracture types, indicating good hyperbolic fit
irrespective of the stress history and the loading mode (Bandis et al. 1983).
 69

3.3 Shear deformation and dilation

For a fracture under normal stress loading, the fracture will have a shear

deformation if the shear stress (τ) is less than the peak shear strength (τpeak) and become

instable and have a fast movement if τ exceeds τpeak. However, for rough surfaces, the

shear dilation caused by shear displacement may prevent the instability. The typical

shear stress–shear displacement curves have three stages, pre-peak, peak, and post-peak

(Figure III-7).

The peak shear strength is a critical parameter to predict the stability of fractures,

faults or the initiation of nonlinear movement under anisotropic stress condition. Barton

(1976) presented a formula (Eq. (3-1)) to predict the peak shear strength τpeak according

to the effective normal stress, the fracture surface roughness JRC, compression wall

strength JCS and residual friction angle φr based on large body of laboratory measured

results under low effective normal stress (σn′<10MPa). But the peak shear strength at

high effective normal stress is independent of JRC, JCS, φr and even the rock type, and

is only dependent on the effective normal stress. Byerlee (1978) developed empirical

formulae (Eqs. (3-10) and (3-11)) based on large body of experimental data on rocks

including sandstone, limestone, granite, gabbro, etc.

τ peak = 0.85σ n' σ n' < 200MPa (3-10)

τ peak = 50 + 0.6σ n' 200MPa < σ n' < 2000MPa (3-11)

 70

According to the shear stress–shear displacement curves (Figure III-7), the pre-

peak curve can be approximated as a line. The slope of the line is the pre-peak shear

stiffness Ks:

τ peak
Ks = (3-12)
Ds − peak

where Ds-peak is the shear displacement when the shear stress reaches the peak value.

The post-peak curve is very complicated and is often treated as a zero slope line, and the

shear stiffness Ks is assumed as zero.

When shearing of two rough surfaces occurs, the opposed asperities slide over

each other and cause an increase in aperture. The increase of fracture aperture induced

by shear deformation was well investigated in laboratory by Bandis et al. (1981). Figure

III-8a shows the shear stress–displacement curves for different block size and Figure

III-8b shows the corresponding aperture increase induced by the shear displacement at

constant normal stresses. The dashed lines show the dilation angle, which is defined as:

 Dn 
φd = tan −1   (3-13)
 Ds 
 71

Figure III-7. Shear stress – shear displacement for joints with different normal stress and JRC
(Barton et al., 1985).
 72

Figure III-8. Cumulative mean shear stress---shear displacement (a) and dilation (b) curves (Bandis
et al., 1981).
 73

3.4 Fracture aperture and permeability

Fracture aperture is the perpendicular distance between adjacent rock walls of a

fracture. The fracture deformation will change the fracture aperture, thereby changing

the fracture permeability. The relation of permeability and aperture for laminar flow

through a pair of smooth parallel plates has been investigated and the cubic law was

derived (Snow, 1965; Iwai, 1976). The flow rate through the fracture (Figure III-9) is:

w3f dp
q=− (3-14)
12µ dx

Compared with Darcy’s law, the fracture permeability is:

w2f
kf = (3-15)
12

x wf

Figure III-9. Laminar flow through a pair of smooth parallel plates.

The natural fracture is not completely open, and the surfaces are not smooth.

Therefore, Eq. (3-15) cannot be applied to the natural fracture directly. However,

Witherspoon et al. (1980) found that the cubic law was still valid for partially closed

fractures by laboratory investigations. The investigated fracture aperture ranges from

4µm to 250µm and the rock types include basalt, granite and marble. The fracture
 74

conductivity still has a cubic relation with the average fracture aperture. But Eq. (3-15)

requires a correction coefficient f to be valid for partially closed fracture.

w2f
kf = (3-16)
12 f

The correction coefficient in their investigation varied from 1.04 to 1.65.

Barton et al. (1985) argued that Witherspoon et al. (1980) did not measure the

real mechanical aperture, and that the aperture they used was an approximate hydraulic

aperture. Barton et al. (1985) still used Eq. (3-15) to relate fracture permeability to

aperture, but substituted effective hydraulic fracture aperture for mechanical aperture.

Based on published experimental data (Figure III-10), they developed an empirical

formula to relate the hydraulic fracture aperture to mechanical aperture:

JRC 2.5
wef =
(w f / wef )2
(3-17)

The unit of wef and wf is µm.

 75

Figure III-10. Comparison of mechanical aperture and hydraulic aperture (Barton et al., 1985;
Olson and Barton, 2001).

3.5 Chapter conclusions

This chapter described the characteristics of fracture surfaces, nonlinear Barton-

Bandis model of fracture deformation, and the relation of fracture permeability to

fracture aperture in rough fractures. In the nonlinear Barton-Bandis model of fracture

deformation, the relation of normal stress and fracture closure is represented by a

hyperbolic formula. The relation of shear stress and shear displacement is linear before

yielding and too complicated to represent using simple functions after yielding. The

model also includes shear dilation which is the fracture opening caused by shear
 76

displacement. The peak shear strength can be determined from the effective normal

stress, JRC, JCS and friction angle. The fracture permeability has a cubic relation to the

effective hydraulic aperture but not the average mechanical aperture. The effective

hydraulic aperture is related with the average mechanical aperture using JRC.

The next chapter will combine the DDM, the nonlinear Barton-Bandis model of fracture

deformation, and an FDM to determine the pore pressure change in fractures and in turn

to determine the change of fracture aperture and permeability due to production from a

fractured reservoir.
 77

CHAPTER IV

MODELING OF THE FRACTURE APERTURE AND

PERMEABILITY CHANGE IN FRACTURED RESERVOIRS

Throughout this study, the fractured reservoir is treated as a fracture network in a

porous medium saturated with a compressible single-phase fluid. As in dual porosity

models, the fracture network provides the main flow channels and the porous media

provides the main storage media. On production, the fluid flows from matrix to

fractures, then in fractures to the well. The fluid pressure change induces effective stress

change and fracture aperture change, which in turn causes permeability changes in the

fractures, the nature of which was addressed in the Chapter III. The fracture

permeability change in turn influences fluid flow. Fluid flow in the fracture network is

solved using a finite difference method (FDM). The change of effective stress on the

fractures induces fracture deformation including normal and shear deformation. The

fracture deformation also disturbs the stress distribution in the fracture network. A new

numerical method is developed in this chapter to determine the fluid pressure, fracture

aperture change and stress change implicitly using an FDM to solve the diffusion

equation for fluid flow in fractures, a fully coupled displacement discontinuity method

(DDM) to determine the global fracture deformation relation, and the nonlinear Barton-

Bandis fracture deformation model to determine the local fracture deformation relation.

This chapter will start with building and discretizing the equation for fluid flow

in fracture network in Section 4.1. And then Section 4.2 will describe a method for
 78

combining the global and local relations between stress and displacement to fracture

deformation. Section 4.3 will present a new numerical method combining an FDM for

the diffusivity equation governing fluid flow in fractures, a fully coupled DDM for

determining the global fracture deformations, and a nonlinear fracture deformation

model for determining the local fracture deformations. In addition to the fully coupled

method, an uncoupled method will be presented that saves computation time in cases

where the effect of solid deformation on fluid flow is small. Finally, Section 4.4 will

give conclusions of this chapter.

4.1 Fluid flow in the fracture network

The apertures of real fractures vary in space (Figure IV-1) and the fluid flow

inside is very complicated due to the rough surfaces. But Witherspoon et al. (1980)

verified that Darcy’s law is still valid and the rough fracture can be represented by a

fracture with an average fracture aperture, as in Figure IV-2. The one dimensional fluid

material balance equation in the fracture including flow from the connected fractures and

the interface flow from the connected matrices is given by

∂ (ρ f q f ) ∂ (ρ f n w f ∆L )
=− − ∆Lρ f qint − ρ f qs (4-1)
∂x ∂t

where ρf is the fluid density; qf is the flow rate in the fracture per unit formation

thickness; qint is the interface flow rate per fracture length per unit formation thickness;

∆L (given previously as 2a for the well fracture) is the length of fracture segment; qs is

the production rate per unit formation thickness; n is the ratio of actual fracture void
 79

volume (Vf) to the effective fracture void volume for fluid flow (Vef). In Eq. (4-1), the

∂ (ρ f q f )
left term, , is the net mass flow rate out of the fracture, the first right term,
∂x

∂ (ρ f n w f ∆L )
, is rate of fluid mass change in the fracture, the second right term,
∂t

∆Lρ f qint , is the mass flow rate between fracture and the connected matrix, and the third

right term, ρ f qs , is a production term, for example for a producing well. The flow

rate in the fracture can be obtained by using Darcy’s law:

qint

qin qs qout

qint

Figure IV-1. Fluid flow through a rough fracture.

k f w f ∂p
qf = − (4-2)
µ ∂x

where kf is the fracture permeability determined from the fracture aperture (3-15).

Combining Eqs. (2-23) and (4-2), the net fracture flow rate term becomes:

∂ (ρ f q f ) ρ f k f w f ∂ 2 p co ρ f k f w f  ∂p  2
=− −   (4-3)
∂x µ ∂x 2 µ  ∂x 
 80

qint

qin qs qout

qint

Figure IV-2. Fluid flow through an artificial fracture represented using average fracture aperture.

The second term with squared pressure gradient multiplied by the small compressibility

can be neglected (Lee et al., 2003), and the net fracture flow rate is approximated as:

∂ (ρ f q f ) ρ f k f wf ∂2 p
=− (4-4)
∂x µ ∂x 2

The fluid mass change in the fracture includes two parts, one is due to fracture volume

change and another one is due to fluid density change. The fracture volume change is

mainly from the fracture aperture change:

∂V f ∂w f
= n ∆L (4-5)
∂t ∂t

The fracture aperture change can be related with the fracture closure Dn:

∂w f ∂Dn
=− (4-6)
∂t ∂t

Eq. (4-5) can be rewritten by substituting Dn for wf:

∂V f ∂Dn
= −n ∆L (4-7)
∂t ∂t

The fluid mass change due to fluid density change is:

 81

∂m ∂ρ
= − n w f ∆L f (4-8)
∂t ∂t

Substituting Eq. (2-23) into Eq. (4-8) yields:

∂m ∂p
= −n co ρ f w f ∆L (4-9)
∂t ∂t

Combining Eqs. (4-1) – (4-9) yields the pressure diffusion equation:

k f wf ∂2 p ∂p ∂D
= n w f ∆Lco − n∆L n + ∆Lqint + qs (4-10)
µ ∂x 2
∂t ∂t

k f wf ∂2 p
In Eq. (4-10), the left term, , is the net flow rate in the fracture, the first right
µ ∂x 2

∂p
term, n w f ∆Lco , is the fluid volume change due to fluid compression or expansion
∂t

∂Dn
(fluid density change), the second right term, n∆L , is the fluid volume change due
∂t

to fracture deformation, the third right term, ∆Lqint , is the interface flow rate per

formation thickness between fracture and the matrix, and the last term, qs , is the

production rate per unit formation thickness.

4.2. Mechanical coupling of fracture deformation

In a fracture network, the change of stress and fracture deformation for any

fracture obeys the constitutive relations for fracture deformation. There is a local relation

for each fracture or fracture segment between its stress and deformation, and there are

global relations for stress and fracture deformation among fractures in the fracture

network.
 82

4.2.1 Local relation between stress and displacement to fracture deformation

For any fracture in the fracture network (Figure IV-3), the deformation must

comply with the fracture deformation model. The relation between effective normal

stress change ∆σn′ and normal displacement ∆Dn of the ith fracture segment is:

i i i
∆ σ n' = − K n ∆ Dn (4-11)

The normal stiffness Kn is a coefficient which is dependent on the fracture closure (Eq.

(3-6)) or stress (Eq. (3-7)). The effective stress (tension is treated as positive) is defined

as:

σ n' = σ n + α p (4-12)

where α = 1 − K m / K s as before in Chapter II. For a fracture, when the bulk modulus of

system Km is much less than the solid bulk modulus Ks, the Biot coefficient becomes

unity, and the effective stress is given by:

σ n' = σ n + p (4-13)

Substituting Eq. (4-13) for effective stress in Eq. (4-11) yields (for each fracture

segment):

i i i
∆ σ n + ∆ p = −Kn ∆ Dn (4-14)

The relation of shear stress change ∆σs and shear displacement ∆Ds is:

i i i
∆ σ s = K s ∆ Ds (4-15)

The shear stiffness is a constant before yielding and reduces to zero after yielding. The

normal deformation ∆Dn-dilation due to shear dilation is:

 83

i i
∆ D n − dilation = −∆ D s tan φd (4-16)

The dilation angle is defined in Eq. (3-13). Eq. (4-14) must be rewritten when the

normal deformation induced by shear dilation is considered:

i i i
 i i

∆ σ n + ∆ p = − K n  ∆ D n + ∆ D s tan φd  (4-17)
 

σs
σn′

Figure IV-3. Local relation of fracture deformation.

4.2.2 Global relation between stress and displacement to fracture deformation

In the fracture network with m fracture segments, there are interactions among

fractures. The stress change of the ith fracture segment is influenced by the deformation

of all the fracture segments in the system. For the elastic DDM (Eq. (2-11)), the change

of normal and shear stresses of the ith fracture segment is related with the normal and

shear deformation of all the fracture segments as:

 84

i m ij j m ij j
∆ σ n = ∑ A ∆ Dn + ∑ B ∆ Ds
j =1 j =1

(4-18)
i m ij j m ij j
∆ σ s = ∑ E ∆ Dn + ∑ F ∆ Ds
j =1 j =1

For the poroelastic DDM, the interface flow rate between fracture and matrix also has an

impact on the stress change. Therefore, the change of normal and shear stresses of ith

fracture segment depends on the interface flow rate in addition to the normal and shear

deformation of all fracture segments according to Eq. (2-40).

i m ij j m ij j m ij j
∆ σ n = ∑ A ∆ D n + ∑ B ∆ D s + ∑ C q int
j =1 j =1 j =1

(4-19)
i m ij j m ij j m ij j
∆ σ s = ∑ E ∆ D n + ∑ F ∆ D s + ∑ K q int
j =1 j =1 j =1

The change of fluid pressure of ith fracture also depends on the interface flow rate,

normal and shear deformation of all fracture segments according to Eq. (2-40).

i m ij j m ij j m ij j
∆ p = ∑ L ∆ D n + ∑ H ∆ D s + ∑ N q int (4-20)
j =1 j =1 j =1

4.3 Uncoupled and coupled solution methods

The changes in fracture apertures due to production can be determined by solving

the pressure diffusion equation in fracture network, the fracture deformation model for

local stress—displacement relations and the DDM for global stress–displacement

relations. The result can be determined using an uncoupled method or a coupled

solution method. The uncoupled method saves computation time and provides a suitable
 85

approximation when the effect of solid deformation on fluid flow is small. The

uncoupled method first solves for the fluid pressure change from the diffusivity equation,

and then uses the resulted fluid pressure change as a boundary condition to determine the

fracture deformation by combining the constitutive equations for fracture deformation

(Eqs.(4-15) and (4-17)) and stress–displacement relations from the elastic DDM (Eq. (4-

18)). The coupled method simultaneously obtains the fluid pressure change, interface

flow rate, fracture deformation by solving together the diffusivity equation (Eq. (4-10)),

constitutive equations for fracture deformation (Eqs. (4-15) and (4-17)), and stress–

displacement relations from the poroelastic DDM (Eqs. (4-19) and (4-20)).

4.3.1 Uncoupled method

The change of normal fracture closure is related with the pore pressure change

according to Eq. (4-14) by defining a fracture compressibility parameter cfr.

∂Dn ∂p
= − w f c fr (4-21)
∂t ∂t

Substituting Eq. (4-21) into Eq. (4-10) yields:

k f wf ∂2 p ∂p
= ct n w f ∆L + ∆Lqint + qs (4-22)
µ ∂x 2
∂t

where ct=co+cfr is the total compressibility. Eq. (4-22) can be discretized for a given

fracture network using an implicit finite difference method.

m ij j i i i
∑ C p pl +1 = C pc pl − ∆L qint − qs
j =1
(4-23)
 86

where m is total fracture elements, Cp is coefficient matrix, the subscript l+1 indicates

i
new time level and the subscript l indicates the old time level, and qs is the production

from the ith fracture element. The interface flow rate qint is an unknown and can be

determined using an iterative method. For every time step, the interface flow rate qint is

assumed as zero for the first iteration step. Then Eq. (4-23) can be solved to obtain the

fluid pressure distribution in the fracture network. The new fluid pressure in fractures

can be taken as the boundary conditions for every matrix element (Figure IV-4) and the

fluid flow between the matrix and fractures around it can be obtained by finite difference

solution of the uncoupled diffusivity equation (Eq. (4-24)) in the matrix. After the

pressure distribution in the matrix is determined, the flow rate at the boundary between

the matrix element and surrounding fractures can be obtained from Darcy’s law

 ∂ 2 p ∂ 2 p  φµcmt ∂p
 2 + 2  = (4-24)
 ∂x ∂y  k ∂t

i1
where cmt is the total compressibility of fluid and matrix. Then interface flow rate q int is

i2
used to solve Eq. (4-23) in the second iteration. A new interface flow rate q int can be

obtained as for the second iteration, and this process is repeated until the difference

between successive interface flow rate values is smaller than the accuracy needed to the

problem. At that point, the iteration terminates and the calculation begins for the next

time step. The pressure distribution at the last iteration is taken as the result for that time

step.
 87

Matrix
Element Fracture
element

Figure IV-4. Interface flow rate between fracture and matrix.

The effective stress change in a fracture resulting from changes in pore pressure

and the total stress is illustrated in Figure IV-5. The effect is that of a set of springs

between two plates, and the stress acting on the springs represents the effective stress.

The compression effective stress (-∆σn′) increases with the decrease of pore pressure (∆p)

and the increase of compression total stress (-∆σn). After the pressure change, ∆p, is

solved for every time step, a set of linear equations for effective stress change is

obtained by combining Eqs. (4-15), (4-17) and (4-18).

 88

m ij j m ij j i i i i i
∑ A ∆ D n + ∑ B ∆ D s + K n ∆ D n + K n tan φd ∆ D s = −∆ p
j =1 j =1

(4-25)
m ij j m ij j i i
∑E∆D +∑F ∆D
j =1
n
j =1
s − K s ∆ Ds = 0

The normal and shear displacement for every time step can be obtained by solving the

linear equation (4-25).

∆σn

∆p

Figure IV-5. Illustration of effective stress change on fracture.

After solving the fracture displacement, the fracture aperture and permeability

are updated.

i i i
wlf+1 = wlf − ∆ D n (4-26)

The fracture aperture at the new time step is determined by subtracting the fracture

closure determined from the previous time step. Then the fracture permeability is

updated according to Eq. (3-15). If the difference between mechanical hydraulic

aperture is to be considered, Eq. (3-17) is used to convert the mechanical aperture into

the hydraulic aperture to update the fracture permeability and diffusivity equation is

modified to use the hydraulic aperture, wef, instead of the mechanical aperture, wf, used

in Eq. (4-10).
 89

k f wef ∂ 2 p ∂p ∂D
= n w f ∆Lc f − n∆L n + ∆Lqint + qs (4-27)
µ ∂x 2
∂t ∂t

4.3.2 Coupled method

The fluid pressure change induces fracture deformation and the fracture

deformation also influences the fluid pressure distribution. In the coupled method, the

equations for fluid pressure, interface flow rate, and normal and shear fracture

displacement are solved simultaneously.

The poroelastic DDM solutions are both space and time dependent, and the

fundamental solutions are based on constant displacement discontinuities and constant

interface or source flow rates. However, for practical applications, the displacement

discontinuities and interface flow rates in Eqs. (4-19) and (4-20) are time dependent.

The time marching scheme shown in Figure II-9 is used to allow source strengths (the

displacement discontinuities and interface flow rate) to change with time. Starting each

boundary integration from an initial homogeneous status avoids the need for volumetric

integration (Carvalho, 1990). Therefore, all the previous increments of source strengths

must be included while numerically integrating the effect of source strengths at each

time step. According to Eq. (2-44), the induced stress and pore pressure on the ith

fracture segment by the increments of source strengths are:

 90

jξ jξ jξ
∆ σ n (t ) = ∑ A(t − τ ξ )∆ D n + ∑ B (t − τ ξ )∆ D s + ∑ C (t − τ ξ )∆ q int
i m ij m ij m ij

j =1 j =1 j =1
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh
+∑
h=0
∑ A(t − τ h )∆ D n + ∑
j =1 h=0
∑ B(t − τ h )∆ D s + ∑
j =1 h=0
∑ C (t − τ h )∆ q int
j =1

jξ jξ jξ
∆ σ s (t ) = ∑ E (t − τ ξ )∆ D n + ∑ F (t − τ ξ )∆ D s + ∑ K (t − τ ξ )∆ q int
i m ij m ij m ij

j =1 j =1 j =1
ξ −1 m ξ −1 m ξ −1 m
(4-28)
ij jh ij jh ij jh
+∑
h=0
∑ E (t − τ )∆ D + ∑ ∑ F (t − τ )∆ D + ∑ ∑ K (t − τ )∆ q
j =1
h n
h=0 j =1
h s
h=0 j =1
h int

jξ jξ jξ
∆ p(t ) = ∑ L(t − τ ξ )∆ D n + ∑ H (t − τ ξ )∆ D s + ∑ N (t − τ ξ )∆ q int
i m ij m ij m ij

j =1 j =1 j =1
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh
+∑
h=0
∑ L(t − τ h )∆ D n + ∑
j =1 h=0
∑ H (t − τ h )∆ D s + ∑
j =1 h =0
∑ N (t − τ h )∆ q int
j =1

jξ jξ jξ
where ∆ D n , ∆ D s and ∆ q int are the source strength increments for the jth fracture

jh jh jh
segment at the current time step, ξ; ∆ D n , ∆ D s and ∆ q int are the previous source

strength increments of for the jth fracture segment at time step h, which indexed from 1

ij ij ij ij ij ij ij
to ξ-1. A(t − τ h ) , B(t − τ h ) , C (t − τ h ) , E (t − τ h ) , F (t − τ h ) , K (t − τ h ) , L(t − τ h ) ,

ij ij
H (t − τ h ) , and N (t − τ h ) are the influence coefficients of jth fracture element on the ith

fracture element at time step h as defined in Eq. (2-38).

Using the same time discretization, the effective normal stress change (Eq. (4-17))

and shear stress change (Eq. (4-15)) in the ith fracture segment can be rewritten as:
 91

 iξ iξ
 i  ξ −1 ih ξ −1

∆ σ n (t ) + p(t ) − p 0 = − K n  ∆ D n + ∆ D s tan φd  − K n  ∑ ∆ D n + tan φd ∑ ∆ D s
i i i i ih

   h=0 h=0 
(4-29)
iξ ξ −1
∆ σ s (t ) = K s ∆ D s + K s
i i i ih
∑∆ D
h=0
s

where p(t ) is the fluid pressure in the ith fracture segment at time t and p 0 is the initial
i i

fluid pressure in the ith fracture segment. Substituting Eq. (4-29) into Eq. (4-28), and

substituting p(t ) − p 0 for ∆ p(t ) in Eq. (4-28) yield:

i i i

jξ iξ jξ iξ
p(t ) + ∑ A(t − τ ξ )∆ D n + K n ∆ D n + ∑ B (t − τ ξ )∆ D s + K n tan φd ∆ D s
i m ij i m ij i

j =1 j =1

jξ ξ −1 ξ −1
+ ∑ C (t − τ ξ ) q int = −∑
m ij m ij jh m ij jh

j =1 h=0
∑ A(t − τ )∆ D − ∑ ∑ B(t − τ )∆ D
j =1
h n
h=0 j =1
h s

ξ −1 i  ξ −1 ξ −1
mij jh
 i
( )
ih ih
−∑ ∑ C t − τ h q int − K n
∑
 ∆ D n + tan φd ∑ ∆ Ds  + p 0
h=0 j =1  h =0 h =0 

jξ jξ iξ jξ

∑ E (t − τ ξ )∆ D + ∑ F (t − τ ξ )∆ D − K s ∆ D s + ∑ K (t − τ ξ ) q int =
m ij m ij i m ij
n s
j =1 j =1 j =1
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh
−∑
h=0
∑ E (t − τ h )∆ D n − ∑
j =1 h =0
∑ F (t − τ h )∆ D s − ∑
j =1 h =0
∑ K (t − τ h ) q int
j =1
(4-30)

i ξ −1 ih
+ K s ∑ ∆ Ds
h =0

jξ jξ jξ
− p(t ) + ∑ L(t − τ ξ )∆ D n + ∑ H (t − τ ξ )∆ D s + ∑ N (t − τ ξ ) q int =
i m ij m ij m ij

j =1 j =1 j =1
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh

∑ L(t − τ h )∆ D n − ∑ ∑ H (t − τ h )∆ D s − ∑ ∑ N (t − τ h ) q int
i
− p0 − ∑
h =0 j =1 h=0 j =1 h =0 j =1
 92

The diffusivity equation (4-10) is discretized in space and time for a given

fracture network using an implicit finite difference method like that given in Appendix C

for a regular fracture network. For the ith fracture segment at the time step, ξ,

m ij j iξ iξ i ξ −1 ih ξ ih

∑ C p p(t ) −n∆L∆ D n + ∆L∆ qint = n w f ∆Lc f p(τ ξ ) − ∆L∑ ∆ qint − ∑ q s

j =1 h=0 h=0
(4-31)

ij
where C p is the fluid coefficient matrix. The production rate from ith fracture segment

ih
q s is also discretized in time in Eq. (4-31). All left terms in Eqs. (4-30) and (4-31) are

unknown and all right terms are known. Appendix D gives an example matrix for the set

of linear equations built from Eqs. (4-30) and (4-31) for a given fracture network. When

the production rate and initial reservoir pressure are given, the normal and shear fracture

displacement, interface flow rate, and fluid pressure can be obtained by solving the

linear equation Eqs. (4-30) and (4-31). Unlike for the uncoupled method, the interface
 93

flow rate is solved implicitly, and there is no need for the FDM determination of the

interface flow rate. The treatment for fracture permeability is the same as that for the

uncoupled method.

4.4 Chapter conclusions

This chapter presented a new numerical method to solve the fluid pressure,

fracture aperture change and stress change simultaneously by combining a finite

difference method (FDM) solution for the diffusivity equation for fluid flow in fractures,

a fully coupled displacement discontinuity method (DDM) for the global relation of

fracture deformation, and the Barton-Bandis fracture deformation model for the local

relation of fracture deformation. The fracture permeability changes with the fracture

aperture change. Applications of this method are shown in the next chapter.
 94

CHAPTER V

MODEL APPLICATIONS

The coupled method described in Chapter IV applies when a single phase fluid is

produced from a naturally fractured reservoir. Pressure decrease causes effective stress

change, thereby inducing fracture aperture and permeability change in the natural

fractures. The coupled method is applied to quantitatively predict the fracture aperture

and permeability change during production under different in situ stress conditions for

rock and fracture parameters that can be measured in laboratories and/or from

production data.

This chapter will illustrate that under isotropic stress conditions the effective

stress increases with reservoir pressure drop, and fracture aperture and permeability

decrease with time. Further we will show that under highly anisotropic stress conditions,

fracture aperture and permeability in some fractures may not decrease, or may even

increase.

This chapter will start with applications under isotropic in situ stress conditions

in Section 5.1. Next will be applications under high anisotropic in situ stress conditions

in Section 5.2. Finally chapter conclusions are in Section 5.3.

5.1 Fracture aperture and permeability change under isotropic conditions

In this section the coupled solution method is applied to a case under isotropic

stress conditions. The results of reservoir pressure change, stress change, fracture
 95

aperture and permeability change are shown, and the interactions of these changes are

discussed. The influences of input rock and fracture properties on the results are also

investigated.

5.1.1 Parameters and assumptions

In this section the fracture permeability change during production and its effect

on transient wellbore pressure are investigated for a well with constant production rate

(12.6 Res bbl/day) from a unit reservoir thickness of 3.28 ft (1 m) in a formation with a

fracture network consisting of two sets of orthogonal vertical fractures surrounded by an

effectively infinite porous medium as in Figure V-1. The fracture permeability is

calculated from the mechanical aperture using the cubic law for the ratio of hydraulic

aperture to the mechanical aperture (wef/wf) assumed to be 1. (Cases for other ratios will

be discussed later). Only two-dimensional flow and deformation are considered, and

change in the vertical direction is ignored. The in situ stress field before production is

assumed to be isotropic with compression set to 3045 psi. To better illustrate the

geomechanic effects during production, the reservoir pressure is set very close to the in

situ stress at 2900 psi. The two joint parameters, initial normal stiffness and maximum

closure, characterizing the normal deformation of fracture are 2.21×104 psi/ft and 0.0315

in, respectively. The nonlinear relationship between effective normal stress under

compression and fracture closure is shown in Figure V-2. The fracture aperture at the

initial condition (zero effective normal stress) is assumed as 0.0315 in. The fracture

aperture under the initial in situ stress before production is assumed as 0.009 in for all
 96

fractures. Other parameters are listed in Table V-1. Because the fracture permeability

dominates the reservoir permeability, changes in matrix permeability are neglected and

assumed as constant during production.

Matrix rock

Matrix rock Matrix

Matrixrock
rock

Matrix rock

Figure V-1. Well located at the center of a fractured field, which is surrounded by matrix rock of
effectively infinite extent.
 97

1500

Effective normal stress (psi)

1200

Maximum fracture closure

900

600

300

0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035

Fracture closure (in)

Figure V-2. Nonlinear fracture normal deformation.

Table V-1. Rock and fracture parameters in the modeling.

Area (ft2) 3281×3281

Shear modulus G (psi) 8.555×105
Possoin’s ratio υ 0.16
Undrained Possoin’s ratio υu 0.31
Matrix permeability (md) 0.8
Matrix porosity φ 0.2
Biot’s coefficient α 0.83
Fluid viscosity µ (cp) 1
Fluid compressibility (psi-1) 4.69×10-6
Ratio of actual fracture volume to the 10
effective fracture volume n
Fracture spacing S (ft) 310
 98

5.1.2 Results for isotropic stress conditions

Figure V-3 shows the reservoir pressure distribution after 360 days on production.

In this case, the lowest pressure is 2635 psi, and the highest pressure is 2672 psi. The

fracture aperture declines with production as in Figure V-4. The fracture intersected

with the well has the maximum fracture closure with the aperture changing from

9.02×10-3 in to 4.82×10-3 in. The aperture of a fracture on the boundary shows the

minimum fracture closure change from 9.02×10-3 in to 5.5×10-3 in. The effective normal

stress increases with time. The change of effective normal stress and fracture aperture

for the facture intersected with well and for a boundary fracture with minimum change

are shown in Figure V-5. Figure V-6 shows that the fracture permeability calculated

from the fracture aperture using the cubic law has the same trend as the aperture change,

and changes from 4428 darcy to 1266 darcy at the well and from 4428 darcy to 1645

darcy at the boundary. The pressure in the fracture intersected with the well is assumed

as the bottomhole pressure. Figure V-7 compares the bottomhole pressure versus time

behavior for the stress-dependent fracture permeability to that for the fixed fracture

permeability case. At early time stage while most of the fluid production from the

fracture network is mainly driven by the contraction of fracture volume and fluid

expansion, both the pressure drop and pressure derivative show a unit slope trend. At

the medium stage, the pressure derivative for the fixed fracture permeability case shows

infinite-acting radial flow behavior, and the stress-dependent fracture permeability case

shows a higher derivative level indicating lower reservoir permeability. At the late stage,
 99

both pressure drop and derivative behavior show the boundary of the fracture system.

The late stage behavior is actually a transition to the infinite-acting radial flow trend for

flow in the surrounding porous medium, as seen in Figure V-8. But the pressure

derivative for the stress dependent fracture permeability case still increases at very late

stage showed in Figure V-8 because the fracture permeability decreases with production.

p psi

well

Figure V-3. Pore pressure distribution after 360 days production.

 100

0.010

0.009
Fracture aperture (in)

0.008

Boundary
0.007

0.006

Well
0.005

0.004
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Time (hr)

Figure V-4. Fracture aperture declines with time.

350 0.010
Effective normal stress (psi)

300 wf 0.008

Fracture aperture (in)

250 0.006

200 0.004

σn′
150 0.002

100 0.000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Time (hr)

Figure V-5. Effective normal stress and fracture aperture change with time for the fracture
intersected with well.
 101

4500

4000
Fracture permeability (darcy)

3500

3000

2500 Boundary

2000

Well
1500

1000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Time (hr)

Figure V-6. Fracture permeability declines with time.

1000

100
Dp, dp/dlnt (psi)

pressure drop
fixed fracture permeability
10

pressure derivative
1

stress-dependent fracture permeability

0.1
0.001 0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-7. Comparison of transient pressure behavior at bottom hole with constant production
rate between fixed fracture permeability and stress-dependent fracture permeability case.
 102

10000

1000

100
Dp, dp/dlnt (psi)

fixed fracture permeability

pressure drop
10

1
pressure derivative
0.1
stress-dependent fracture permeability

0.01
0.001 0.01 0.1 1 10 100 1000 10000 100000 1000000
Time (hr)

Figure V-8. Comparison of transient pressure behavior at bottom hole with constant production
rate between fixed fracture permeability and stress-dependent fracture permeability case for a long
time production to show the flow behavior in the surrounded matrix rock.

The next example shown in Figure V-9 compares the previous stress dependent

fracture network case to that of a well producing from the unfractured porous medium.

In this comparison the pressure of the fracture intersected by the well is assumed as the

bottomhole pressure, and the pressure in a small square fracture element with cross

section area equal to that of the well is used for bottomhole pressure for the well in the

unfractured reservoir. (For example, if the well radius is 0.328 ft, both the length and

aperture of the fracture element is 0.581 ft.) Because the fracture element is meant to

represent the well, the fracture aperture and length are fixed during the production. The

comparison shows that the bottomhole pressure drops much less for the case with a

fracture network. The early time pressure derivative trends indicate that the effective

permeability of the fracture system is much larger than that of the matrix for the case

without any fracture. In late time the infinite-acting radial flow is the same for both cases.
 103

Figure V-10 compares the previous stress dependent fracture network case to a

well intersecting the only fracture in the reservoir. The fractures and matrix properties

are the same for the two cases, and the fracture length for the fracture intersected by the

well is the fracture spacing (310 ft) in the fractured reservoir case. Again, the

bottomhole pressure drops much less for the fractured network case (Figure V-11). The

infinite-acting radial flow behavior for both cases is the same in late time.

10000

without fracture dp

1000
dp/dlnt
dp, dp/dlnt (psi)

100

dp
10

fracture network
dp/dlnt
1

0.1
0.001 0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-9. Comparison of transient pressure behavior at bottom hole with constant production
rate between the case with a well connected with a fracture network and the case without any
fracture.
 104

310 ft

well

reservoir

Figure V-10. A well is intersected with a fracture in a non-fractured reservoir.

10000
dp

1000 dp/dlnt
dp, dp/dlnt (psi)

100
only one fracture
dp
10

dp/dlnt
1

fracture network
0.1
0.001 0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-11. Comparison of transient pressure behavior at bottom hole with constant production
rate between the case with a well connected with a fracture network and the case with only one
fracture in the reservoir.
 105

Figure V-12 compares the previous stress dependent fracture network case to a

well that does not intersect any natural fracture, with the well located at the center of the

matrix element in the center of the fracture network. Except for that the fracture spacing

of 290 ft (adjusted to make the fracture network area the same as in the other cases), all

other parameters are the same as the case in which the well is connected with the

fracture network. The bottomhole pressure drops much more compared with the case of

a well connected with a fracture network (Figure V-13). Initially the pressure derivative

for the case of the well that does not intersect a fracture shows the trend for infinite-

acting radial flow in the matrix permeability. Later, when the pressure disturbance

reaches the fracture network, the higher permeability in the fractures causes a leveling in

the pressure change. At the late stage, for both cases the fracture network conducts the

pressure disturbance to the outer matrix, and both cases have identical pressure

derivative trends.

From the perspective of pressure transient testing, the case with the well not

intersecting the fracture network is quite intriguing because it exhibits apparent classic

dual porosity behavior, but for the opposite reason from that usually applied for this

response. The initial and final dual porosity trend is that of the matrix, and not that of the

natural fractures, and the valley trend in the pressure derivative does not represent

recharge from the matrix; instead, it represents the higher permeability natural fractures

encountered before the pressure disturbance encounters the outer matrix with effectively

infinite extent.
 106

Matrix rock

Matrix rock Matrix rock

well

Matrix rock

Figure V-12. A well is located at the center of a matrix in a fractured network surrounded by matrix
rock.
 107

10000
well at the center of matrix in fracture network
dp

1000
dp/dlnt
dp, dp/dlnt (psi)

100

well connected with fracture network dp

10

dp/dlnt
1

0.1
0.001 0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-13. Comparison of transient pressure behavior at bottom hole with constant production
rate between the case with a well connected with a fracture network and the case with a well at the
center of a matrix in a fracture network.

For the original stress dependent fracture network case as the fracture

permeability declines with production, build up tests at different times show the change

in the fracture network permeability. Figure V-14 shows three successive simulated

build ups tests, conducted at different times. The rate history is listed in Table V-2. The

pressure derivative level is higher before the transition to the outer matrix behavior with

successively later buildup tests because the reservoir permeability declines with

production. It is difficult to use a single buildup test to determine the rock and joint

properties. However, these examples show that any one buildup test may indicate

whether the natural fracture system is stress sensitive, and manual history matching with

multiple pressure buildup tests may enable quantification of rock and joint properties.
 108

10

399 days

1 37 days
dp/dlntsup (psi)

t=5 days

0.1

0.01
0.001 0.01 0.1 1 10 100
∆t (hr)

Figure V-14. Pressure derivative curves for successive build ups.

Table V-2. Production rate history.

Duration Production rate

(days) (Res bbl/day)
5 12.6
2 0
30 12.6
2 0
360 12.6
2 0

5.1.3 Effect of initial effective normal stress

The slope of the trend in Figure V-2 gives the normal fracture stiffness, which

changes with the effective stress, from a small value at small effective stress to a rapidly

increasing value at high effective stress. As such, the fracture is more deformable when
 109

the reservoir pressure is close to the in situ stress than when there is a large contrast

between them. To study the influence of a higher stiffness, consider the same initial

fracture aperture of 9.02×10-3 in and fracture permeability of 4428 darcy before

production, but set the initial in situ stress to a value that increases the effective stress

while all other properties remain same. Figure V-15 shows the fracture permeability

change at the well for different effective in situ stress conditions. The influence of

production on the fracture permeability change strongly depends on the initial effective

stress condition, and decreases rapidly with increase in the effective in situ stress. The

fracture permeability only reduces 3.3% of the initial permeability of 4428 darcy for the

case with an effective in situ stress of 1450 psi. However fracture permeability loss for

the case with an effective in situ stress of 145 psi is 84.7% of the initial permeability.

4500
1450 psi

4000 725 psi

Fracture permeability (darcy)

3500

3000
290 psi
2500

2000

1500
Effective in situ stress=145 psi
1000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Time (hr)

Figure V-15. Effect of initial effective in situ stress on the fracture permeability change.
 110

5.1.4 Effect of ratio of hydraulic fracture aperture to mechanical fracture aperture

(wef /wf)

The ratio of effective hydraulic fracture aperture to mechanical fracture aperture

(wef/wf) is assumed as 1 in the above analysis. This assumption is only valid for fractures

with wide fracture apertures and smooth fracture surfaces. The effective hydraulic

fracture aperture wef is less than the mechanical fracture aperture wf, and the ratio wef/wf

is dependent on wf and the joint roughness coefficient (JRC) (Eq. (3-17)).

Figure V-16 compares cases with three different values for the wef/wf ratio. In

each case, the maximum fracture closure is 0.0393 in, and the initial mechanical fracture

aperture wf before production is 0.0131 in. In addition, the fracture aperture without

stress loading is assumed to be 0.0393 in and it is assumed to be reduced to 0.0131 in for

all fractures due to the compression in the reservoir before production. All other

parameters remain the same as in the previous examples. Figure V-16 shows that the

ratio wef/wf increases linearly with the increase of wf, the slope is a function of JRC and

decreases with the decrease of JRC. But the ratio wef/wf cannot exceed the limit value 1.

Three cases are investigated for unit ratio wef/wf, JRC=10.2 and JRC=12, respectively.

The fracture permeability is calculated from wef using cubic law (Eq. (3-15)) and updated

with the change of wef during the simulation.

 111

1.2

wef=wf
1

0.8
wef / wf

JRC=10.2
0.6

0.4
JRC=12
0.2

0
0 0.003 0.006 0.009 0.012 0.015
Mechanical aperture wf (in)

Figure V-16. The ratio wef/wf as a function of wf.

For the same mechanical aperture, the hydraulic aperture and permeability for the

case with JRC=12 is lower than the other two cases. Consequently, the pressure drop for

JRC=12 is higher than that in the other two cases, as seen in Figure V-17 and Figure

V-18. The higher pressure drop in turn causes higher mechanical aperture change seen

in Figure V-19 leading to higher hydraulic aperture change seen in Figure V-20, and

thereby the high permeability loss seen in Figure V-21.

 112

2900

2800

wef = wf
p (psi)

2700

JRC=12 JRC=10.2
2600

2500
0 2000 4000 6000 8000 10000

Time (hr)

Figure V-17. Bottom hole pressure declines with time for three cases: wef=wf, JRC=10.2 and
JRC=12.

100
Wef=Wf
JRC=10.2
10 JRC=12
dp/dlnt (psi)

0.1

0.01
0.0001 0.001 0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-18. Log-log plot of the pressure derivatives for three cases: wef=wf, JRC=10.2 and JRC=12.
 113

0.014
Mechanical aperture w f (in)

0.012

0.010

wef = wf
0.008
JRC=10.2
JRC=12

0.006
0 2000 4000 6000 8000 10000
Time (hr)

Figure V-19. The mechanical aperture of fracture intersected with well changes with time for three
cases: wef=wf, JRC=10.2 and JRC=12.

0.014
Effective hydraulicl aperture wef (in)

0.012

0.010

0.008 wef = wf

0.006
JRC=10.2
0.004
JRC=12

0.002
0 2000 4000 6000 8000 10000
Time (hr)

Figure V-20. The effective hydraulic aperture of fracture intersected with well changes with time for
three cases: wef=wf, JRC=10.2 and JRC=12.
 114

10000
Fracture permeability (darcy)

8000

6000

4000
wef = wf

2000 JRC=10.2
JRC=12

0
0 2000 4000 6000 8000 10000

Time (hr)

Figure V-21. The permeability of fracture intersected with well changes with time for three cases:
wef=wf, JRC=10.2 and JRC=12.

5.2 Fracture aperture and permeability change under high anisotropic in situ stress

conditions

The examples in the previous section all assumed isotropic in situ stress

conditions. This section considers anisotropic in situ stress conditions. The shear

deformation of a fracture is approximately linear before yielding and is treated as linear

here, as is characteristic of a constant shear stiffness value. The shear stiffness is

abruptly reduced to zero after yielding as in Figure V-22. The yielding stress can be

calculated using Eq. (3-1). But the simplified formula given in Eq. (5-1) is used in this

study to calculate the yielding stress according the effective normal stress and the

internal friction angle.

τ peak = σ n' tan φi (5-1)

 115

where φi is the internal friction angle. For reservoirs already at the critical stress

conditions the fractures are already yielded. Therefore the fractures are very week and

the shear stress disturbance can result in large shear deformation. The shear deformation

will induce some normal deformation by dilation.

tanφ i σn′
Shear stress

Ks

Shear displacement
Figure V-22. The relation of shear stress and shear displacement used in the modeling.

In Figure V-23 a fractured reservoir with high anisotropic in situ stress (σ1=4350

psi, σ3=3335 psi) has are two sets of fractures with an angle of 60°. The shear stiffness

before yielding is 3.7×105 psi/in, the internal friction angle is 30°, the dilation angle is

5°, the fracture spacing is 437 ft, and all other parameters are the same as those in the

isotropic case listed in Table V-1. All fractures are already yielded before production

and the production with a constant rate of 12.6 Res bbl/day induces not only the normal

deformation but also large shear deformation. Figure V-24 shows the direction and
 116

magnitude of the shear displacement after 360 days production. If the shear dilation

induces more openness of the fracture than the closure induced by the increase of the

effective normal stress, the fracture permeability will increase with production instead of

reduction. Figure V-25 shows the fracture permeability distribution after 360 days

production. There is still reduction of fracture permeability for those fractures in dark

blue. But the fracture permeability for other fractures increases compared with the

initial fracture permeability of 4428 darcy. The fracture permeability and shear

displacement are compared and show consistent increase (Figure V-26). Figure V-27

shows that the fracture permeability increases with production both for the fracture

intersected by the well and for a fracture at the boundary with the maximum

enhancement. Figure V-28 shows the change and derivative of the bottomhole pressure,

which also shows the enhancement of fracture permeability with production compared

with the case of fixed fracture permeability. Therefore, under highly anisotropic stress

conditions production may increase the fracture permeability.

 117

Matrix rock

Matrix rock Matrix rock

Matrix rock

Figure V-23. Well located at the center of a fractured field under anisotropic stress field and the
fractured network is surrounded by matrix rock.
 118

Ds in

Well

Figure V-24. Shear displacement distribution after 360 days production for the case fractures are
already yielded before production. The arrow represents the shear direction.

kf darcy

Well

Initial kf=4428 darcy

Figure V-25. Fracture permeability distribution after 360 days production for the case fractures are
already yielded before production.
 119

kf darcy

Well

Initial kf=4428 darcy

Figure V-26. Distribution of fracture permeability and shear displacement (shown with arrows)
after 360 days production for the case fractures are already yielded before production.

12000
Maximum increase
Fracture permeability (darcy)

10000

Fracture intersected with well

8000

6000

4000
0 2000 4000 6000 8000 10000

Time (hr)

Figure V-27. Fracture permeability increases with production for the case the fracture are already
yielded before production.
 120

1000

blue lines: fixed fracture permeability

red lines: stress-dependent fracture permeability

100
Dp, dp/dlnt (psi)

dp
10

1
dp/dlnt

0.1
0.01 0.1 1 10 100 1000 10000
Time (hr)

Figure V-28. Log-log plot of pressure drop and pressure derivative for the case in which the fracture
permeability of most fractures are enhanced by production.

In Figure V-26 the shear displacement and fracture permeability distribution are

not symmetric to lines through the well and parallel to x and y directions. For this case,

neither the fracture network nor the fracture intersected with the well are symmetric.

Before further comment on symmetries that do appear in this case, it is instructive to

consider the example shown in Figure V-29 for a well producing from four fractures

located at the center of the fracture network. In this case the resulting fracture network is

symmetric about the well, and both the permeability and aperture changes are symmetric

in x and y directions. It is now apparent that the asymmetries in Figure 5-27 arise from

the asymmetries in both inner and outer boundary conditions.

In both cases permeability is enhanced in a similar way. Those fractures at the

top and bottom which incline toward inside of the fracture network have larger shear

displacement, thereby inducing higher permeability. As the whole fracture network is

 121

compressed and moves inside with the reduction of reservoir pressure and the direction

of maximum principal stress (Figure V-23) tends to have a larger displacement than the

direction of minimum principal stress.

Figure V-29. Distribution of fracture permeability and shear displacement (shown with arrows)
after 360 days production for the case with symmetric fracture network and production wells.

5.3 Chapter conclusions

This chapter provided applications of the method described in Chapter IV under

isotropic in situ stress conditions and highly anisotropic in situ stress conditions.

Fracture aperture and permeability decrease with pressure depletion in naturally

fractured reservoirs under isotropic stress conditions, and the magnitude of the decrease

is dependent on the initial effective in situ stress. For low initial effective in situ stress
 122

(the reservoir pressure is very close to the magnitude of stress), the normal stiffness is

small if the initial normal stiffness is small, i.e., weak fractures. The small change of

reservoir pressure and effective stress can induce large fracture closure and permeability

loss. But for hard rock (high initial normal stiffness) or high effective in situ stress, the

normal stiffness is large, and the changes in fracture aperture and permeability are small

even for large reservoir pressure change. For rough fractures, the effective hydraulic

aperture is smaller than the average mechanical aperture. If the difference is neglected,

the influence of production on the fracture permeability reduction at isotropic stress

conditions will be underestimated. For highly anisotropic stress, the fractures can be at

the critical stress condition and even a small change in the shear stress can induce large

shear displacement. As a result, the fracture aperture and permeability can be enhanced

due to shear dilation while the reservoir pressure is declining.

 123

CHAPTER VI

CONCLUSIONS AND RECOMMENDATIONS

Production in naturally fractured reservoirs will cause reservoir pressure change,

thereby changing the stress. The stress change will change the fracture aperture and

permeability, thereby influencing the production. The coupled interactions exist in the

fractured porous media: (i) fluid pressure change induces solid deformation and stress

change; (ii) stress change induces fluid volume change and fluid pressure change; (iii)

fracture deformation induces the change of pore pressure and stress in the whole field

(the influence disappears at infinity); (iv) the change of pore pressure and stress at any

point has an influence on the fracture and induces fracture deformation. A method is

developed in this study to consider all of these coupled processes to model the fracture

aperture and permeability change during production in naturally fractured reservoirs.

6.1 Conclusions

The main contributions and conclusions from this study are summarized as

follows:

1. A method is developed to combine the fully coupled DDM with the Barton-

Bandis model of fracture deformation. The fully coupled DDM gives the

global fracture deformation and the Barton-Bandis fracture deformation

model gives the local fracture deformation. The combination of the fully
 124

coupled DDM and the Barton-Bandis fracture deformation model makes

every fracture deformation comply with both local and global relations.

2. Fracture aperture and permeability decrease with pressure reduction caused

by production in naturally fractured reservoirs under isotropic stress

conditions, but the magnitude of the changes are dependent on the initial

effective in situ stress. For low initial effective in situ stress (the reservoir

pressure is very close to the magnitude of stress), the normal stiffness is small

if the initial normal stiffness is small, i.e., weak fractures. The small change

of reservoir pressure and effective stress can induce large fracture closure and

permeability loss. But for hard rock (high initial normal stiffness) or high

effective in situ stress, the normal stiffness is large. The change of fracture

aperture and permeability is small even for large reservoir pressure change.

Therefore, whether the reservoir is stress sensitive can be decided by

laboratory tests on the properties of fractures and field tests of the in situ

stress. For stress sensitive fractured reservoirs, the method developed in this

study can be applied to evaluate the change of fracture permeability during

production and its influence on production.

3. For rough fractures, the effective hydraulic aperture is smaller than the

average mechanical aperture. If the difference is neglected, the influence of

pressure reduction caused by production on the fracture permeability

reduction under isotropic stress conditions will be underestimated.

 125

4. For highly anisotropic stress, the fractures can be at the critical stress

condition, and a small change of the shear stress can induce large shear

displacement. The fracture aperture and permeability can be enhanced due to

shear dilation while the reservoir pressure is decreasing.

6.2 Recommendations

The model is only for two-dimensional single-phase flow in a naturally fractured

porous medium. A three-dimensional model will be better to consider the influences

from all three principal stresses – the maximum horizontal stress, the minimum

horizontal stress and the vertical stress. Single-phase flow rate is a simplified case for

oil and gas reservoirs, and future work should consider two-phase or three-phase flow.
 126

NOMENCLATURE

α = Biot’s poroelastic coefficient

β = Angle counterclockwise from the x-axis to fracture

segment.

ij
γ = π/2+βi-βj

φ = porosity

φd = dilation angle

φb = basic friction angle

φi = internal friction angle

φr = residual friction angle

ρf = fluid density

σ = stress tensor

σc = shear strength

σ′n = effective normal stress

τ = shear strength

τ ξ , τh = step time

τpeak = peak shear strength

µ = fluid viscosity

υ = Poisson’s ratio

υu = undrained Poisson’s ratio

 127

ζp = pore space change

ζs = solid grain deformation

∆L = fracture length

a = half length of fracture segment

aa, b = constants related with fracture normal deformation

c = fluid diffusivity

c0 = cohesive strength

cf = fluid compressibility

cfr = fracture compressibility

ct = total compressibility of fracture

cmt = total compressibility of matrix

e = strain tensor

f (x,y) = a function defined in Eq. (2-2)

f (x , y ) = a function defined in Eq. (2-5)

k = permeability

kf = fracture permeability

n = Vef/Vf

p = pore pressure

q = flow rate

qf = flow rate through fracture

qint = interface flow rate between fracture and matrix /fluid

source
 128

qs = injection/production rate

t = time

u = displacement

v = interface flow

wf = fracture aperture

wef = effective hydraulic fracture aperture

x,y = co-ordinate positions in the global co-ordinate system

x, y = co-ordinate positions in the local co-ordinate system

ij
A = influence coefficient for normal stress by the normal
displacement discontinuity defined in Eq. (2-10) for the
elastic DDM or in Eq. (2-38) for the poroelastic DDM

ij
B = influence coefficient for normal stress by the shear
displacement discontinuity defined in Eq. (2-10) for the
elastic DDM or in Eq. (2-38) for the poroelastic DDM

ij
C = influence coefficient for normal stress by fluid
source/interface flow rate defined in Eq. (2-38)

D = fracture displacement

Dn = normal fracture displacement

Ds = shear fracture displacment

Dnmax = maximum possible closure

ij
E = influence coefficient for shear stress by the normal
displacement discontinuity defined in Eq. (2-10) for the
elastic DDM or in Eq. (2-38) for the poroelastic DDM
 129

ij
F = influence coefficient for shear stress by the shear
displacement discontinuity defined in Eq. (2-10) for the
elastic DDM or in Eq. (2-38) for the poroelastic DDM

G = shear modulus

ij
H = influence coefficient for pore pressure by the shear
displacement discontinuity defined in Eq. (2-38)

JCS = joint compressive strength

JRC = joint roughness coefficient

ij
K = influence coefficient for shear stress by fluid
source/interface flow rate defined in Eq. (2-38)

Km = system bulk modulus

Kni = initial normal stiffness

Kn = normal stiffness

Ks = shear stiffness or solid bulk modulus

ij
L = influence coefficient for pore pressure by the normal
displacement discontinuity defined in Eq. (2-38)

M = Biot Modulus

ij
N = influence coefficient for pore pressure by fluid
source/interface flow rate defined in Eq. (2-38)

S = fracture spacing

Vf = actual fracture void volume or pore space

Vef = effective fracture void volume for fluid flow

 130

Subscripts

0 = initial

ξ = current time step

d = dilation

ef = effective

f = fluid or fracture

fr = fracture

h = index of time step

i, j = index of fracture segment

i = internal

inj = injection

int = interface

kk = bulk value

m = porous media system

mt = total of the porous media system

max = maximum

n = normal

o = oil

p = pore space

s = shear, solid/porous matrix, or source term

x,y,z = co-ordinate direction in the global co-ordinate system

x, y = co-ordinate direction in local co-ordinate system

 131

Superscripts

q = fluid injection source/interface flow rate between fracture

and matrix

dn = normal displacement discontinuity source

ds = shear displacement discontinuity source

Over scripts

ξ = Current time step

h = index of time step

i, j = index of fracture segment

 132

REFERENCES

Aguilera, R., 1995. Naturally Fractured Reservoirs (Second Edition). Tulsa, Oklahoma:
Pennwell Books.

Asgian M., 1988. A Numerical Study of Fluid Flow in Deformable Naturally Fractured
Reservoirs. Ph.D. dissertation, University of Minnesota, Minneapolis, Minnesota.

Asgian M., 1989. A Numerical-Model of Fluid-Flow in Deformable Naturally Fractured

Rock Masses. Int J. Rock Mech. Min. Sci. 26 (3-4): 317-328.

Bagheri, M., 2006. Modeling Geomechanical Effects on the Flow Properties of

Fractured Reservoirs. Ph.D. dissertation, University of Calgary, Calgary, Alberta,
Canada.

Bagheri, M. and Setteri, A., 2008. Modeling of Geomechanics in Naturally Fractured

Reservoirs. SPEREE 11 (1): 108-118.

Bandis, S.C., Lumsden, A.C. and Barton, N.R., 1981. Experimental Studies of Scale
Effects on the Shear Behavior of Rock Joints. Int J. Rock Mech. Min. Sci. and Geomech.
Abstr. 18 (1): 1-21.

Bandis, S.C., Lumsden, A.C. and Barton, N.R., 1983. Fundamentals of Rock Joint
Deformation. Int J. Rock Mech. Min. Sci. and Geomech. Abstr. 20 (6): 249-268.

Barton N., 1976. The Shear Strength of Rock and Rock Joints. Int. J. Rock Mech. Min.
Sci. & Geomech. Abstr. 13 (9): 255-279.

Barton, N. and Choubey, V., 1977. The Shear Strength of Rock Joints in Theory and
Practice. Rock Mech. 10 (1-2): 1-54.

Barton, N., 1982. Modeliing Rock Joint Behaviour from In Situ Block Tests:
Implications for Nuclear Waste Repository Design. Office of Nuclear Waste Isolation,
Columbus, Ohio, ONWI-308, p. 96.

Barton, N., Bandis, S. and Bakhtar, K., 1985. Strength, Deformation and Conductivity
Coupling of Rock Joints. Int J. Rock Mech. Min. Sci. and Geomech. Abstr. 22 (3): 121-
140.

Biot, M.A.,1941. General Theory of Three-Dimensional Consolidation. J. Appl. Phys.

12 (2): 155-164.
 133

Biot, M.A., 1956. General Solutions of the Equations of Elasticity and Consolidation for
a Porous Material. J. Appl. Mech., Trans. ASME 78, 91-96.

Boulon, M. J., Selvadurai, A. P. S., Benjelloun, Z. H., and Feuga, B., 1993. Influence of
Rock Joint Degradation on Hydraulic Conductivity. Int. J. Rock Mech. Min. Sci.,
Geomech. Abstr. 30 (7): 1311-1317.

Buchsteiner, H., Warpinski, N.R. and Economides, M.J., 1993. Stress-Induced

Permeability Reduction in Fissured Reservoirs. Paper SPE 26513 presented at 1993 SPE
Annual Technical Conference and Exhibition, Houston, Texas, 3-6 October.

Byerlee, J. D., 1978. Friction of Rocks. Pure Appl. Geophys. 116 (4-5): 615–626.

Carvalho, J.L., 1990. Poroelastic Effects and Influence of Material Interfaces on

Hydraulic Fracturing. Ph.D. dissertation, University of Toronto, Toronto, Canada.

Chen, H.Y., and Teufel, L.W., 1997. Coupling Fluid-Flow and Geomechanics in Dual-
Porosity Modeling of Naturally Fractured Reservoirs. Paper SPE 38884 presented at
1997 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, 5-8
October.

Cheng, A.H.-D. and Predeleanu, M., 1987. Transient Boundary Element Formulation for
Linear Poroelasticity. Applied Mathematical Modelling 11 (4): 285–290.

Chin, L. Y., Raghavan, R., and Thomas, L. K., 1998. Fully Coupled Analysis of Well
Responses in Stress-Sensitive Reservoirs. Paper SPE 48967 presented at the 1998 SPE
Annual Technical Conference & Exhibition, New Orleans, Louisiana, 27-30 September.

Cinco, H. and Samaniego, F., 1982. Pressure Transient Analysis for Naturally Fractured
Reservoirs. Paper SPE 11026 presented at the 1982 SPE Annual Technical Conference
and Exhibition, New Orleans, Louisiana, 26-29 September.

Crouch S.L. and Starfield A.M, 1983. Boundary Element Methods in Solid Mechanics.
London: Allen & Unwin.

Curran, J.H. and Carvalho, J.L., 1987. A Displacement Discontinuity Model for Fluid-
Saturated Porous Media. Proc. 6th Cong. Int. Soc. Rock Mech., Vol. 1, Montreal, 1987.

Davies, J. P., and Davies, D. K., 1999. Stress-Dependent Permeability: Characterization

and Modeling. Paper SPE 56813 presented at the 1999 SPE Annual Technical
Conference and Exhibition, Houston, Texas, 3-6 October.

De Swaan, A., 1976. Analytical Solutions for Determining Naturally Fractured

Reservoir Properties by Well Testing. SPEJ 16 (3): 117-122.
 134

Detournay, E. and Cheng, A.H.-D., 1988. Poroelastic Response of a Borehole in a Non-

Hydrostatic Stress Field. Int. J. Rock Mech. Min. Sci. Geomech. Abstr. 25 (3): 171–182.

Goodman R.E., 1976. Methods of Geological Engineering in Discontinuous Rocks. New

York: West Publication Company.

Gray, D. H., Fatt, I., and Bergamini, G., 1963. The Effect of Stress on Permeability of
Sandstone Cores. SPEJ 3 (2): 95-100.

Gutierrez, M., and Makurat, A.. 1997. Coupled THM Modelling of Cold Water Injection
in Fractured Hydrocarbon Reservoirs. Int J. Rock Mech. Min. Sci. 34 (3–4): 429-443.

Huang, T.H., Chang, C.S. and Chao C.Y., 2002. Experimental and Mathematical
Modeling for Fracture of Rock Joint with Regular Asperities. Eng. Fract. Mech. 69 (17):
1977–1996.

ISRM Commission on Standardization of Laboratory and Field Test, 1978. Suggested

Methods for the Quantitative Description of Discontinuities in Rock Masses. Int. J. Rock
Mech. Min. Sci. Geomech. Abstr. 15 (6): 319–368.

Iwai, K., 1976. Fundamental Studies of Fluid Flow Through a Single Fracture. Ph.D.
dissertation, University of California, Berkeley, California.

Jones, F.O. and Owens, W.W., 1980. A Laboratory Study of Low-Permeability Gas
Sands. JPT 32 (9): 1631-1640.

Lee, H.S. and Cho, T.F., 2002. Hydraulic Characteristics of Rough Fractures in Linear
Flow Under Normal and Shear Load. Rock Mech. Rock Eng. 35 (4): 299–318.

Lee, J., Rollins, J. and Spivey, J., 2003. Pressure Transient Testing. Richardson, Texas:
SPE Textbook Series Vol. 9.

Lewis, R.W., and Ghafouri, H.R., 1997. A Novel Finite Element Double Porosity Model
for Multiphase Flow Through Deformable Fractured Porous Media. Int. J. Numer. Anal.
Methods Geomech. 21 (11): 789–816.

Meng, F., 1998. Three-Dimensional Modeling of Two-Phase Fluid Flow in Deformable

Fractured Reservoirs. Ph.D. dissertation, University of Oklahoma, Norman, Oklahoma.

Min, K.-B., Rutqvist, J., Tsang, C. -F., Jing, L., 2004. Stress-Dependent Permeability of
Fractured Rock Masses: A Numerical Study. Int. J. Rock Mech. Min. Sci. 41 (7): 1191–
1210.

Najurieta, H., 1980. A Theory for Pressure Transient Analysis in Naturally Fractured
Reservoirs. JPT 32 (7): 1241-1250.
 135

Nelson, R.A., 1985. Geologic Analysis of Naturally Fractured Reservoirs. Houston,

Texas: Gulf Publishing Company.

Olson, R. and Barton, N., 2001. An Improved Model for Hydromechanical Coupling
During Shearing of Rock Joints. Int. J. Rock Mech. Min. Sci. 38 (3): 317–329.

Ostensen, R. W., 1986. The Effect of Stress-Dependent Permeability on Gas Production

and Well Testing. SPEFE 1 (3): 227-235.

Raghavan, R., Scorer, J. D. T., and Miller, F. G., 1972. An Investigation by Numerical
Methods of the Effect of Pressure-Dependent Rock and Fluid Properties on Well Flow
Tests. SPEJ 12 (3): 267-275.

Rice, J.R. and Cleary, M.P., 1976. Some Basic Stress Diffusion Solutions for Fluid-
Saturated Elastic Porous Media With Compressible Constituents. Rev. Geophys. 14 (2):
227–241.

Samaniego V, F., Brigham, W. E., and Miller, F. G., 1976. Performance-Prediction

Procedure for Transient Flow of Fluids Through Pressure-Sensitive Formations. Paper
SPE 6051 presented at the SPE-AIME 51st Annual Fall Technical Conference &
Exhibition, New Orleans, Louisiana, 3-6 October.

Samaniego V., F., Brigham, W. E., and Miller, F. G., 1977. An Investigation of
Transient Flow of Reservoir Fluids Considering Pressure-Dependent Rock and Fluid
Properties. SPEJ 17 (2): 140-150.

Samaniego V., F. and Cinco-Ley, H., 1989. On the Determination of the Pressure-
Dependent Characteristics of a Reservoir Through Transient Pressure Testing. Paper
SPE 19774 presented at the 1989 SPE Annual Technical Conference and Exhibition, San
Antonio, Texas, 8-11 October.

Shu, Z., 1999. A Dual-Porosity Model for Two-Phase Flow in Deformable Porous
Media. Ph.D. dissertation, University of Oklahoma, Norman, Oklahoma.

Sneddon, I.N., 1951. Fourier Transforms. New York: McGraw-Hill.

Snow,D . T., 1965. A Parallel Plate Model of Fractured Permeable Media. Ph.D.
dissertation, University of California, Berkeley, California.

Sun, Y., 1994. A Three-Dimensional Model for Transient Fluid Flow Through Fractured
Porous Media. Ph.D. dissertation, University of Minnesota, Minneapolis, Minnesota.

Sun, Z., Gerrard, C. and Stephanson, O., 1985. Rock Joint Compliance Tests for
Compression and Shear Load. Int. J. Rock Mech. Min. Sci. & Geomech. Abstr. 22 (4):
197-213.
 136

Thomas, R. D. and Ward, D. C., 1972. Effect of Overburden Pressure and Water
Saturation on Gas Permeability of Tight Sandstone Cores. JPT 24 (2): 120-124.

Vairogs J., Hearn, C. L., Dareing, D. W., and Rhoades, V. W., 1971. Effect of Rock
Stress on Gas Production From Low-Permeability Reservoirs. JPT 23 (9): 1161-1167.

Vairogs J. and Rhoades, V.W., 1973. Pressure Transient Test in Formations Having
Stress-Sensitive Permeability. JPT 25 (8): 965-970.

Vandamme L.M. and Roegiers J.C., 1990. Poroelasticity in Hydraulic Fracturing

Simulators. JPT 42 (9): 1199-1203.

Warren, R.E. and Root, P.J., 1963. The Behavior of Naturally Fractured Reservoirs.
SPEJ 3 (3): 245-255.

Witherspoon, P.A., Wang, J.S.Y., Iwai, K., and Gale, J. E., 1980. Validity of Cubic Law
for Fluid Flow in a Deformable Rock Fracture. Water Resour. Res. 16 (6): 1016-1024.
 137

APPENDIX A

FUNDAMENTAL SOLUTIONS FOR FLUID SOURCE

Induced pore pressure p q , displacement u q and stress σ q by continuous unit fluid

source along a line fracture segment.

(
r 2 = x − x' + y 2 ) 2
(A-1)

where x′ varies from –a to +a.

∞ e−u
E1 ( x ) = ∫ du (A-2)
x u

r
ξ= (A-3)
2 ct

Induced pore pressure:

µ
pq =
4πk ∫−a
a
( )
Ei ξ 2 dx ' (A-4)

Induced displacement:

a
α µ (1 − 2υ )  r 2 E1 (ξ 2 ) 
u xq = −ξ
− − 2ct (ln r 2 + E1 (ξ 2 ) )
2

 2ct e (A-5)
16π k G (1 − υ )  2  −a

 a
α µ (1 − 2υ )   x − x ' 
u =
q
y − 4ct arctan
16π k G (1 − υ ) 

a

−a
( )
+ y ∫ E1 ξ 2 dx '
 y  −a
 (A-6)


a e−ξ 2

− 4cty ∫ '
dx 
−a r2 
 138

Induced stress:

 a 
α µ (1 − 2υ )   e −ξ 
2

− (x − x ) 2 − 2 + E1 (ξ ) ( )
'  1
a 
σ = q 2
− 2 ∫ E1 ξ dx 
2 '
(A-7)
8π k (1 − υ )  ξ ξ
xx
−a
 −a 
 

a
α µ (1 − 2υ )   e −ξ 
2

σ =
q
yy 
8π k (1 − υ ) 
(x − x )
'  1
ξ 2
−
ξ 2
+ E1 (ξ 2 )

(A-8)
  −a

a
α µ (1 − 2υ )   1 e −ξ 
2

σ xyq = − y 2 − 2 + E1 (ξ 2 ) (A-9)

8π k (1 − υ )   ξ ξ  −a
 139

APPENDIX B

FUNDAMENTAL SOLUTION FOR DISPLACEMENT

DISCONTINUITIES SOURCE

1. Induced pore pressure, displacement and stress by the continuous unit normal

displacement discontinuity along a line fracture segment.

Induced pore pressure:

( )(
a
G (υu − υ )
p =−
dn  2 x − x'
−
2παr 2 (1 − 2υ )(1 − υu )  r2
−ξ 2 
1− e  ) (B-1)
−a

Induced displacement:

1   υu − υ  
(1 − 2υ ) ln r − 
   
( )
E1 (ξ 2 )  1 − e −ξ 
2

u =−   ln r + 2  − 2ξ 2 
dn

4π (1 − υ )   1 − υu  
x
 
a (B-2)
y2   υ − υ  1 e −ξ 2  
+ 2 1 +  u
1 − 2 + 2  
r   1 − υu  ξ ξ 
  − a

  x − x' 
− 2(1 − υ ) arctan
1
u =−
dn
 ln r
4π (1 − υ ) 
y
 y 
(B-3)
(x − x )y 1 +  υ
a
u −υ 
 1 e −ξ  
2
'
  
−
r2  1 − υ 1 − ξ 2 + ξ 2 
  u    − a
 140

Induced stress:

σ dn
=−
G ( ) ( )
 x − x ' 3 − x − x ' y 2  υ − υ 
+  u 

2π (1 − υ )   1 − υu 
xx
r4
a (B-4)

( ) ( )+
( ) ( )
−
( 
)
 x − x ' 3 − x − x ' y 2 3 x − x ' y 2 − x − x ' 3 1 − e−ξ 2 2 x − x ' y 2e −ξ 2  
 r4 r4 ξ2 r4   − a

σ dn
=−
G ( ) ( )
 x − x ' 3 + 3 x − x ' y 2  υ − υ 
+  u 

2π (1 − υ )   1 − υu 
yy
r4
a (B-5)

( ) ( +
) ( ) ( ) −
(
 x − x ' 3 + 3 x − x ' y 2 x − x ' 3 − 3 x − x ' y 2 1 − e −ξ 2 2 x − x ' 3 e −ξ 2  

)
 r4 r4 ξ2 r4   − a

σ dn
=−
G ( )
 x − x ' 2 y − y 3  υ − υ 
+  u 

2π (1 − υ )   1 − υu 
xy
r4
(B-6)
( ) ( ) ( )
a
 x − x ' 2 y − y 3 3 x − x ' 2 y − y 3 1 − e −ξ 2 2 x − x ' 2 ye−ξ 2  
 − + 
 r4 r4 ξ2 r4   − a

where vu is the undrained Poisson’s ratio.

2. Induced pore pressure, displacement and stress by the continuous unit shear

displacement discontinuity along a line fracture segment.

Induced pore pressure:

a
G (υu − υ )
p =−
ds 2y

2παr (1 − 2υ )(1 − υu )  r
2 2
(
2 
1 − e −ξ 

) (B-7)
−a
 141

Induced displacement:

  x − x' 
− 2(1 − υ ) arctan
1
u =−
ds
 ln r
4π (1 − υ ) 
x
 y 
(B-8)
(x − x )y 1 +  υ
a
u −υ 
 1 e −ξ  
2
'
 
+
r2  1 − υ 1 − ξ 2 + ξ 2 
  u    − a

u =−ds 1   υu − υ  
− (1 − 2υ ) ln r +    ln r +
E1 (ξ 2 )  1 − e −ξ
 +
2
( )
4π (1 − υ )  − υ 2ξ 2
y
 1 
u   2  
a (B-9)
y   υu − υ  1 e −ξ  
2
2
+ 2 1 +   1− +  
r   1 − υu  ξ 2 ξ 2 
  − a

Induced stress:

σ =−
G ( )
 3 x − x ' 2 y + y 3  υ − υ 
+  u 
−
ds

2π (1 − υ )  − υ
xx 4
r  1 u 

(B-10)
( ) ( )
a
 3 x − x ' 2 y + y 3 3 x − x ' 2 y − y 3 1 − e −ξ 2 2 y 3e −ξ 2  
− + + 
 r4 r4 ξ2 r 4  
−a

σ ds
=−
G ( )
 x − x ' 2 y − y 3  υ − υ 
+  u 

2π (1 − υ )  − υ
yy 4
r  1 u 

(B-11)
( ) ( ) ( )
a
 x − x ' 2 y − y 3 3 x − x ' 2 y − y 3 1 − e −ξ 2 2 x − x ' 2 ye−ξ 2  
 − + 
 r4 r4 ξ2 r4   − a
 142

σ ds
=−
G ( ) ( )
 x − x ' 3 − x − x ' y 2  υ − υ 
+  u 

2π (1 − υ )   1 − υu 
xy
r4
a (B-12)

( ) ( )
+
( ) ( ) −
( )
 x − x ' 3 − x − x ' y 2 3 x − x ' y 2 − x − x ' 3 1 − e −ξ 2 2 x − x ' y 2e −ξ 2  

 r 4
r 4
ξ 2
r4   − a
 143

APPENDIX C

ij
COEFFICIENT MATRIX, C p , FOR FLUID DIFFUSION IN A REGULAR

FRACTURE NETWORK

For a regular fracture network with nc columns and nr rows, the fracture segment

divided by the intersection points is the discretized fracture element. The fracture

segment is numbered according to the row (ir) and column (jc) as:

nf = (ir − 1) × nc + jc (C-1)

where nf is the index of fracture segment in the discretized fracture network. The

fracture segments are divided into two types – type (a) and type (b) according to the

orientation shown in Figure C-1. For any fracture segment (i, j) of type (a), there are 6

fracture segments directly connected with the segment and they are (i, j-1), (i, j+1), (i-1,

j), (i+1, j), (i-1, j-1), (i+1, j+1) shown in Figure C-2. For any fracture segment (i, j) of

type (b), there are also 6 fracture segments directly connected with the segment and they

are (i, j-1), (i, j+1), (i-1, j), (i+1, j), (i-1, j+1), (i+1, j-1) shown in Figure C-3. Finally,

for any fracture segment (i, j) of either type (a) or (b), the connected fracture segments

can be expressed as (i, j-1), (i, j+1), (i-1, j), (i+1, j), (i-1, j-(-1)j+1×(-1)i+1), (i+1, j+(-

1)j+1×(-1)i+1).

Using the Darcy’s law, the flow rate from the directly connected fracture

segments to the fracture segment (i, j) can be calculated.

 144

From (i, j-1):

p l +1 (i, j − 1) − p l +1 (i, j )
qW = (C-2)
 a (i, j − 1) a (i, j ) 
µ + 
 k f (i, j − 1) w f (i, j − 1) k f (i, j ) w f (i, j )

From (i, j+1):

p l +1 (i, j + 1) − p l +1 (i, j )
qE = (C-3)
 a (i, j + 1) a (i, j ) 
µ + 
 k f (i, j + 1) w f (i, j + 1) k f (i, j ) w f (i, j )

From (i-1, j):

p l +1 (i − 1, j ) − p l +1 (i, j )
qS 1 = (C-4)
 a(i − 1, j ) a(i, j ) 
µ + 
 k f (i − 1, j ) w f (i − 1, j ) k f (i, j ) w f (i, j ) 

From (i+1, j):

p l +1 (i + 1, j ) − p l +1 (i, j )
qN 1 = (C-5)
 a(i + 1, j ) a (i, j ) 
µ + 
 k f (i + 1, j ) w f (i + 1, j ) k f (i, j ) w f (i, j )

From (i-1, j-(-1)j+1×(-1)i+1):

qS 2 =
( )
p l +1 i − 1, j − (−1) j +1 (−1)i +1 − p l +1 (i, j )
(C-6)

µ
(
a i − 1, j − (−1) (−1) j +1 i +1
+
) a(i, j ) 

( j +1 i +1
) ( j +1
 k f i − 1, j − (−1) (−1) w f i − 1, j − (−1) (−1)
i +1
)
k f (i, j ) w f (i, j ) 

From (i+1, j+(-1)j+1×(-1)i+1):

qN 2 =
( )
p l +1 i + 1, j + (−1) j +1 (−1)i +1 − p l +1 (i, j )
(C-7)

µ
(
a i + 1, j + (−1) (−1) j +1 i +1
+
) a(i, j ) 

( j +1 i +1
) ( j +1
 k f i + 1, j + (−1) (−1) w f i + 1, j + (−1) (−1)
i +1
)
k f (i, j ) w f (i, j ) 
 145

where the superscript l+1 denote the new time step, a is the half length of fracture

segment, kf is the fracture permeability, wf is the fracture aperture. a(i, j), kf(i, j) and wf(i,

j) denote the half length, permeability and aperture of the fracture segment (i, j). The net

flow rate into the fracture segment (i, j) is:

qnet = qW + qE + qS 1 + qN 1 + qS 2 + qN 2 (C-8)

Substituting Eqs. (C-2)-(C-7) into Eq. (C-8) yields:

qnet =
COEW p l +1 (i, j − 1) + COEE p l +1 (i, j + 1) + COES 1 p l +1 (i − 1, j ) + COEN 1 p l +1 (i + 1, j )
( ) ( )
(C-9)
+ COES 2 p l +1 i − 1, j − (−1) j +1 (−1)i +1 + COEN 2 p l +1 i + 1, j + (−1) j +1 (−1)i +1
− (COEW + COEE + COES 1 + COE N 1 + COES 2 + COE N 2 ) p l +1 (i, j )

where

1
COEW =
 a(i, j − 1) a(i, j ) 
µ + 
 k f (i, j − 1) w f (i, j − 1) k f (i, j ) w f (i, j )
1
COEE =
 a(i, j + 1) a(i, j ) 
µ + 
 k f (i, j + 1) w f (i, j + 1) k f (i, j ) w f (i, j ) 
1
COES 1 =
 a (i − 1, j ) a (i, j ) 
µ + 
 k f (i − 1, j ) w f (i − 1, j ) k f (i, j ) w f (i, j ) 
(C-10)
1
COE N 1 =
 a(i + 1, j ) a(i, j ) 
µ + 
 k f (i + 1, j ) w f (i + 1, j ) k f (i, j ) w f (i, j )
 146

1
COES 2 =

µ
(
a i − 1, j − (−1) (−1)i +1
j +1
) +
a(i, j ) 

( j +1 i +1
) ( j +1
 k f i − 1, j − (−1) (−1) w f i − 1, j − (−1) (−1)
i +1
)
k f (i, j ) w f (i, j )
1
COEN 2 =
µ
 (
a i + 1, j + (−1) (−1)i +1
j + 1
) +
a(i, j ) 

( j +1 i +1
) ( j +1
 k f i + 1, j + (−1) (−1) w f i + 1, j + (−1) (−1)
i +1
)
k f (i, j ) w f (i, j ) 

The fluid volume change (the first right term in Eq. (4-10)) from the old time step l to

new time step l+1 in the fracture segment is:

( )
COEc p l +1 (i, j ) − p l +1 (i, j ) (C-11)

where

COEc = 2n w f (i, j )a(i, j )co (C-12)

The fracture length ∆L is substituted by 2a in Eq. (C-11).

Combining Eqs. (C-9) and (C-11), Eq. (4-10) for the fracture network in Figure C-1 is

discretized into:

COEW p l +1 (i, j − 1) + COEE p l +1 (i, j + 1) + COES 1 p l +1 (i − 1, j ) + COE N 1 p l +1 (i + 1, j )

( ) (
+ COES 2 p l +1 i − 1, j − (−1) j +1 (−1)i +1 + COE N 2 p l +1 i + 1, j + (−1) j +1 (−1)i +1 ) (C-13)
− (COEW + COEE + COES 1 + COE N 1 + COES 2 + COE N 2 + COEc ) p l +1 (i, j )
= −COEc p l (i, j ) − 2na(i, j )∆Dn + 2a(i, j )qint + qs
 147

The index of fracture segment can be calculated from the row number and column

number using Eq. (C-1). Therefore, Eq. (C-13) is rewritten using the index of fracture

segment.

nf nf W nf W nf nf E nf E nf nf S 1 nf S 1 nf nf N 1 nf N 1 nf nf S 2 nf S 2 nf nf N 2 nf N 2 nf nf nf
C p p + C p p l +1 + C p p l +1 + C p p l +1 + C p p l +1 + C p p l +1 + C p p l +1
l +1

(C-14)
nf nf nf nf nf nf
= COEc p l + 2n a ∆ D n − 2 a q int − q s

where

nf = (i − 1) × nc + j
nfW = (i − 1) × nc + j − 1
nf E = (i − 1) × nc + j + 1
nf S 1 = (i − 2) × nc + j (C-15)
nf N 1 = i× nc + j
nf S 2 = (i − 2) × nc + j − (−1) j +1 (−1)i +1
nf N 2 = i × nc + j + (−1) j +1 (−1)i +1

nf nf W nf nf E nf nf S 1
C p = −COEW C p = − COEE C p = −COES1
nf nf N 1 nf nf S 2 nf nf N 2
C p = −COE N 1 C p = −COES 2 C p = −COE N 2 (C-16)
nf nf
C p = COEW + COEE + COES 1 + COE N 1 + COES 2 + COE N 2 + COEc
 148

16

15

14

13

12

11

10

9
a b
8
7

ir=1

jc=1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Figure C-1. A regular fracture network.

 149

16

15

14

13

12

11

10

9 (i+1, j) (i+1, j+1)

8 (i, j-1) (i, j) (i, j+1)

7 (i-1, j-1) (i-1, j)

ir=1

jc=1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Figure C-2. Discretization of type (a) fracture segment.

 150

16

15

14

13

12

11

10

9 (i+1, j-1) (i+1, j)

8 (i, j-1) (i, j) (i, j+1)

7 (i-1, j) (i-1, j+1)

ir=1

jc=1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Figure C-3. Discretization of type (b) fracture segment.

 151

APPENDIX D

COMBINED SET OF LINEAR EQUATIONS IN MATRIX FORM FOR A

REGULAR FRACTURE NETWORK

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256

225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224

193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208

177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176

145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160

129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144

113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128

97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112

81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80

49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Figure D-1. A regular fracture network with indices of fracture segments.

For the fracture network and discretized fracture segments shown in Figure D-1,

a set of linear equations can be built from Eqs. (4-30) and (4-31), and the unknowns (p,
 152

∆Dn , ∆Ds and ∆qint ) for all fracture segments can be obtained by solving the following

equations in matrix form.

   1 
 1 C  p(t )   bD n 
1,1 1 1,1 1 1,1 1,i 1,i 1,i 1, 256 1, 256 1, 256 1
 A+ K n B + K n tan φd C  0 A B C  0 A B
 1,1 1,1 1 1,1 1,i 1,i 1,i 1, 256 1, 256 1, 256  1ξ   1 
 0 E F− K s K  0 E E K  0 A B C  ∆ D n   bD s

1ξ 1
 −1    
1,1 1,1 1,1 1,i 1,i 1,i 1, 256 1, 256 1, 256
L H N  0 L H N  0 L H N ∆ Ds bq
 1,1  1ξ   1 
int
1 1 1,i 1, 256
 Cp − 2n a 0 2a  Cp 0 0 0  C p 0 0 0  ∆ q int   bp 
   
                    
 i ,1 i ,1 i ,1 i ,i i i ,i i i ,i i , 256 i , 256 i , 256 
C  p(t )   bD n 
i i
 0 A B C  1 A+ K n B + K n tan φd C  0 A B
 i ,1 i ,1 i ,1 i ,i i ,i i i ,i i , 256 i , 256 i , 256  iξ   i 
 0 E E K  E F− K s K  0 E F K  ∆ D n  =  bD s 
 i ,1 i ,1 i ,1 i ,i i ,i i ,i i , 256 i , 256 i , 256  iξ
  i 
 i ,01 L H N  −1 L H N  0 L H N  ∆ D s   bq int 
iξ
 Cp 0  ∆ q int   bp 
i ,i i i i , 256 i
0 0 0  Cp − 2n a 0 2a  C 0 0

p
   
                    
  256   256 
C  p (t )   bD n 
256 ,1 256 ,1 256 ,1 256 ,i 256 ,i 256 ,i 256 , 256 256 256 , 256 256 256 , 256

 0 A B C  0 A B C  1 A +Kn B + K n tan φd
256ξ 256
 0 K  ∆ D n   bD s 
256 ,1 256 ,1 256 ,1 256 ,i 256 ,i 256 ,i 256 , 256 256 , 256 256 256 , 256
E F K  0 E F K  0 E F −Ks
 256 ,1 256 ,1 256 ,1 256 ,i 256 ,i 256 ,i 256 , 256 256 , 256 256 , 256  256ξ   256 
 0 L H N  0 L H N  −1 L H N  ∆ D s   bq int 
 256,1 256 ,i 256 , 256 256 256  256ξ   256  (D-1)
C p 0 0 0  C p 0 0 0  C p − 2n a 0 2 a  ∆ q int   bp 

where

ξ −1 ξ −1 ξ −1 i  ξ −1 ξ −1
m ij jh m ij jh m ij jh
 i
∑ A(t − τ h )∆ D n − ∑ ∑ B(t − τ h )∆ D s − ∑ ( )
i ih ih
bD n = −∑ ∑ C t − τ h q int − K n
∑
 ∆ D n + tan φ d∑ ∆ D s  + p 0
h=0 j =1 h=0 j =1 h=0 j =1  h=0 h=0 
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh ξ −1

∑ E (t − τ h )∆ D n − ∑ ∑ F (t − τ h )∆ D s − ∑ ∑ K (t − τ h ) q int + K s
i i ih
bD s = −∑ ∑ ∆ Ds
h=0 j =1 h=0 j =1 h=0 j =1 h=0
(D-2)
ξ −1 m ij jh ξ −1 m ij jh ξ −1 m ij jh

∑ L(t − τ )∆ D − ∑ ∑ H (t − τ )∆ D − ∑ ∑ N (t − τ ) q
i i
bq int = − p 0 − ∑ h n h s h int
h=0 j =1 h=0 j =1 h=0 j =1

i i i ξ −1 ih ξ ih
bp = 2n w f a c f p (τ ξ ) − ∆L∑ ∆ q int − ∑ q s
h=0 h=0

153
 154

VITA

Name: Qingfeng Tao

Address: Department of Petroleum Engineering, Texas A&M University,

College Station, TX 77840

Email Address: [email protected]

Education: B.S., Geology, China University of Geosciences, 2000

M.S., Geological Engineering, The University of North Dakota, 2006