You are on page 1of 161

FINITE DIFFERENCE MODELING OF OIL RECOVERY

BY WATERFLOODING USING HORIZONTAL

WELL INJECTORS

by

SOHAIL ARSHED FARUQI, B.S.P.E., M.S.P.E.

A DISSERTATION

IN

INTERDISCIPLINARY ENGINEERING

Submitted to the Graduate Faculty


of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
DOCTOR OF PHILOSOPHY

Approved

Accepted

December, 1998
© 1998. Sohail Arshed Faruql
All Rights Reserved
ACKNOWLEDGEMENTS

The author is deeply grateful to Dr. John J. Day, Professor and


Chairman of Petroleum Engineering Department, and would like to show
great appreciation for his devotion, guidance, support, and
encouragement. The author is truly Indebted to Dr. Scott M. Frailey,
Assistant Professor of Petroleum Engineering, for the enormous amount of
time that he spent to provide guidance and helpful constructive critique of
the author's work. The author also wishes to express his sincere thanks to
Dr. Lloyd R. Heinze, Associate Professor of Petroleum Engineering for his
help and motivation. Without the help of these faculty members, this
research was not possible.
The author would also like to sincerely thank Dr. George B. Asqulth,
Professor of Geosciences, and Dr. Ralph H. Ramsey, Associate Professor
of Civil Engineering, for their help. The help and support provided by the
Petroleum Engineering staff throughout my studies is greatly appreciated.
I would also like to thank the Graduate School for their help,
understanding, and giving me several opportunities to finish my work.
The author is indebted to his dear colleagues Mohammad M.
Rahman and Farid Uddin Khan for their reassuring motivation and
inspiration.
This dissertation is dedicated to my dear and loving wife, Zarfishan,
son, Bilal, parents, Shaukat and Saghira, and brother, Saud, who have
given me everiasting love and support and who bring joy to my heart.

II
TABLE OF CONTENTS

ACKNOWLEDGEMENTS ii

ABSTRACT vi

LIST OF TABLES vii

LIST OF FIGURES viii

NOMENCLATURE xil

CHAPTER

1 INTRODUCTION 1

1.1 Research Objectives 2

2 LITERATURE REVIEW 3

2.1 Reservoir Simulation 3

2.1.1 Physical Model 4

2.1.2 Mathematical Modef 4

2.1.3 Numerical Model 7

2.1.4 Solution Techniques 12

2.1.5 Computer Model 13

2.2 Horizontal Versus Vertical Wells 13

2.3 Formation Damage 15

2.4 Performance Prediction of Horizontal Wells 16

2.5 Well Models 19


3 THEORY 33

3.1 Mathematical Model 33

3.1.1 Auxiliary Relations 35

3.1.2 External Boundary Conditions 37

3.2 Numerical Model 38

3.2.1 Formulation Technique 39

3.2.2 Discretization of the Flux Term 40

3.2.3 Discretization of the Accumulation Term 41

3.3 Well Model 46

3.3.1 Representation of Wells in this Study 47

3.4 Solution Technique 50

3.4.1 The Point Successive Over Relaxation

Method (PSOR) 50

4 DISCUSSION OF RESULTS 55

4.1 Validation of the Reservoir Simulator 56

4.1.1 Analytical Solution 56

4.1.2 Volumetric Material Balance 57

4.2 Sensitivity Analysis 58

4.2.1 Absolute Permeability 61

4.2.2 Porosity 62

4.2.3 Rock Compressibility 63

IV
4.2.4 Relative Pemneability 63

4.2.5 Capillary Pressure 64

4.2.6 Formation Thickness 64

4.2.7 Oil Viscosity 65

4.2.8 Oil Density 65

4.2.9 Oil Fomnation Volume Factor 66

4.2.10 Summary of Sensitivity Analysis 67

4.3 Potential of Horizontal Well Injectors in Waterflooding.. 67

4.3.1 Results Based on Oil Recovery 69

4.3.2 Results Based on Water Cut 74

5 CONCLUSIONS 127

6 RECOMMENDATIONS 129

REFERENCES 130

APPENDICES

A DERIVATION OF FLOW EQUATIONS 134

B DERIVATION OF FINITE DIFFERENCE FORM


OF EQUATIONS 140
ABSTRACT

Waterflooding is the most commonly used injection method for


secondary recovery of oil reservoirs. The selection of a horizontal well or a
vertical well as an injector Is an important Issue in waterflooding because
these two types of wells can behave differently due to their orientation in
the reservoir. Horizontal wells, due to their geometry, possess great
apparent potential in injection processes because these wells have large
contact with the formation as compared to the vertical wells. The
performance of vertical well injectors in waterflooding an oil reservoir have
been extensively investigated and reported in the literature. The detailed
analysis of the performance of horizontal well injectors, on the other hand,
is not found in the literature.
The objective of this research was to investigate the potential of
horizontal well injectors in waterflood operation with the help of a reservoir
simulator. A two-phase, black oil model was developed in this research to
study the potential of horizontal well injectors in waterflooding.
The results showed that vertical to horizontal permeability ratio and
formation thickness are the two main factors that can affect the
performance of a horizontal well injector as compared to a vertical well
injector. As the vertical to horizontal permeability ratio decreases, the
advantage of a horizontal well injector over a vertical well Injector
decreases. This change in the permeability ratio does not have any
significant effect on the performance of a vertical well injector. Also, the
smaller the fomiation thickness, the better the performance of the
horizontal well injectors.

VI
LIST OF TABLES

4.1 Data for Validation of Simulation model by Analytical Solution.... 78

4.2 Data for Validation of Simulation Model by Material Balance 78

4.3 Grid Dimensions, Well Flow Rates, and Rock Properties

Used for Sensitivity Analysis 79

4.4 Pressure Dependent Variables 80

4.5 Fluid Properties 80

4.6 Grid Dimensions, Well Flow Rates, and Rock Properties

Used for Simulation Results 81

4.7 Oil Recovered Versus Pore Volume Injected for h=7 ft 82

4.8 Oil Recovered Versus Pore Volume Injected for h=42 ft 82

4.9 Water Cut as a Function of Pore Volume Injected for h=7 ft 82

4.10 Water Cut as a Function of Pore Volume Injected for h=7 ft 82

VII
LIST OF FIGURES

2.1 An Example of a Reservoir Divided into Several Gridblocks 25

2.2 Cartesian Coordinate System 26

2.3 Block-centered and Point-distributed Grids 27

2.4 Hybrid Grids 28

2.5 Vomoi Grids 29

2.6 The External Boundaries of a Reservoir as Shown on a Grid 30

2.7 Multilateral Horizontal Wells 31

2.8 A Single Horizontal Well 32

3.1 Mass Balance 54

4.1 Physical Model for the Analytical Solution 83

4.2 Grid Model Used for the Simulator Validation by


Analytical Solution 83

4.3 Validation Results Using an Analytical Solution


(Early Time Data) 84

4.4 Validation Results Using an Analytical Solution

(Late Time Data) 85

4.5 Validation Results Using Volumetric Material Balance 86

4.6 HW-configuration (Top View) 87

4.7 VW-configuration (Top View) 88

4.8 HW-configuratlon (Side View) 89

4.9 VW-configuration (Side View) 89

viii
4.10 X, y, and z Direction Orientation 90

4.11 Relative Permeability Curves Used for the Simulation Results.... 91

4.12 Capillary Pressure Curve Used for the Simulation Results 92

4.13 Results of Sensitivity Analysis for Absolute Pemrieability 93

4.14 Results of Sensitivity Analysis for Pemrieability


Ratio of 0.1 and 0.3 94

4.15 Results of Sensitivity Analysis for Porosity


Values of 0.09 and 0.25 95

4.16 Results of Sensitivity Analysis for Rock Compressibility

Values of 2E-6 1/psi and 25E-6 1/psi 96

4.17 Curves Used for Relative Permeability Sensitivity (case 1) 97

4.18 Curves Used for Relative Permeability Sensitivity (case 2) 98

4.19 Results of Sensitivity Analysis for Relative Permeability

Data Given in Figures 4.17-4.18 99

4.20 Capillary Pressure Sensitivity Data (case 1) 100

4.21 Capillary Pressure Sensitivity Data (case 2) 101

4.22 Results of Sensitivity Analysis for Capillary Pressure


Data Given In Figures 4.20-4.21 102
4.23 Results of Sensitivity Analysis for Formation Thickness
of 21 feet and 105 feet 103
4.24 Results of Sensitivity Analysis for Oil Viscosity of
0.8 cp and 40 cp 104

4.25 Results of Sensitivity Analysis for Oil Density Values


of 40 Ib/cfl and 58 Ib/cft 105

IX
4.26 Data Used for Sensitivity Analysis of Oil Formation
Volume Factor 106
4.27 Results of Sensitivity Analysis for Oil Formation
Volume Factor Data Given in Figures 4.26 107

4.28 HW- and VW-configuration Response for Different


Permeability Ratios when h=7feet 108

4.29 HW- and VW-configuration Response for Different


Permeability Ratios when h=14feet 109

4.30 HW- and VW-configuration Response for Different


Permeability Ratios when h=21 feet 110

4.31 HW- and VW-configuration Response for Different


Permeability Ratios when h=28feet 111

4.32 HW- and VW-configuration Response for Different


Permeability Ratios when h=35 feet 112

4.33 HW- and VW-configuration Response for Different


Permeability Ratios when h=42feet 113

4.34 HW- and VW-configuration Response for


kv/kh=0.05 when h=7 ft and 14 ft 114

4.35 HW- and VW-configuration Response for


kv/kh=0.5 when h=7ft and 14ft 115

4.36 HW- and VW-configuration Response for


kv/kh=0.05 when h=7 ft and 28 ft 116

4.37 HW- and VW-configuration Response for


kv/kh=0.5 when h=7 ft and 28 ft 117

4.38 HW- and VW-configuration Response for


kv/kh=0.05 when h=7 ft and 42 ft 118

4.39 HW- and VW-configuration Response for


kv/kh=0.5 when h=7 ft and 42 ft 119

X
4.40 HW- and VW-configuration Response for Different
Permeability Ratios when h=7feet 120

4.41 HW- and VW-configuration Response for Different


Pemrieability Ratios when h=14feet 121

4.42 HW- and VW-configuration Response for Different


Permeability Ratios when h=21 feet 122

4.43 HW- and VW-configuration Response for Different


Permeability Ratios when h=28 feet 123

4.44 HW- and VW-configuration Response for Different


Permeability Ratios when h=35feet 124

4.45 HW- and VW-configuration Response for Different


Permeability Ratios when h=42feet 125

4.46 Production Rates for HW and VW-configurations


when h = 7 ft and kv/kh = 0.05 126

B.I Example of Ax^^i/z ''^^

XI
NOMENCLATURE

a extension of drainage volume of well in x direction, ft


A area, ft^
b extension of drainage volume of well in y direction, ft
b reciprocal of the F V F , STB/RB
b' derivative of b w.r.t. pressure
B F V F , RB/STB
c compressibility, psi'^
h extension of drainage volume of well in z direction, ft
kr relative permeability
kx permeability in x direction, md
ky permeability in y direction, md
kz permeability in z direction, md
nx number of grid blocks in x direction
nz number of grid blocks in z direction
p average pressure in drainage volume, psi
Po simulator wellblock pressure, psi
Po pressure in the oil phase, psi
Pw pressure in the water phase, psi
Pwf flowing bottomhole wellbore, psi
Pc capillary pressure, psi
q source/sink term
Q flow rate, S T B / D a y
ro equivalent wellblock radius, ft
rw wellbore radius, ft
S saturation
S' saturation derivative w.r.t. Pc

xii
t time, days
T transmissibility, STB/Day/psi
(Xw.Zw) well coordinates, (ft,ft)
Ax grid spacing in x direction, ft
Ay grid spacing in y direction, ft
Az grid spacing in z direction.ft

Symbols:
X integers (Babu et al.'s model)
X mobility, md/cp
y density in terms of pressure/distance
ji viscosity, cp
p density, Ibm/ft^
V integers
(j) porosity
A difference operator for s p a c e derivative
At difference operator for time derivative
V del operator
CO relaxation parameter

XIII
CHAPTER 1
INTRODUCTION

Waterflooding is the process of injecting water into oil reservoirs for


the purpose of pressure maintenance and secondary recovery of oil by
displacement with water. Generally, theoretical studies of waterflood field
performance are based on either analytical solutions or numerical reservoir
simulation studies. Analysis based on analytical solutions have the
advantage of being fast as compared to reservoir simulation studies.
These analytical solutions evaluate well test pressure behavior around a
horizontal well and do not account for layering, relative permeability and
capillary pressure effects, after water breakthrough performance, and
multiphase flow. On the other hand the reservoir simulation studies can
take into account the effect of relative permeability and capillary pressure,
layering, water breakthrough, and multiphase flow.
The selection of a horizontal well or a vertical well as an injector is
an important issue in waterflooding because these two types of wells can
behave differently due to their orientation in the reservoir. Horizontal wells,
due to their geometry, possess great apparent potential in injection
processes. The performance of an injection well in waterflooding an oil
reservoir depends upon the reservoir area invaded by the injected water.
This injected water displaces oil towards the producing well. As more area
in the reservoir is invaded by the Injected water, more oil Is displaced and
produced at the production well. In other words the perfomnance (fraction
of original oil in place recovered at the producer) of a horizontal well
injector or a vertical well injector can be analyzed by the amount of oil
recovered.
The decision to choose either a horizontal well or a vertical well as
an injector can be made by comparing the perfomnance of these two wells
in waterflooding an oil reservoir. Furthermore, a horizontal well injector Is
located parallel to the bedding plane (along the horizontal section of the
reservoir), so the injector water from the horizontal well invades the
reservoir in vertical direction and at the same time displace the oil from the
invaded region in the horizontal direction. The injected water from the
vertical well primarily travels in the horizontal direction while displacing the
oil towards the producer. Therefore, the increase or decrease in the
formation thickness and vertical permeability can have some effect on the
performance of a horizontal well injector.
The performance of vertical well injectors in waterflooding an oil
reservoir have been extensively investigated and reported in the
literature.^"^ The detailed analysis of the performance of horizontal well
injectors, on the other hand, is not found in the literature.

1.1 Research Objectives


The main objective of this research was to investigate the potential
of horizontal well Injectors in waterflood operations by:
1. developing a reservoir simulation model for waterflooding,
2. comparing the performance of horizontal well injectors with
vertical well injectors in waterflooding an oil reservoir,
3. evaluating the effect of change in vertical permeability on the
performance of a horizontal well injector,
4. analyzing the effect of fomriation thickness on the perfomnance of
the horizontal well injector.
CHAPTER 2
LITERATURE REVIEW

As mentioned in Chapter 1, the main objective of this study is to


analyze the potential of horizontal well injectors as compared to vertical
well injectors in waterfooding an oil reservoir. This was done with the help
of a reservoir simulator, developed for this study. An extensive review of
the literature was performed to understand the procedure in building a
simulator with proper representation of the horizontal well in a reservoir
simulator. The following is an overview of the literature.

2.1 Reservoir Simulation


The main purpose of reservoir simulation is to understand the
fluid flow behavior of a petroleum reservoir in order to optimize the
hydrocarbon recovery. There are four major stages to the modeling
process for reservoir simulation."*
First, a physical model is constructed. This model represents the
physics of fluid flow processes in the reservoir. Second, a mathematical
model is built based on the physical model. This involves nonlinear partial
differential equations. The third stage constitutes the transformation of the
mathematical model into the discretized numerical model. This model is
capable of producing solutions representing the basic physical features in
a reservoir. Finally, an algorithm followed by a computer program is
developed, which can efficiently perform the necessary computations for
solving the discretized numerical model.
In order to understand the complexities of the entire modeling
process, it is important to comprehend the physical behavior of the
recovery process.
2.1.1 Physical Model
The physical model consists of the physics that detennine the flow
mechanisms in the reservoir. As a well is drilled in a reservoir and starts
producing hydrocarbons, the decrease in the reservoir pressure
sun^ounding the well causes the hydrocarbons in the reservoir to expand
and continue moving towards the well. So the recovery of hydrocarbons is
accomplished by natural reservoir energy created by hydrocarbon
expansion. This type of hydrocarbon recovery is termed as primary
recovery.
The reservoir pressure decreases as the reservoir is depleted
through primary recovery, which causes the hydrocarbon flow rates to
decrease. At this stage of hydrocarbon production, water may be injected
into some wells (either new wells drilled for injection purpose or old
production wells converted to injectors) while the hydrocarbons are
produced from the other wells. This process, known as waterflooding, not
only maintains relatively high reservoir pressure and flow rates but also
physically displaces the oil towards the production wells. The process of
injecting water to displace the hydrocarbons towards the production wells
Is also known as secondary recovery. Waterflooding is the most
commonly used injection method. This is due to the general availability of
water, ease with which water can be Injected, the ability of water to invade
the formation, and water's efficiency in displacing oil.^

2.1.2 Mathematical Model


The physical laws that govern fluid flow in a reservoir are based on
conservation of mass, momentum, and energy. In addition to these laws,
Darcy's law is also incorporated into the flow equations. The resulting flow
equations are partial differential equations, which model the basic
processes that occur within the physical system. The flow equation for
each phase, present in the resen/oir, Is referred to as a single-phase
equation. The flow equations will be discussed in detail in Chapter 3 and
Apendix A. As an example, if a reservoir has two phases present, oil and
water, the single-phase equation for the oil phase will look like:

1 _
-V V, ^0 .^. As (2.1)
_Bo B.
CI _"^0 °

where VQ = oil velocity, given by Darcy's law as:

kk
Vo= ^(VPo-YoVh), (2.2)
Mo

and
1 -
-V- V, = mass flow into reservoir - mass flow out, (flux term), (2.3)
B,

qo = rate of withdrawal/injection from the reservoir, (2.4)

^ i ^ Q = change of mass in reservoir or rate of accumulation. (2.5)

A major difficulty in the modeling procedures is the choice of a set of


governing equations that accurately describe the complex physical
process. When the reservoir hydrocarbons consist of only two pseudo-
components, a non-volatile component (oil) and a volatile component (gas,
soluble in oil), the reservoir can be described using a mathematical model
known as a black oil model.^ The oil is only present in a liquid state in a
reservoir described by a black oil model. Equation (2.1) is an example of
the mathematical formulation of the black oil model. If more than two
hydrocarbon pseudo-components are present in the reservoir, the
reservoir can be described by a mathematical model known as a
compositional model. The presence of more than two hydrocarbon
components increases the complexity of the physical model by increasing
the number of single-phase equations.
An example of a reservoir that generally requires a compositional
simulation model is a light oil reservoir that contains gas condensates or
volatile oils. Hence the oil and gas are both present in liquid and vapor
states. The vapor/liquid equilibrium depends on the composition of oil and
gas components as well as the reservoir pressure.
The partial differential equations describing the flow in the reservoir
for either black oil model or compositional model are nonlinear in nature.
This nonlinear nature of the partial differential equations can be explained
by using the example of equation (2.1), which represents a black oil model.
The first term in equation (2.1), also referred to as the flux term (given by
equation (2.3)), has both saturation and pressure dependent variables.
Darcy's velocity (given by equation (2.2)) in the flux tenn has relative
permeability which is a saturation dependent variable and viscosity which
is a pressure dependent variable. The formation volume factor in equation
(2.3) is also a pressure dependent variable. Therefore, the change in the
flux term given by equation (2.3) is a function two independent variables,
saturation and pressure. This dependence of the flux term on two
independent variables is termed as a nonlinear behavior of the partial
differential equations. The nonlinearltles in the partial differential equations
can be grouped as weak nonlinearltles and strong nonlinearltles.^ The
variables that are functions of pressure of one phase can be considered
weak nonlinearltles. These Include fomriation volume factor, B, reciprocal
of the derivative of the formation volume factor (1/B), and viscosity, ^i. The
effect of these nonlinearities depends upon the degree of the pressure
change; moreover, this effect disappears in cases where pressure remains
constant. It is reasonable even in the case of variable pressure to evaluate
pressure dependent functions one step behind (at the previous time step,
where a time step is defined as the time for which the simulator solves for
the change in the values of independent variables defining a reservoir).
These independent variables are oil and water pressure and saturations.
These variables will be discussed in Chapter 3.
The variables that depend on saturation or capillary pressure are
termed as strong nonlinearities because the change in variables that
depend upon saturation or capillary pressure is usually more as compared
the variables which depend on pressure. Relative permeability, k^, and
saturation derivative with respect to capillary pressure, S w, fall in this group
(in equation (2.1) Sw results when the accumulation term is simplified as
will be shown in Chapter 3). The nonlinearity due to S w disappears if the
change in capillary pressure is linear over a time step. Hence kr exhibits
the sole nonlinearity in the system. This concludes the discussion of the
partial differential equations that form the mathematical model. The
different aspects of the numerical model will be discussed next.

2.1.3 Numerical Model


The next step after the selection of the mathematical model is the
selection of the formulation technique to be incorporated in the numerical
model. There are two basic methods for the fomnulation of multiphase
equations:
1. implicit pressure-explicit saturation method (IMPES),
2. fully implicit method.
The basic assumption In the IMPES technique is that there is no change in
the capillary pressure over a time step. The capillary pressure is updated
after each time step. The IMPES technique combines the two, single-
phase equations (for oil and water) into one multiphase equation that
represents both oil and water. The multiphase equation Is solved implicitly
for the pressure. The saturations (oil and water) are then calculated
explicitly for each spatial point in the discretized resen/oir model.
The fully implicit method involves two, single-phase equations (for oil
and water) in a form where the saturation derivatives with respect to time
(as shown on the right hand side of equation (2.1)) are converted to
pressure derivatives. This procedure of converting the saturation
derivatives into pressure derivatives is described in Appendix A. The two,
single-phase equations are solved implicitly for pressures in the oil and
water phase. The saturations are then calculated implicitly using capillary
pressure relations. The formulation techniques for the two methods are
outlined in detail by several authors.^^^
To convert a mathematical model into a numerical model, the
reservoir is divided into a number of blocks as shown in Figure 2.1. This
network of blocks imposed upon a reservoir is known as a grid and the
blocks are temried as gridblocks. Only two dimensions (x and y) are shown
in Figure 2.1 (to show the areal extent of a reservoir) for simplicity. The
actual reservoir also extends in the third dimension (z). The three-
dimensional location of gridblocks can be explained by a Cartesian
coordinate system shown in Figure 2.2. i, j , k refer to the location of a
gridblock in x, y, and z dimensions. The reservoir simulator predicts the
performance of a reservoir by solving the nonlinear flow equations for each

8
gridblock. Normally, the grid is selected with one or more of the following
considerations:^^
1. geology, size, and available data of the reservoir,
2. fluid displacement or depletion process to be modeled,
3. field history and future development plan,
4. accuracy desired for the numerical model,
5. software/hardware options available,
6. main goals of the study,
7. competence of the simulation team (experience in building a grid
system),
8. availability of computer resources, time, and money.
According to Aziz^^, commonly used grids are constructed by
aligning the gridblocks along orthogonal coordinate directions. The two
examples of these kinds of grids are the block-centered grid and the point-
distributed grid (Figure 2.3). In the block-centered grid (Figure 2.3a), the
reservoir is divided into gridblocks, and the gridblock points are located at
the centers of the gridblocks. In the point-distributed grid (Figure 2.3b), the
gridblock points are located, and the gridblock boundaries are placed
midway between these gridblock points. According to Azlz,^^ the block-
centered grid is advantageous for calculating accumulation terms, while
the point-distributed grid is more accurate for calculating flow between
gridblocks, but the differences between these two types of grids diminish
when the grid is almost uniform. This is because If uniform gridblocks (all
the gridblocks have same dimensions) are used in a block-centered grid,
the gridblock points are located at equal distance from each other.
Conversely, in the case of the point-distributed grid, if the gridblock points
are located at equal distances from each other, the resulting gridblocks
would have equal dimensions. The block-centered grid has been
extensively used due to its simplicity.^^
Aziz^^ also showed a hybrid grid model in his paper. As shown in
Figure 2.4, the hybrid grid, due to its shape provides better resolution in
the vicinity of the wellbore. Another kind of grid system, known as Vornoi
grids, is shown in Figure 2.5. It is evident from Figure 2.5, that the Vornoi
grid system does not constitute a single gridblock pattem. A Vornoi grid
can comprise of several kinds of gridblocks as shown in Figure 2.5, which
can range from hexagonal to hybrid or a combination of these two. Aziz^^
also acknowledges that unless there are compelling reasons (e.g., coning
problems or faults) the grid should be orthogonal and as uniform as
possible so that the computation time can be minimized, which is more in
case of Vornoi grids. For example, the Vornoi grids are flexible to
accommodate the presence of a fault in a resen/oir as compared to the
uniform orthogonal grids, but Vornoi grids require more computation than
uniform grids.
To solve the partial differential equations for each gridblock, the
partial derivatives are calculated. These partial derivatives in the single-
phase equations are approximated using a finite difference technique.
This procedure of approximating the partial derivatives is known as
discretization. The discretized, single-phase equations for oil and water
are solved for each gridblock for each time step. An example of the
discretization technique can be given by showing the calculation of a
pressure derivative with respect to location as follows:

gP_ Ap^Pi.i-Pi ^ (2.6)


ax Ax x,^i - Xj '

10
where i and i+1 are the location of two adjacent gridblocks in the x-
direction. A similar discretization technique can be applied to equation
(2.1). The detailed discretized form of the mathematical model used in this
study is presented in Chapter 3 and Appendix B.
The first comprehensive treatment of a displacement process
(waterflooding) described by including the capillary pressure effects in the
flow equations was reported by Douglas, Blair, and Wagner.^^ In their
work a numerical model was developed for determining the behavior of a
linear waterflood with the inclusion of capillary pressure. Their work was
limited to flow in a single dimension.
In 1959, Douglas, Peaceman, and Rachford® presented their work
which was based on the numerical solution to partial differential equations
for the multi-dimensional flow system describing the waterflooding process.
Douglas, Peaceman, and Rachford® also pointed out that the outcome of
the waterflood is related to the specific well locations, the reservoir and
well geometry, and the rock and fluid properties. It is evident that only a
sophisticated method (reservoir simulation), which takes into account all
the factors pointed out in the above discussion, can be successful In a
detailed prediction of reservoir perfomnance, that is, the amount of
hydrocarbons produced, amount of water produced, and change in
pressure and saturation. Mathematical and numerical models representing
multi-dimensional, multiphase flow have also been discussed in the
literature by other authors.^^^
In addition to the flow equations that are solved for each gridblock, a
set of boundary conditions is also specified to describe the flow Into or out
of the reservoir model. The boundary conditions are separated into
external and internal boundary conditions. The external boundary
conditions relate to any external flow (or no-flow) Into the resen/oir

11
boundary as shown in Figure 2.6. The internal boundary conditions
specify the location of wells and the flow condition (injection or production)
of the wells within individual gridblocks. The geometry (horizontal or
vertical) of the wells is also specified.

2.1.4 Solution Techniques


The solution to the numerical model can be computed by using
either direct or iterative methods. The direct methods are not suitable for
solving the system of partial differential equations described in section
2.1.2 as these equations are usually solved for a large number of
gridblocks and the roundoff error compiles as the final results are
generated. This is because the roundoff enror in direct methods is added
for each mathematical step taken towards the final result. In other words,
the final roundoff en'or in direct methods is an addition of all the roundoff
enters generated in each mathematical step. The roundoff error for the
iterative methods is only limited to one iteration. This is because the
roundoff error at the end of each iteration results in a different estimate of
the next iteration and hence, the final result only has the rounoff en"or
accumulated from the final iteration. Point Successive Over Relaxation
(PSOR) method Is a simple iterative method used in this study. PSOR is
an improvement of an iterative method known as the Gaus-Seidel method,
which is an improvement of another iterative method known as Point
Jacobi method. The algorithm of the three iterative methods, mentioned
here, will be discussed with the help of flow equations used in this study in
Chapter 3.

12
2.1.5 Computer Model
After the numerical model is derived and the solution technique is
chosen, a computer model is constructed. The computer model is a
transfomnation of the discretized form of the partial differential equations
into a computer program. The computer program is written in a computer
language that is appropriate for handling the calculations in the discritized
equations and solution techniques. This computer program Is called a
reservoir simulator. The commercial simulators are generally written in
FORTRAN or C*^

2.2 Horizontal Versus Vertical Wells


The main objective of a horizontal well injector is to increase the
well's contact with the reservoir and thereby improving the injection
process. The increased contact enables the horizontal well Injector to
invade parts of the reservoir which are not accessed by the vertical well
injector. Generally, a horizontal well is drilled parallel to the bedding plane
whereas a vertical well is drilled perpendicular to the bedding plane. In
order to drill horizontally, the deviation is achieved form vertical by using
'directional drilling practices'. The progress in the field of directional drilling
in the recent years has contributed to the development of the multilateral
well technology. A multilateral well is defined as one vertical wellbore
draining from two or more horizontal wells as shown in Figure 2.7. This is
very useful in the cases where one or more vertical permeability barriers
are present or the surface is environmentally sensitive. Therefore,
horizontal portion of the wells can be drilled fomn a single vertical wellbore
to access different parts of the resen/oir, hence, bypassing the
permeability banners. Furthemnore, if the environmental conditions on the

13
surface are not feasible for drilling several individual wells, one multilateral
well can be drilled to perform the function of several individual wells.
Horizontal wells are also very advantageous in the case of offshore
development, which can be explained by the term 'slot cost' (S), which is
defined as the ratio of total pipeline and platfomn cost (P) to the number of
well slots (N).^® Cade and Joshi^^ reported that in the case of the Dan
Field, North Sea, the ratio of H+S to V+S is about 1:4 (H being the cost of
horizontal well and V being the cost of vertical well). If the slot costs are
not included the ratio becomes 2:1. In other words, multilateral well
technology is used in case of horizontal wells to reduce the number of slots
(part of well going from the platform to the formation) thereby reducing the
cost as compared to the vertical wells, where each vertical well has to
have its own individual slot.
Joshi^®'^® evaluated the production perfomnance of horizontal and
vertical wells. He suggested not to use horizontal wells in homogeneous
formations thicker than 200 feet. This is because the advantage of a
horizontal well In a thick formation diminishes as compared to a fully
penetrating vertical well.
Taber and Seright^° reported the benefits of horizontal wells over
vertical wells, in waterflooding, based on analytical equations. Their study
showed that horizontal wells showed better areal sweep efficiencies,
higher flooding rates, and lower injection pressures as compared to vertical
wells. The ratio of the area invaded by the injected water to the area of the
reservoir is known as the areal sweep efficiency. The above mentioned
properties of horizontal wells make them very beneficial for all EOR
methods. The analytical equations used in their study, however, do not
account for capillary pressure, geologic layering, or after water
breakthrough performance.

14
Kossack, Kleppe, and Aasen^^ published the investigation of oil
production from the Troll Field in the Nonvegian North Sea. This field
comprises of a thin formation in deep water environments. A problem
specific to the Troll Field is that the conventional wells cone gas and/or
water within 2-3 years of production. If too few vertical well producers are
drilled, the wells will be shut in due to a high gas-oil ratio before the oil
recovery is satisfactory. If too many vertical wells are drilled, they interfere
with each other and gas coning problems may arise eariier. As a result of
this reasoning, drilling of horizontal wells was proposed in their study.
Kossack et al.^^ also showed a comparison of horizontal and vertical wells
for the Troll Field in the North Sea. These researchers reported that
horizontal wells exhibit much better performance in production of thin oil
zones than vertical wells. Specifically, Kossack et al.^^ showed in their
study that a 1500 foot long horizontal well gave better oil recovery than a
vertical well fully penetrating In the 43 feet thick fomnation.

2.3 Formation Damage


Due to its large contact area with the formation and long duration of
formation exposure to drilling fluids, drilling a horizontal well can cause
severe formation damage, and can drastically reduce the permeability in
the vicinity of the horizontal section of the wellbore as compared to drilling
a vertical well. The fomnation damage will be higher at the heal and will
decrease towards the toe. The portion of the horizontal well that is near
the vertical part of the well is known as heal and the end that extends
horizontally into the fomnation is known as toe as shown in Figure 2.8.
Byrom and Coulter^^ recently reported formation damage due to the
drill string rotation. This is not an issue in vertical wells, but it may be
significant for horizontal wells because the rotating drill string is forced

15
against the lower side of the wellbore by gravity for a considerable time.
The lower side of the wellbore can be crushed into small particles which
can block the pores thereby increasing the fomnation damage.
A stimulation treatment by matrix accidizing is not as effective in
horizontal wells as it is in vertical wells. One of the reasons is that the
production interval to be treated is an order of magnitude larger in
horizontal wells than in vertical wells.^ The specific acid volume per unit
length of the treatment inten/al conventionally used in vertical wells would
be more In horizontal wells because due to the extended length of the
horizontal wells, a large amount of acid is required. The time needed to
perform the stimulation treatment of a horizontal well should be limited to
avoid corrosion of the formation in the vicinity of the heal of the horizontal
well. This is because the heal of the horizontal well is continuously
exposed to the acid while the toe of the horizontal well is being treated with
the acid.

2.4 Performance Prediction of Horizontal Wells


Horizontal well performance can be predicted by using either
analytical solutions or reservoir simulation techniques. A detailed
discussion of these analytical solutions is beyond the scope of this study
so a brief overview will be given here.
Clonts and Ramey^"* presented the transient dimensionless pressure
for a horizontal well. Their analytical solution was based on a uniform flux
approximation for a horizontal well in an anisotropic reservoir of finite
thickness. The unifomn flux approximation means that flow is unifomnly
distributed along the horizontal wellbore. Log-log type curves were
presented in this work which can be used to determine reservoir
characteristics including directional pemneability or the horizontal well's half

16
length. They also stated named the late-time flow regime for a horizontal
well as pseudo-radial.
Goode and Thambynayagam^^ published a transient analytical
solution for the horizontal well pressure response during a drawdown and
buildup, in an anisotropic undersaturated oil zone. The analytical model in
their equations was developed by solving three-dimensional, diffusion
equations with the use of Laplace and Fourier transforms. The model
developed can be used to calculate the directional permeabilities, average
reservoir pressure, and skin factor.
Mutalik, Godbole, and Joshi^® evaluated the effect of drainage area
shapes on the productivity of horizontal wells during transient and pseudo-
steady state flow periods. Their research concluded that at very large
values of dimensionless horizontal well length the horizontal well pressure
response asymptotically approaches that of a fully penetrating infinite
conductivity vertical fracture. This dimensionless length is a function of
horizontal well length, reservoir thickness, and horizontal and vertical
permeaibilities. Moreover, off-centered wells are less efficient as
compared to the centrally located wells in producing a given reservoir
volume. The dimensionless length in their work is given by the following
equation:

L.=^j;^. (2.7)
2hA^k

where L is the productive length of the horizontal well, h is the formation


thickness, and ky and kh are the vertical and horizontal pemneability,
respectively.

17
Babu and Odeh^^ used a uniform flux approximation to provide an
analytical solution to evaluate the pseudo-steady state pressure behavior
of a horizontal well. They studied the effects of well length, well location
and degree of penetration of the horizontal well within a drainage area, and
vertical and horizontal permeability on well productivity. The degree of
penetration relates to the length of the horizontal well as compared to the
extent of the drainage area. Their study concluded that the well length and
degree of penetration has the strongest effect on the productivity of a
horizontal well.
In 1991, Goode and Kuchuck^® studied the inflow performance
formulas for horizontal wells during pseudo-steady state flow conditions.
The model developed by Goode and Kuchuck^® can be applied to
horizontal wells producing from a closed rectangular region where the well
is placed at any location in the region.
The analytical models briefly discussed above provide analysis of
horizontal well pressure response. These models account for directional
permeability, the drainage area around the horizontal well, and the
horizontal well length but do not completely account for geologic layering,
relative permeability, capillary pressure effects, after water breakthrough
performance, and multiphase flow. Numerical models address all of these
problems.
The numerical models are derived from mathematical models. An
example of a mathematical model is given in equation (2.1). The second
term, q, in equation (2.1) represents the source/sink term which represents
the withdrawal/injection form the resen/oir. In other words, the source/sink
terms represent the wells located in the resen/oir. The discussion of
representation of wells in the simulator will be given next.

18
2.5 Well Models
The reservoir simulator calculates pressures for each gridblock. In
order to relate the wellblock pressure (pressure of the gridblock where well
is located) to the flowing wellbore pressure, the concept of equivalent
wellblock radius, ro, is used. A comprehensive discussion of this
parameter is thus pertinent. The most significant treatment of well models
until the late 1980s was given by Peaceman.^^ This model is based on the
assumption of steady state flow condition at the boundary of the drainage
area of the well. According to Peaceman,^ the wellblock pressure, Po, is
related to the bottomhole flowing pressure, Pwf, for a vertical well by:

Po-Pv.=—-^t^ln^. (2.8)
27r(k,ky)/2Az "^w

Peaceman^ reported that the equivalent wellblock radius for an isotropic


system (kx=ky) is given by:

r, =0.14(Ax'+Ay')^. (2.9)

The equivalent well-block radius for an anisotropic medium was given as:

fck,/k.y^Ax-.(k./k,))^Ayf
r,=0.28'^^ ;^ ^y 7 '\y' ' • (2.10)
(ky/kxF^^(k./k,K^

Babu, Odeh, AL-Khalifa, and McCann^ presented a well model for


horizontal wells, which was derived by solving the finite difference form of
the heat flow equation in a rectangle with pseudo-steady state flow
19
assumption at the boundary of the drainage area of the well and uniform
flux along the wellbore. They used a technique based on an infinite series
sum, which provided expressions for the summations that generate the
exact solution for the finite difference heat flow equation. The analytical
solution in their study relates the wellblock pressure, Po, to the average
pressure, p, in the drainage volume of the well.

^ 887q|iB ^ 27ca M J_ '^w , '^w


P-Po = + (SJ (2.11)
V2nbylkJ{.) h VkAs a a^ 12n:

where the single series sum is given by:

(
cos'
\
^l(i-rXi-:-)
SX2„ = (2.12)
^Tcn^
V 2 n J Vi+af(l-x„-^"0
sin

and the grid aspect ratio is given as:

^Tin^
ttn =asin (2.13)
v2n,y

( r a k,
Ax (2.14)
a=\ AzA/k \ hVk,

Xn^(an+V^+^) .
(2.15)

20
v^(2i^+1),i^=0,1....(n,-1), (2.16)

and X = ( 2 j , + l ) . j , =0.l....(n,-1). (2.17)

where the well is located in the center of the gridblock with the well location
given by:

1
(2.18)

An analytical formula relating Pwf with p was given as:

7 1 x^ x2w >
w
^ 887q^iB ^ r„ h A/k V' a a^;
P-Pwc =
'w
B.. (2.19)
VInbyjk^, J k (
+0.25ln^-ln sin^=^ 1.84
TTL \

k, I h ;

where the boundary term is given as:

r27iz,'\
BE =ln(l-Ei) + 0.5ln 1-2C0S Ei+E? (2.20)
'^ h ;

where

27tmin(x„,a-x^) [k^
Ei = exp (2.21)
k^

21
The boundary term (BE) is usually considered negligible unless the well is
located close to the boundary of the drainage area. The equations (2.11)-
(2.19) are used to solve for (Po-Pwf). The resulting expression is substituted
in the following equation to obtain a formula for equivalent wellblock radius:

r 887q|iB ^
^r ^
Po-Pwf = In 'o.B (2.22)
V
27rb.yig< z y V^w 7

and the equation for wellblock radius is given by:

A ( 7ra 'A
VB k.^ . r . Tiz >
In + 0.25lri-'' -IH sin^=^ - 1 . 8 4 - 8 , - 8 , (2.23)
6hn^ \ k, J T.J I h J
V

In 1991, Peaceman^^ also provided guidelines regarding the


representation of horizontal wells in a numerical reservoir simulator. He
suggested that for a horizontal well, it is sufficient to interchange Ay and
Az, as well as ky and kz in his previous equations for vertical wells
(equations (2.8) and equation (2.10)). This is because of the fact that the
horizontal well is located along the horizontal plane and its performance is
effected by permeability and block dimensions in the vertical/z-direction
and x-direction; y-direction is not needed here because horizontal well is
located along the y-directlon. Hence, interchanging Ay and Az, as well as
ky and kz in equation (2.8) and equation (2.10) yields:

q^ In-^, (2.24)
Pwf = Po -
27c(k,k,)^Ay ""w

22
and

Peaceman^^ pointed out that for Babu et al.'s^ model, the following
rcorrection for equivalent wellblock radius should be used:

r _ r '^0,B
O (2.26)
o.5(k,/k,y'4+(k^/ky4

where ro.e is the equivalent wellblock radius defined by Babu et al.^° in


equation (2.23). It can be seen from the equations (2.8-2.25) for the two
well models, discussed above, that Peaceman's modeP^ is very simple
and Babu et al.'s model^ is very complex.
Brigham^^ and Peaceman^^-^^-^ pointed out that as the pemneability
in the vertical direction (kz) and the horizontal direction (kx, ky) differ (as
they do in most resen/oirs, kx and ky > kz), the circular wellbore acts
elongated (as an ellipse). Peaceman's modeP^ takes inoto account this
elongated wellbore, whereas Babu et al.'s model^ does not take into
account this elliptical behavior of the wellbore.
As mentioned in an eariier section, Kossack et al.^^ proposed a
development plan for oil production from the Troll Field using horizontal
wells. The horizontal well in their work is represented in the grid system by
one foot thick layer. The perforated section is represented by assigning
very high permeability to the well gridblocks. This is done by multiplying
the rock permeability of each gridblock by a large factor (10^-10*"). A

23
sensitivity study was made on the multiplication factor. The factor was
varied fomn 1.0 to 10^

24
> X

Figure 2.1: An Example of a Reservoir Divided into Several Gridblocks

25
i.j+1.k+;

-> X
i-1.j. k+X i j . k + l / i + 1 . j , k+j

*• X

•• X

Figure 2.2: Cartesian Coordinate System.

26
•0
c(0

</)
0)
^-^
c
D
^"L •c "C
M 0
•a Q.
3
• Q.
.O •0 0
k.
•fi
I
0 0)
•0
.s 0)
o ^•^ ^
a. 3 CD
k_
(/> c
4-: 4-^ •?A
T3
0
1.
v^
(U
C 3
—9 0 £
Q.
U -4i T3 ^CO
C 05
CO .i= O)
-0 0 '^
^ >;
i
0
•0 s->
£ ?»-"
~9
i" ^
c
0) s
0 • : U)
^ - - -9"^— — 9
ock-

i
^ £
0)
•c
oa m • S "
••
CO
CNJ
^i • •
mm 1
• c 0)
— ..0 — — — 0 D 3
o> 0
ca UL (/)

i-- — 0
1

M—

27
Figure 2.4: Hybrid Grid

Source: Aziz, K.: "Resen/oir Simulation Grids: Opportunities and


Problems," JPT, July. 1993.

28
1
1
(a) Canesian (b) locally rcfiiied Cartesian

(c) Cylindrical (d) Hexagonal (e) Curvilinear

(0 Hybrid-Canesian (£) Hybrid-hexagonal

Figure 2.5: Vornoi Grids

Source: Aziz, K.: "Reservoir Simulation Grids: Opportunities and


Problems," JPT, July, 1993.

29
The wide lines represent the external
boundaries of the reservoir model

Figure 2.6: The External Boundaries of a Reservoir as Shown on a Grid

30
Vertical section of the wellbore

Multilateral sections of horizontal wells

Figure 2.7: Multilateral Horizontal Wells

31
^Vertical section of the
wellbore

Horizontal section of
the wellbore

Heal Toe

Figure 2.8: A Single Horizontal Well

32
CHAPTER 3
THEORY

3.1 Mathematical Model


In order to formulate the flow equations, the following assumptions
were made:
1. flow is laminar and viscous,
2. t w o phases exist: oil and water,
3. oil and water are immiscible.
T h e single-phase flow equations for a black oil model were discussed in
Chapter 2 . T h e single phase equations used in this study for both oil and
water are given as:

-V-
1_ As (3.1)
V, Ho ~ ->.
B, at B„ °

and for water is:

1 . 4) (3.2)
V- V
Bw w
Hw
a Bw ^^

where V is the differential operator and its operation in the above


equations can be explained as:

^1 > r-i ^
a —V
M -> 1B„ J (3.3)
+ +
VB„ % dx. 3y 8z

33
Hence equations (3.1) and (3.2) represent three-dimensional flow via the V
(del) operator. In order to explain the different terms in these equations, a
reservoir in the shape of a block is shown in Figure 3.1, where the flow is
r 1 -. >i
in the x-direction only. The first term -V- —V in equations (3.1) and
V B AJ

(3.2) is known as the flux term and represents mass flow across the cross-
sectional area (y-z) at x minus mass flow across the cross-sectional area
(y-z) at x+Ax. Furthermore, the del operator in this term represents partial
differentials with respect to space as evident from the space coordinates
(x, y, z) in equation (3.3). The second term (q) in these equations Is the
source/sink term or the net rate of withdrawal/injection from the reservoir.
(• A
The third term is the change of amount of mass inside the
Va
B jy

reservoir or the rate of accumulation. Also, the third term has a partial
differential with respect to time.
Darcy's velocity, v, in equations (3.1) and (3.2) for the two phases
are formulated as:

kk / \
Vp=—^VPp-YpVh), (3.4)

and p = o, w, (p as a subscript will be used in this chapter consistently to


represent oil or water).
Combining equations (3.1) and (3.2) with equation (3.3) results in the
single-phase equations for a black oil model for two phases (oil and water).
These are the partial differential equations:

34
Vk(Vp„-YoVh)]-q„ = As (3.5)
a
V[UVp„-Y„Vh)]-q„=|- As (3.6)
c\

where (3.7)
^^pBp

A detailed derivation of the flow equations is provided in Appendix A.


In addition to the above partial differential equations, certain other relations
known as auxiliary relations are also required to complete the
mathematical model. These auxiliary relations will be given next.

3.1.1 Auxiliary Relations


The ratio of the volume that a phase occupies to the pore volume
(void space in the reservoir) is called the saturation of that phase. The
saturations of all phases present in the reservoir add to 1.

So+S,=1. (3.8)

Any two phases are misclble when they can be mixed together and
remain single phase. Oil and water do not mix and hence they are
immiscible. When oil and water come in contact with each other, they
remain separated by an interface. Molecules near the interface are
unevenly attracted by their neighbors on the opposite side of the interface,
thereby, creating a tension known as interfacial tension. Capillary
pressure, Pc, in the porous media is defined as the pressure difference
existing across the interface that separates the two immiscible phases (oil

35
and water). The phase that wets the surface of the reservoir rock in
preference to the other phase is known as the wetting phase. The
capillary pressure is generally defined as the pressure In the nonwetting
phase minus the pressure in the wetting phase. Assuming a water wet
rock, the capillary pressure is given as:

Pc=Po-Pw (3.9)

The formation volume factor of a phase (oil or water) is defined as


the volume of that respective phase at the reservoir pressure and
temperature required to produce one barrel of that same phase at the
stock tank pressure and temperature. The formation volume factor is
given as:

B =-^^, (3.10)
^P.STC

where RC = reservoir conditions (pressure and temperature),


STC = stock tank conditions (pressure and temperature).
The porosity of a porous medium (<t)) Is defined as the ratio of void
space (pore volume) in a rock, to the total or bulk volume of the same rock.

* = (t)ref[l + C , ^ . ( P o - P r e f ) ] . (^•''^)

where ^^ef is the porosity at reference pressure Pref, the pressure at which
the porosity (fef) was measured. Generally, porosity is measured at the
initial resen/oir pressure. Furthermore, the rock compressibility, Crock, is

36
comparable in magnitude to water compressibility and can be assumed
constant.^
Similarty, the formation volume factor of water (Bw) can be
determined by using the following equation:

BW=T; T^ Vi. (3.12)

where B^^^^ is the reference water formation volume factor at a reference


pressure, Pref. Also, the water compressibility is usually a very small
number and can be assumed constant.^

3.1.2 External Boundary Conditions


As discussed in the previous chapter, external boundary conditions
for the reservoir have to be specified for a system in order to complete the
mathematical model. A no-flow (or closed boundary) is used as external
boundary condition in this study. So there is no flow of any phase (oil or
water) across the external boundary of the reservoir model. In other
words, the component of velocity perpendicular to the external boundary is
zero. This can be expressed as a dot product as shown below:

v«n = 0, (3.13)

where v is the Darcy velocity given by equation (3.3) and Pi is the unit
vector normal to the boundary of the reservoir.

37
3.2 Numerical Model
The finite difference form of the single-phase equations In section
3.1 are as follows:

[AT,A(p,-y„h)]:;^-Q-,=V,,^A. (3.14)
vBoy

/^AC >
KA(P, - Ywh)]::: -Q:-I,=V,, i;4^ (3.15)
At \ B , ^

where the transmissibility in the x-direction is given as:

(3.16)
"^ ''P Ax

Similarty the transmissibility in the y-direction and z-direction is:

AxAz (3.17)
T,P = ^,P
Ay

AxAy
T.P = ^.P
(3.18)
Az

The superscript 'n+1' refer to the cun-ent time step. Also equations (3.14)
and (3.15) can be written in a simplified form as:

[AT„A*J^,;'-Q-J,=V,,1A, (3.19)

38
r^fi >
[AT„ AO J - - Q - , = V,,, ^A, f ^ , (3.20)

so that O, known as phase potential, is given by:

^o=Po-Yoh. (3.21)

^w=Pw-Ywh- (3.22)

The only difference between equations (3.1)-(3.2) and equations (3.14)-


(3.15) is that the former pair of equations is in mathematical form and the
latter pair of equations is in numerical form. Furthermore, equations
(3.19)-(3.20) are a simplified form of equations (3.14)-(3.15). Analogous to
the description of equations (3.1)-(3.2) given in section 3.1, the first term in
equations (3.19)-(3.20) represents the flux term, the second term, Q,
represents the source/sink term, and the term on the right hand side is the
change of amount of mass inside the reservoir or the rate of accumulation.
As discussed in Chapter 2, there are two formulation techniques which can
be used to derive final discretized form of equations (3.19) and (3.20).

3.2.1 Formulation Technique


Two formulation techniques are described in Chapter 2, namely fully
implicit and IMPES. A fully implicit finite difference fomnulation technique is
highly recommended^®'^"* for both modeling and computational reasons
because it solves for both pressure (oil and water) and saturations (oil and
water) implicitly as compared to IMPES technique which solves for

39
pressures implicitly and then calculates the saturations explicitly.
Therefore, a fully implicit finite difference technique was used in this study.

3.2.2 Discretization of the Flux Tenn


The discretized forms of the oil and water flux terms in equations
(3.19) and (3.20) for a three-dimensional model in Cartesian coordinates
(Figure 2.2) are given below:

^ o^^oJi.j.k= 'xo.i+1/2.j.k|Po.i+1,j.k ~Po,i.j.k ~ Yo.i+1/2,j.k (hj+ij^n - hj j J J +


' xo.i-1 / 2.j.k iPo.i-l,j.k ~ Po.i.j.k ~ Yo.i-1 / 2.j.k \^i-1.j.k ~ '^i.j.k j . +
'yo,i.j+1/2,kLPo,i.j+1.k ~ Po.i,j.k ~ Yo,i.j+l/2.k\ni,j+1.k ~''i.j.k/J"^
(3.23)
'yo.i.j-1/2.klPo.i.j-1.k ~ Po.i,j.k ~ Yo.i.j-l/2.k\ni.j-l,k ~^\,\,k). -I-

Tzo,i,j,k+1/2 Po.i,j.k+1 ~Po.i.j.k ~Yo.i.j.k+1/2Vni.j.k+1 " ' ' i . j . k / . +


'zo,i.j.k-1/2 Po.i,j.k-1 ~Po,i.j.k ~Yo.i.j.k-1/2V''i.j.k-l ' " ' ' i . j . k /

[A 'wA^vvJi.j,k~ 'xw.i+1/2,j.kLPw.i+1,j.k ~ Pw.i,>.k ~ Yw.i+1/2.j.k V'^i+I.j.k ~''i.j.k/J"*"

'xw.i-1/2,j.klPw,i-1.j.k ~ Pw.iJ.k ~ Yw.i-1/2,j.k\ni-1.j,k ~''i.j.k jj"'"

'yw,i.j+1/2.klPw.i.j+1.k "Pw.iJ.k ~ Yw.i,j+1/2.k v'i.j+l.k ~ ' ' i . j . k / J " ^


(3.24)
'yw,i.j-1/2J(LPw.i.j-1.k ~ Pw.i.j.k "" Yw.i.j-1/2.k vPi.j-l.k " ' ^ i . j . k / J ' ^

'zw.i.j.k+1/2lPw.i.j.k+1 ~ PwJ.j.k ~ Yw.i.jJ(+1/2V''i.j.k+1 ""'i.j.kj. +

Tzw.i.j.k-1/2|Pw,i.j.k-1 ~ Pw.i.j.k ~ Yw.i.j.k-1/2\ni.j.k-1 " h j j k j

The superscripts are dropped from the RHS of equations (3.23)-(3.24) for
simplicity. '1+1/2' stands for the boundary between gridblock i and 1+1,
whereas 'i-1/2' stands for the boundary between gridblock i-1 and i. To
further explain the calculation of variables at i+1 or i+1/2 in equations
(3.23)-(3.24), an example of pressure and transmissibility is used.
Pressure is an independent variable so its individual value for different

40
gridblocks is used. Whereas the transmissibility is a dependent variable,
which depends on rock and fluid properties, so transmissibility Is calculated
by averaging the properties of two gridblocks, across which the flow is
being calculated. This means that the transmissibility is not for a
respective gridblock but rather at the boundary between two gridblocks.
This explains the presence of '1/2' in the transmissibility subscripts.

3.2.3 Discretization of the Accumulation Term


The discretization of the accumulation term in equation (3.19) and
(3.20) is given by the following equations:

n+—
2 1> 1
n+- n+-
1 - S .w 2 b'o ^A,p, +
At ^'^^°^°>».i^ - At
IV J (3.25)
1
n+—
-b,2S-(A,p,-A,pJ

1 1
V"(t)
^A.(S.bJ,, n+— n+—
w ^+
At ^w w '-'tr'
__ V.
(3.26)
1
n+-
bJS:(A,p,-A,pJ

A derivation along with equations defining different parameters of the finite


difference equations (equation (3.23)-(3.26)) is given in Appendix B. Using
equations (3.19)-(3.26), a simplified finite difference fomi of the oil
equation is as follows:

41
(G0),,p-1,_, +(EO),,,p-,,, +(CO),^p-,^, -(AO),,PSI;,

+ (B0),,P-,,, +(D0),,P-U +(FO),,P:I;,. (3.27)


-(AAO),,p-\,-Q„..^=(ACUMO),,

Where the different coefficients in equation (3.27) are fomiulated as


follows:

(GO)i_j,, = T,oXiy-v2 . (3.28)

^ ^^'i.j.k ~ ' yo.i.j-1/2.k > (3.29)

V^^>'i.j,k - 'xo.i-1/2.jJ( . (3.30)

V"^/i,j,k ~ 'xoj+1/2.j.k . (3.31)

(DO),j, = T,„,,,i,j+1/2J< . (3.32)

^ ^^'ij.k ~ '^zoj.j.k+1/2 . (3.33)

(AG),, = (GO),, +(E0),, +(C0),, +(B0),, +(D0),, + ( F 0 ) , , +


V(D "+- "+- n+- n+— (3.34)
^ b' 2 _ b ' 2S 2 _ b 2S'n
At V y

(AAO),,=^br^S:. (3.35)

42
f 1 1 1 1 A f 1 ^
' n+- n+- n + - n+x^ —
(ACUMOX,, = ^ - b'-'o
' 2 + b ' o 2 ^$ w 2 ^ b o 2 $
^ 'w" -p" n+- (3.36)
b 2$'"
V J ^o ^w

V J_

Similarty, the final finite difference form of the water equation is as follows:

(GVV),,,P::;,., + ( E W ) , , P : ^ _ , , + ( C W ) , , P - L , , , -(Aw),.,p-r,
+(BW),,p-:,,, +(DW),,p";^,,, +(FW),^,p-l,i.k+1 (3.37)
-(AAW)^,,p-], - Q , . , , =(ACUMW),,

where the coefficients for the water equation (3.37) are given by the
following equations:

(GW),,, = T„,,
i.i.k-1/2 »
(3.38)

(EW),,,
i,j.k = T
' yw.i.j-1/2.k '
(3.39)

(CW),, = Txw,i-1/2,j.k » (3.40)

(BW);,,
i.j.k =
~ T
' xw.k-1/2.j.k •
(3.41)

(DW),j, = Tyw.i,j+1/2J( ' (3.42)

(FW),,
ij.k =
~ T
' zw.i.j.k+1/2 •
(3.43)

43
(AW),, = (GW),, +(EW),, +(CW\,, +(BW),, + ( D W ) J , + ( F W ) , . +
\
V(l) / n+—1 n+—1 1
n+-
(3.4.)
b' 2$ w 2 _ b w 23'n
At w w
V J

(AAW),,=^b:^S:, (3.45)

f 1 1 1 >\ ^ n4 A
n+— n+— n+-
(ACUMW),, = ^ w -b' 2$ 2+b 2$'" - P : w
(3.46)
w
V J

The coefficients in equations (3.27) and (3.37) are functions of both


dependent and independent variables. The Independent variables are oil
pressure (Po), water pressure (Pw), and water saturation (Sw). The oil
pressure dependent variables are the reciprocal of formation volume factor
(bo), derivative of reciprocal of formation volume factor with respect to oil
pressure (bo), viscosity (|jo), and porosity (cj)). The water pressure
dependent variables are reciprocal of formation volume factor (bw),
derivative of reciprocal of formation volume factor with respect to water
pressure (bw), and viscosity (\y^. The water saturation dependent
variables are relative permeability (kr) and saturation derivative with
respect to capillary pressure (S w).
These dependent variables need to be evaluated to generate the
coefficients in equations (3.27) and (3.37) for the next time step/level, so
that the dependent variables can be calculated. The current time step,
'n+1', in simulation is referred to as the cunrent time, where the dependent
variables in the flow equations (3.31) and (3.41) are to be evaluated. The
previous time step, 'n' is the time where the dependent variables have
already been calculated. Generally, it Is assumed that the change in
dependent variables is very small over a small time step. Therefore, S'w is
evaluated at the previous time step (n). The other parameters such as bo,

44
bw, b'o, b'w, ^o. and ^w are evaluated at the averaged value of pressure at
n+1/2, which is the average value of the pressure between n and n+1. The
saturation dependent parameters krw and kro are evaluated at n+1/2. The
pressure (oil and water) and saturation (water) values are averaged in time
between the previous time step (n) values and extrapolated current time
step (n+1) values. Hence, the averaged values of time, at time step n+1/2.
The extrapolation is performed because in order to solve for the values of
independent variables at the cun-ent time step, the values of the dependent
variables at the cunrent time step are required.
The extrapolated values of water saturation, oil pressure and water
pressure are calculated using the following equation:^

u-i=u"+^(u"-u"-0, (3.47)

where u represents the independent variables (pressure in the oil and


water phases and water saturation). The values at 'n+1/2' time level are
evaluated as follows:

^n.-^u_j^ (3.48)

So the values of the independent variables are extrapolated to the time


step (n+1) using equation (3.47) and then are averaged in time to n+1/2
using equation (3.48). The mathematical and numerical models have been
discussed in the previous sections. The source/sink temris in these models
refer to injection or productions wells. These wells can be either horizontal

45
or vertical. Therefore, it is pertinent here to discuss the pros and cons of
the well models presented in the literature review.

3.3 Well Model


A review of well models given by Babu et al.^° and Peaceman^^ was
provided in the Chapter 2. As mentioned in Chapter 2, Babu et al.'s
model^ uses a pseudo-steady state approximation at the reservoir
boundary. Furthermore, Babu et al.^ assumed that there Is uniform flux
along the wellbore. Peaceman's modeP^ is based on steady state flow
conditions at the reservoir boundary. The reservoir boundary in Babu et
al.^^ and Peaceman^ models refers to the drainage area of the well.
Brigham^^ and Peaceman^^-^^-^ presented that the circular radius of
a wellbore (rw) in anisotropic reservoirs should be corrected because in an
anisotropic reservoir, rw becomes elongated and behaves as an ellipse
rather than a circle. Babu et al.'s^ model does not provide con-ection for
this elliptical behavior of the circular wellbore in anisotropic reservoirs.
Peaceman^^ also provided a relation (equation (2.25)) to incorporate this
wellbore correction in Babu et al.'s model.^
In Kossack et al.'s^^ model, the perforated section is represented by
assigning a very high permeability to the well grid blocks by multiplying the
rock permeability by a large factor (10^-10^). Their description of a
horizontal well as a row of high permeability and high porosity blocks is an
unusual and unrealistic assumption about the physical nature of the
system because in actuality their simulator is still using a vertical well
representation model. The only difference between representation of a
vertical well and a horizontal well in their study is that they assign high
permeability to the gridblock which is suppose to have a horizontal well.
Kossack et al.^^ also acknowledge that their description of the horizontal

46
well as a row of very high permeability and porosity gridblocks is only an
approximation of the physical situation. The largest uncertainty in this
model is the multiplication factor used to increase the permeability of the
gridblock where the horizontal well is located.
It should be noted that injection processes usually create steady
state flow conditions due to pressure maintenance. Therefore, ideally the
well model should be based on steady state flow for injection processes.
Peaceman's modeP^'^ was used for both vertical wells and horizontal
wells in this study because it is based on steady state flow assumption,
incorporated a correction for anisotropy in rw, and is simple to implement.

3.3.1 Representation of Wells in this Study


Given the total liquid flow rate, qy, for a production well, the oil and
water flow rates are calculated using the following equations:

f X1 well.oT ^
qoT = qx. (3-49)
V^well.oT "^ ^well.wT J

^ X ^
ClwT =
q,, (3.50)
V^well.oT "^ ^well.wT J

where A^eii. in equations (3.49)-(3.50), represents the total oil and water
mobility which are given by the following equations:

f X\ vvetl,o \
(3.51)
^well.oT ~ 2-(
k=1 V^welLo "^^well.wy k

47
\
'weil.w
'well,wT (3.52)
k=1 V^well.o "^ ^well.w J^

where K is the number of gridblocks in which the well is completed (also


referred to as well nodes). The oil and water mobility at each well node
are given as:

\ -_I2_ (3.53)
'^well.o "" D '

\ — (3.54)
weil.w l^wB w

It is evident from equations (3.53) and (3.54) that the oil and water mobility
depend upon both saturation dependent (kr) and pressure dependent (^,
B) variables. The oil and water flow rates in the case of a production well
in each well node are calculated using the following relations:

(W1X^„J,
lok toT K
(3.55)
Z(WI;L^,O\
k=1

f'
lwk,prod Hok
'weil.w (3.56)
V ^well,o j

The water rates for the individual well nodes in the case of an injection well
are given by:

48
Q _ (W^we<l.w)h
^wk,inj ~ HT.inj K (3.57)
S(wa^,.„l
k=1

Where Wl in equations (3.55) and (3.57) is the well index. The Wl for a
vertical well is given by:

(wi).=^!#>, (3.58)
In

where ro is the equivalent wellblock radius discussed in Chapter 2. ro for a


vertical well is given by equation (2.10)

tk,/k,)>^Ax^^(k,/k,)^Ayf
ro = 0.28 (2.10)
{klKY^^ikJkV^

The Wl for the horizontal well is:

(3.59)
irt'O
V^wy

where ro is given by equation (2.25) as:

r oogk/^)^Ax^^(k,/k,y^Az (2.25)
(KIK/'AKIKV'

49
The equation (3.58) for Wlv contains Az because the vertical well is located
along the z-direction. Equation (3.59) for WIH, on the other hand contains
Ay, because horizontal well in this study is located along the y-direction in
the reservoir. Furthermore, the Wl shown in equation (3.55) and (3.57)
can be calculated by using either equation (3.58) or (3.59) depending on
the orientation of the well (vertical or horizontal).
Up to this point in Chapter 3, fomriulation of the flow equations for a
black oil model have been discussed. The next step is to discuss the
solution of these flow equations.

3.4 Solution Technique


The Point Successive Over Relaxation method (PSOR) was used to
solve the finite difference equations (3.27) and (3.37) because it is an
iterative mthod which is very simple to implement as will be shown in the
next section. As discussed in Chapter 2, iterative methods are more suited
for solving a large system of equations (for example, the single-phase
equations described in section 3.1 for each gridblock in the grid create a
large system of equations). The solution calculated by PSOR is faster as
compared to other similar iterative methods.

3.4.1 The Point Successive Over Relaxation Method (PSOR)


The finite difference equations (3.27)-(3.37) for oil and water are
solved using the PSOR technique. Equation (3.27) for oil will be used to
give an example of the PSOR technique. When the finite difference
equation for oil (equation (3.27)) is solved for the oil pressure of a
respective block, it can be written as:

50
^>' = "(A5)~K^'^°)y* -(GO),,,pri-, -(EO),^p-U
-(co),,prAj. -(BO),,pr;i.i. -(DO),,P;;;,, (3.6O)
-(F0),,p-J,., +(AAO),,p-r, +Q,,,,

In order to show the utility of PSOR Method it is customary to start with the
Point Jacobi Method,^ according to which, the pressure in equation (3.60)
is calculated by perfomriing several iterations. These iteration levels are
shown as superscripts in the following equation:

^ ^ ~m:;^^^^^^^^'^' -(Go),,pa... -(EO),,PI^,,,


-(CO),,pLt.,, -(BO),,p!,i,,, -(DO),,pLti.k . (3.61)
-(FO),,pak.i+(AAO),,pt:l,,+Q,,,

where (v = 0, 1, 2, ...) is the number of iterations or the iteration level. The


iteration can be performed on the above equation (3.61) starting at v=0
with some initial estimated value for Po^°^ (initial estimate was 3000 in this
study). The iteration should be continued until:

n(v^i)_r,(v) < s, (3.62)


r'o.i.j.k r^o,i.jJ(

where s is some acceptable tolerance, generally, between 10"^ to 10"^. A


tolerance of 0.001 was used in this study. It should be noted that the
maximum iteration level is different than time step level. Several iterations
are required to converge (equation (3.62)) on a solution at the cunrent time
step.

51
The Point Gauss-Seidel Method is a slight modification of the Jacobi
Method. According to this method, the "latest" available values of the
different variables (pressures and saturations) on the right hand side of
equation (3.61) are used for the iteration level. For example, when solving
for oil pressure (at the current iteration level v+1) of a gridblock at a
location (i, j , k), the values of oil pressure (at the iteration level v+1) for the
gridblock at a location (i-1, j - 1 , k-1) are already known (previously
calculated). In other words, when solving for pJ^'^^J, the values of

pLj-uk.pLrAk. and p[)';^i, are known and can be used in the calculation of

Pi7ii • Therefore, the equation for this method can be written as:

^ ^ "(A5^[^^^^^^^'-^-^ -(Go)M.p^:ti -(Eo),,p;


'o.i,j-1.k

- (CO),, P i - i , - (BO),, p[%,, - (DO),,., pl,ti.k . (3.63)


-(FO),,pL-Wi+(AAO).|,py,,+Q,.,;

and the stopping criteria is the same as for the Jacobi Method and given
by equation (3.62). It is noticeable that equation (3.63) does not make any
use of the value of pJj^A^ in the computation of p^^^"^.^. However, this value

(pLuk) is utilized in the PSOR method.^ If the value computed by equation

(3.63) is stored as PojjJ^, then the equation for PSOR method is given as:

p^^tJ=(l-a,)pl^i,+a)pS-r. (3-64)

where co is the relaxation parameter. The optimum value of co should give


the best rate of convergence. The optimum value is detennlned by making
trial and error simulation runs for different values of ©. In this study, a

52
value of 1.16 for co gave faster convergence as compared to the Point
Gauss-Seidel method. So in the above example, the PSOR method
modified the gridblock pressure generated by Gauss-Seidel method by
using the value of pressure for the same gridblock at the previous iteration
level. To check the convergence, this new value of pressure is used in
equation (3.62). Furthermore, it is noticeable that the PSOR method
becomes identical to the Gauss-Seidel method for © = 1.
The above discussion of PSOR completes the black oil model and
its solution. The results generated by the black oil model will be discussed
in Chapter 4 next.

53
q (withdrawal or injection)

x+Ax

Face x+Ax
Facex

Figure 3.1: Mass Balance.

54
CHAPTER 4
DISCUSSION OF RESULTS

A three-dimensional, two-phase black oil reservoir simulator based


on the mathematical and numerical models, presented in Chapter 3, was
developed. The main objective of this study was to investigate the
potential of horizontal well injectors in waterflooding an oil reservoir using
the developed reservoir simulator. This objective was achieved by
comparing the performance of horizontal well injectors with the
performance of vertical well injectors in waterflooding an oil reservoir. If a
horizontal well injector is placed in an oil reservoir, which has a vertical
producer, the performance of the horizontal well injector can be evaluated
by the amount of oil displaced towards the producer or in turn the amount
of oil produced at the producer. If the horizontal well injector is replaced by
a vertical well injector in the same reservoir, the performance of the vertical
well injector can also be analyzed in the similar manner as that of the
horizontal well injector. Then the performance of the two injectors can be
compared. This comparison will help determine the conditions under
which it is beneficial to use horizontal wells as injectors in waterflooding.
In order to find meaningful conclusions, the results of the
comparison between the performance of a horizontal well injector and the
performance of a vertical well injector, it is necessary to validate the model
results. This is accomplished by comparing the results generated by the
simulator to an analytical solution and a volumetric material balance
solution. Additionally, it is necessary to check if these results are
dependent on the specific values of different input parameters (rock and
fluid properties) used in the simulator. A sensitivity analysis on the input
parameters must be performed.

55
4.1 Validation of the Reservoir Simulator
To validate the simulator developed in this study, the results
generated by the simulator were compared with two different solutions.
1. single-phase analytical well test solution,
2. volumetric material balance solution.

4.1.1 Analytical Solution


The analytical and simulation results are compared for a single-
phase problem to validate the simulator. The analytical model selected to
perform this task comprises of one well producing at a constant rate. This
well is located in the center of a square reservoir. One side of the
reservoir acts as a constant pressure boundary and the other three sides
behave as no-flow boundaries as shown in Figure 4.1. The analytical
solution for this problem, in the form of dimensionless pressure, is provided
by Ramey et al.^^ To simulate the behavior of one well producing from a
square reservoir with one constant pressure boundary, a closed
rectangular reservoir with two wells was chosen as shown in Figure 4.2. It
can be seen in this figure that the reservoir dimensions are such that if the
reservoir is cut in half between the two wells, it can be visualized as two
square reservoirs joined together. One well produces at a constant rate
and the other well injects at the same rate thereby creating a constant
pressure boundary between the two wells.
To simulate single-phase flow with the two-phase simulator, oil and
water phases are given the same fluid properties. The well located in the
left half of the rectangular reservoir In Figure 4.2 is a producer. Therefore,
the left half of this reservoir simulates the physical model described in
Figure 4.1. The data used to generate the simulator solution is shown in
Table 4.1. The change in the wellbore pressure as a function of time was

56
calculated using the simulator and converted to dimensionless values
using the following equations:

kh r /M
''°^41.3q^B'P'-P"^')J- (4.1)

0.00634 k t
'-=^;i:H:^' (4-2)

where A is the drainage area of the square reservoir described for the
analytical model. The simulator results are plotted against the analytical
results in Figure 4.3 and Figure 4.4, which show a very good match
between the two results.

4.1.2 Volumetric Material Balance


To compare the results from the volumetric material balance with the
results from simulation model, the pressure and formation volume factor
data for Canyon Reef reservoir of the Kelly-Snyder Field^ was considered.
As the simulator developed in this study can only model two phases (oil
and water), only Bo data above the bubble point pressure was used for the
volumetric material balance calculations. This data is given in Table 4.2.
Using this data, the cumulative oil produced is calculated by the following
material balance equation:

Np=N (4.3)
V B.

57
where Np is the cumulative oil produced, N is the Initial oil in place, Boi is
the oil formation volume factor at the initial pressure and Bo is the oil
fomiation volume factor at the cun-ent pressure.
The pressure and Bo data used in generating material balance
results are used in the simulator and Np values are generated. To
generate the simulation results, a 10x10x3 grid is constructed. The
gridblock dimensions in the x and y-direction are 90 feet and in the z-
direction are 45 feet. One well with a production rate of 100 STB/Day is
placed in the center of the grid. The initial oil In place for both material
balance and simulation calculations is kept at 1.313 MMSTB. Np versus
pressure for both simulation and material balance results is plotted in
Figure 4.5. This figure shows that the results from the simulation run are
very close to the results from material balance.

4.2 Sensitivity Analysis


A high and a low value for the input parameters were selected to
perform the sensitivity analysis. A typical high and low for these input
parameters was picked from different sources."*"^ To exemplify the
importance of the sensitivity analysis, take the example of oil density. If
the horizontal well injector shows an advantage over a vertical well injector
for the picked range of oil density values then it can be concluded that this
advantage of the horizontal well Injector is not based upon any specific
value of oil density. Rather the advantage of horizontal well injector is true
within the picked range of values of oil density. The sensitivity analyses
were performed on the following parameters:
1. absolute pemrieability,
2. porosity,
3. rock compressibility,

58
4. relative pemrieability,
5. capillary pressure,
6. formation thickness,
7. oil viscosity,
8. oil density,
9. oil formation volume factor.
Figure 4.6 and Figure 4.7 show the grid configuration used for the
sensitivity analysis. Figure 4.6 shows the top view (x-y direction) of the
grid configuration with a horizontal well injector and a vertical well
producer, whereas Figure 4.7 shows the top view of the grid configuration
with a vertical well injector and a vertical well producer. The grid
configuration with a horizontal well injector will be refen^ed to as HW-
configuration, and the grid configuration with vertical well Injector will be
referred to as VW-configuration in the following text. Also, together these
configurations will be referred to as the two configurations. It can be seen
in these figures (Figure 4.6 and Figure 4.7) that both of the grid
configurations have the same number of gridblocks. The only difference
between the two configurations listed above is the orientation of the
injector (horizontal or vertical). Figure 4.8 shows the side view (x-z
direction) of the HW-configuration, and Figure 4.9 shows the side view of
the VW-configuration. Figure 4.10 further explains the x, y, and z-directlon
of the entire model. The x-direction and y-direction are also refen-ed to as
the horizontal direction and the z-direction is also refenred to as the vertical
direction.
The gridblock dimensions are given in Table 4.3. The x and y
gridblock dimensions are kept constant at 70 feet and the number of
gridblocks in the x and y directions is also same, which is equal to 9. This
can be seen in Figure 4.6 and Figure 4.7. The gridblock dimension in the

59
z-direction is 28 feet for the whole sensitivity analysis except for the
sensitivity of formation thickness, where it is varied from 7 feet to 35 feet.
The number of gridblocks in the z-direction is 3 (Figure 4.8 and Figure 4.9).
The pressure dependent fluid data (Bo, ^o) is given in Table 4.4. The
static fluid property data is listed in Table 4.5. As it is assumed in this
study that there is no gas present in the reservoir, the Bo values are kept
less than 1. It is also assumed that the temperature changes are not
significant and due to the pressure maintenance the change in po is
insignificant. The water and oil relative permeability curves are shown in
Figure 4.11 and the capillary pressure data can be seen in Figure 4.12.
These data were taken from the work performed by Pedrosa.^^ This data
was selected because he used this data to see the behavior of a reservoir
with two-phases (oil and water), and this dissertation also analyzes the
performance of horizontal well injectors in a reservoir where only two
phases (oil and water) are present. The input data discussed in this
section were kept constant for all of the analyses and only the parameter
for which the sensitivity was analyzed was varied.
In the next section, a comparison between the performance of a
horizontal well Injector and a vertical well injector is shown graphically for
different values of each of the parameters listed above. The text provides
a summary on the effect of these parameters on the increasing or
diminishing advantage of the horizontal well injector as compared to a
vertical well injector.

60
4.2.1 Absolute Pemieabilitv
The permeability in the vertical direction (kv) is generally less than
the permeability in the horizontal direction (kh) due to the overburden
pressure of the rock, ky is also refenred to as the permeability in the z-
direction (kz) and kh is also known as the permeability in the x-direction (kx)
and the permeability in the y-direction (ky). In this sensitivity, kh was varied
from 10 md to 600 md. A value 0.3 was used for pemrieability ratio (kj kh)
to determine the kh value to be used for the sensitivity. So when the kh (kx,
ky) is equal to 10 md, kv (kz) is equal to 3 md and when kh is equal to 600
md, kv is equal to 180 md. The fraction of original oil in place recovered as
a function of pore volume injected for these different values of permeability
is shown in Figure 4.13. It should be noted that the fraction of original oil in
place recovered will be refenred to as the oil recovery in the following text.
As the permeability in the horizontal and vertical directions increases the
recovery response for both HW-configuration and VW-configuration
increases. As permeability is the measure of the ability of the reservoir to
transmit fluid, so it is expected that when the pemneability increases, more
oil would be produced. Also the oil recovery response for the HW-
configuration is better than that of the VW-configuration for each
permeability value. The better recovery from the HW-configuration is also
anticipated because of the large contact of the horizontal well with the
fomiation.
For the absolute pemneability sensitivity, the pemneability ratio was
kept constant at 0.3, which makes it important to evaluate if the advantage
shown by the horizontal well injector over the vertical well injector in Figure
4.13, remains the same when the pemneability ratio is varied. This was
analyzed by comparing the oil recovery for ky/kh values of 0.1 with 0.3.
Figure 4.14 shows that there is no change in the oil recovery response of

61
VW-configuration for the two kv/kh values. Also the HW-configuration
shows an advantage over the VW-configuration for kv/kh value of 0.3. This
advantage decreases when kv/kh is 0.1. Thus the analysis from Figure
4.13 and Figure 4.14 show that although the performance of both
horizontal well injector and vertical well injector is sensitive to the change
in kh but only the performance of horizontal well injector is sensitive to a
change in kv In other words, the performance of horizontal well injector is
effected by changes In the permeability ratio. This is because the
horizontal well is located along the horizontal direction and the injected
water from the horizontal well has to travel in both vertical and horizontal
directions whereas the water from the vertical well has to travel primarily in
the horizontal direction only. Hence, the perfomnance of the horizontal well
is effected by the change in vertical pemneability, and the perfomnance of a
vertical well Injector is not effected by the change in vertical pemneability.
The effect of vertical permeability on the horizontal well injector will be
discussed in detail later in this Chapter.

4.2.2 Porosity
The high and low values of porosity used were 9% and 25%. Figure
4.15 shows the oil recovery as a function of pore volume injected. It can
be seen by looking at the oil recovery response that the HW-configuration
is better than the VW-configuration for the two values of porosity. The oil
recovery for HW-configuration or VW-configuration, individually, does not
show any significant change with an increase in porosity. It is expected
that the original oil in place would increase with an increase in porosity but
the fractional oil recovery as a function of fractional pore volume injected
should remain the same. The slight difference in the oil recovery could be

62
due to the numerical enror. This slight difference in terms of relative error
is about 0.7%.

4.2.3 Rock Compressibilitv


The rock compressibility was varied from 2E-06 psi "^ to 25E-06
psi "\ The oil recovery response for the two configurations is shown in
Figure 4.16. It can be seen that the HW-configuration shows more oil
recovery as compared to the VW-configuration. Figure 4.16 does not
show a significant change in the oil recovery from the two configurations,
individually, with a change in rock compressibility. The higher rock
compressibility of a reservoir acts as a drive mechanism when the
reservoir pressure decreases, but in this study the overall pressure in the
reservoir is increasing due to injection. Hence, the higher compressibility
does not show any significant affect on the oil recovery.

4.2.4 Relative Permeability


Two sets of oil and water relative pemneability curves were used to
perfomn the sensitivity analysis. These two sets, refenred to as case 1 and
case 2, are shown in Figure 4.17 and Figure 4.18. These two data sets
were chosen because the rate of increase or decrease in the relative
permeability (of oil and water) due to a change in water saturation is
different for both of them. The oil recovery as a function of pore volume
injected, for the two relative pemneability cases, is plotted in Figure 4.19. It
can be seen that more oil is recovered when the relative pemneability
curves of case 2 are used. This is because at any given value of water
saturation, the relative permeability to both oil and water is higher for case
2 than case 1. Also, the end point relative pemneability for oil and water is
also higher for case 2 as compared to case 1. Figure 4.19 shows that HW-

63
configuration has a similar advantage over the VW-configuration for both
sets of relative permeability curves.

4.2.5 Capillary Pressure


Two different capillary pressure data sets were used to see the
effect on the performance of horizontal well injector and vertical well
injector. These data sets are plotted in Figure 4.20 (case 1) and Figure
4.21 (case 2). In case 1 the capillary pressure reduces from 1.5 psi to 0
psi as the water saturation changes from 0.2 to 0.9, and in case 2 the
capillary pressure varies from 95 psi to 0 psi as the water saturation
increases from 0.2 to 0.9. It can be seen in Figure 4.22 that the oil
recovery for case 2 is slightly more than the oil recovery for case 1 for both
configurations. Furthermore, it can also be seen by the small difference in
oil recovery for the two Pc curves that change in capillary pressure does
not have a significant effect on the results. It was also mentioned in the
literature,'^° that Pc usually has minor effects on the perfomnance of a
homogeneous (non-fractured) reservoir. Pc can be a dominant force in
fractured reservoirs and in reservoirs in which high permeability layers are
in good communication with low pemneability layers.

4.2.6 Formation Thickness


The formation thickness was varied from 21 feet to 105 feet to see
its effect on the perfomnance of horizontal and vertical well injectors. The
oil recovery versus pore volume injected is plotted in Figure 4.23 for the
two configurations. The oil recovery response shows that although the
HW-configuration is advantageous compared to the VW-configuration for
the two values of fomnation thickness, this advantage is less in the case of
higher value of fomnation thickness. This shows that the advantage of

64
horizontal well injector decreases as the fomnation thickness increases.
This will be discussed in detail later in this chapter.

4.2.7 Oil Viscosity


The oil viscosity values of 0.8 cp and 40 cp were used to check the
sensitivity of the results due to this parameter. Figure 4.24 shows that the
oil recovery response for HW-configuration is better than that of VW-
configuration for both values of oil viscosity. It Is also evident from Figure
4.24 that when the viscosity is increased from 0.8 cp to 40 cp, the oil
recovery decreases. The water velocity is greater than the oil velocity due
to higher value of oil viscosity. As the oil viscosity increases, the oil
velocity further decreases and more oil is bypassed by the injected water.
Hence, with an increase in oil viscosity, less oil is produced.

4.2.8 Oil Density


When the oil density is varied from 40 Ib/fl^ to 58 Ib/ft^, the change in
the oil recovery response as a function of pore volume injected is shown in
Figure 4.25. It can be seen in this figure that the horizontal well injector
shows better oil recovery than the vertical well injector for both values of oil
density, and this advantage is similar for both values of oil density. The
lower oil density means relatively higher water potential (O) In the vertical
direction as compared to oil potential (O), which causes more water to flow
towards the bottom layer while moving towards the producer. Hence,
some of the oil, which is in upper layers, is bypassed by water.
Conversely, when the oil density is higher and comparable in magnitude to
the water density, the water potential In the vertical direction is not very
high as compared to the oil potential. This will decrease the more
pronounced movement of water in lower layers and injected water will

65
displace more oil in the upper layers as compared to when oil density was
lower.

4.2.9 Oil Formation Volume Factor


Figure 4.26 shows the two curves used to check the sensitivity of the
results due to a change in the oil formation volume factor (Bo). Bo is varied
form 1.0 RB/STB to 0.965 RB/STB for case 1 and from 1.0 RB/STB to 0.92
RB/STB for case 2, when the pressure increases from 0 psi to 8000 psi.
As it is assumed in this study that there is no gas present in the reservoir,
the Bo values are less than 1 and are varied in very small increments with
the increase in pressure. The oil recovery as a function of pore volume
injected for these two cases is shown in Figure 4.27. It is evident from this
figure that the oil recovery is better for HW-configuration as compared to
VW-configuration for the two cases of Bo. Furthermore, different values of
Bo also have an affect on b' (slope of curves in Figure 4.26), which is only
present in the accumulation term in the flow equations, b' is a very small
number (of the order of 1E-5 or less) and it is evident from equation (3.25)
(Chapter 3) that a variation in b' does not have a pronounced affect on the
accumulation term. The oil recovery for case 2 is slightly more as
compared to the oil recovery for case 1. The Bo values in case 1 are
slightly higher than in case 2 for all pressures. This slight difference in Bo
should not have a pronounced affect on the fractional oil recovery. The
small increase in oil recovery for case 2 could be due to numerical en^or.
This small increase in oil recovery in terms of relative en-or is about 0.7%.

66
4.2.10 Summary of Sensitivity Analysis
It is evident from Figure 4.13 through Figure 4.27 (discussed eariier)
that a horizontal well injector is always more advantageous than the
vertical well injector for the range of values of different parameters
analyzed above. Furthermore, the advantage of a horizontal well injector
over the vertical well injector remains almost of the same magnitude when
a parameter is varied from one value to another. The only exception to
this similar advantage is the change in pemneability ratio and fomnation
thickness. The horizontal well injector's advantage decreases with a
decrease in the permeability ratio and an increase in formation thickness.
Due to the dependence of the performance of horizontal well injector on
permeability ratio and formation thickness, the potential of horizontal well
injectors for different values of these parameters will be explored in section
4.3. Now that the sensitivity analyses have been discussed, further results
generated by the simulator to see the potential of horizontal well injector in
waterflooding can be presented.

4.3 Potential of Horizontal Well Injectors in Waterflooding


The gridblock dimensions are given in Table 4.6. The x and y
gridblock dimensions are kept constant at 70 feet and the number of
gridblocks in the x and y directions is also same, which is equal to 9. The
gridblock dimensions in the z-direction (gridblock heights) are varied from
7 feet to 42 feet to see the affect of fomnation thickness on the
performance of horizontal well injectors in an oil reservoir. The number of
girdblocks in the z-direction is 3, so the total fomnation thickness would be
three times the block height. Therefore, the variation in gridblock height
from 7 feet to 42 feet means that the formation thickness is varied from 21
feet to 126 feet.

67
Table 4.6 also shows the rock properties. Reservoir permeability in
the x-direction (kx) and y-direction (ky) remained constant (200 md) and the
permeability in the z-direction (kz) varied from (10 md to 100 md) in order
to see the performance of a horizontal well injector for different
permeability ratios (kv/kh). As mentioned eariier the pemneability in the x-
direction and y-directlon will be referred to as horizontal permeability (kh)
and the permeability in the z-direction will be called vertical permeability
(kv). Hence kv/kh would be the ratio of vertical permeability to horizontal
permeability. As kh is kept constant, the change in the permeability ratio is
only due to the change in kv values. The effect of vertical permeability will
be discussed later in the discussion that follows.
The pressure dependent fluid property data (Bo, |io) is given in Table
4.4 and the static fluid property data are listed in Table 4.5. The water and
oil relative permeability curves are shown in Figure 4.11, and the capillary
pressure data can be seen in Figure 4.12. This is the same data which
was common for all sensitivity analyses. Figures 4.6 through 4.9
(mentioned eariier) show the two grid configurations.
The pemneability of the reservoir plays an important role in the
performance of a horizontal well injector as compared to a vertical well
injector. This is because the performance of a horizontal well injector is
highly dependent on the permeability of the reservoir in the z-direction
(vertical permeability, kv), which is usually lower than the reservoir
permeability in the x and y directions (horizontal pemneability, kh). This
phenomenon can be explained by looking at the side views (x-z direction)
of the two configurations discussed eariier (Figure 4.6 through Figure 4.9).
It can be seen in Figure 4.6 and Figure 4.8 that the wellbore for a
horizontal well injector is parallel to the y-direction. For the horizontal well
injector to be effective, the injected water has to invade the reservoir in

68
both x and z directions and displace oil towards the producer in the x-
dlrection. On the contrary, the wellbore for the vertical well injector is
parallel to the z-direction, and the water from the vertical well injector has
to invade the region in the x and y directions and then displace oil in the x-
direction towards the producer (Figure 4.7 and Figure 4.9).
The performance of the horizontal well injector based on oil recovery
as a function of pore volume injected will be discussed next.

4.3.1 Results Based on Oil Recovery


The oil recovery as a function of the pore volume injected for a range
of gridblock heights is shown in Figure 4.28 through Figure 4.33. Each of
these figures shows results for both HW-configuration and VW-
configuration for different permeability ratios. As discussed eariier, the
flow from a vertical well injector is primarily in the horizontal direction (x-
and y-directlon) only so the change in vertical permeability should not
affect the performance of a vertical well injector. Figure 4.28 shows the oil
recovery as a function of the pore volume injected for a gridblock height of
7 feet (formation thickness is equal to 21 feet). It can be seen in Figure
4.28 that the oil recovery from VW-configuration does not show any
change for different permeability ratios. The oil production from HW-
configuration, on the other hand, does change for different pemneability
ratios. This is due to the fact that the performance of a horizontal well
injector in a reservoir depends upon permeability in both the horizontal and
vertical direction.
It should be noted that with an increase in the permeability ratio
(Increase in vertical permeability because horizontal permeability is kept
constant), the perfomnance of the horizontal well injector becomes better.
This is because the injected water from the horizontal well can Invade

69
further and faster in the z-direction with the increase in vertical permeability
and then can displace more oil towards the producer. This is evident from
the trend of oil recovered from HW-configuration for different pemneability
ratios in Figure 4.28. Furthermore, it is also clear from this figure that the
performance of a horizontal well injector is better than that of the vertical
well Injector because of the fact that more oil is produced when a
horizontal well is used as an injector. Therefore, a horizontal well injector
proved to cause higher recovery than a vertical well injector in this case.
It should be noted that the length of the horizontal well injector is
unchanged for all of the different gridblock heights, because the horizontal
well is located along the y-direction in the gridblocks, and the y-dimension
of the gridblocks is kept constant. The length of the vertical well injector
increases if the gridblock height is increased because the vertical well is
located along the z-direction and is fully penetrating the three gridblocks as
shown in Figure 4.9 eariier. The length of the vertical or horizontal well in
the formation is also called area of contact. Therefore, in this study, the
area of contact with the formation increases for the vertical well injector as
the gridblock height is increased, whereas the area of contact for the
horizontal well injector remains unchanged.
It can be infenred from the discussion about the area of contact in
last paragraph that the advantage of the horizontal well injector over a
vertical well injector, as evident in Figure 4.28, starts to diminish as the
gridblock height increases (formation thickness becomes larger). Figures
4.29-4.33 show the oil recovery as a function of pore volume injected for
gridblock heights of 14 feet, 21 feet, 28 feet, 35 feet, and 42 feet,
respectively. Figure 4.29 (where the gridblock height is 14 feet) shows that
the oil produced from HW-configuration is greater as compared to the oil
produced from VW-configuration. It can be seen that this difference

70
(between the responses for the two configurations) in oil produced in
Figure 4.29 is slightly less than the difference in oil produced in Figure 4.28
(where the gridblock height is 7 feet). This confimns the diminishing
advantage of horizontal well Injector over a vertical well injector for
increased gridblock height. Comparison of Figure 4.30 through Figure
4.33 also makes it clear that the difference in oil produced between HW-
configuration and VW-configuration decreases as the gridblock height is
increased from 21 feet to 42 feet.
It is clear form Figure 4.28 through Figure 4.33, that after injection of
0.5 pore volume, the fraction of original oil produced is more than 0.4 for
both horizontal well injector and vertical well injector cases. This fractional
amount of oil produced increases to a little over 0.5 for VW-configuration
and to a little over 0.55 for HW-configuration when the pore volume
injected Is 1.0. The fractional oil recovery can be converted into the
percentage of recoverable oil produced with the help of residual oil
saturation. The residual oil saturation used In this study is 10% of the pore
volume and connate water is 20% of the pore volume, which leaves the
recoverable oil equal to 70% of the pore volume. Hence, the maximum
recoverable oil equal to 87.5% of the original oil in place. Therefore, more
than 45% of the recoverable oil is produced with the injection of 0.5 pore
volume and more than 57% of the recoverable oil is produced when pore
volume injected is 1.0 for both configurations.
The advantage of a horizontal well injector can be further explained
by tabulating the oil recovered (shown in Figure 4.28 for h=7 feet) in Table
4.7. This table shows that 48% of the recoverable oil Is produced in case
of VW-configuration with the injection of 0.5 pore volume and 6 1 % of the
recoverable oil is produced with the injection of 1.0 pore volume (kv/kh =
0.05-0.5). Also, 52% and 55% of the recoverable oil is produced for kv/kh

71
values of 0.05 and 0.5, respectively, when 0.5 pore volume is injected in
the case of HW-configuration. The oil produced in case of HW-
configuration increases to 65% and 69% for for kv/kh values of 0.05 and
0.5, respectively, when the pore volume injected is 1.0. Furthemnore, the
curves in Figures 4.28 through Figure 4.33 tend to flatten when the pore
volume injected increases above 0.5, which also show that less amount of
oil is being produced per amount of water injection as pore volume injected
increases above 0.5. A comparison, similar to the one shown in Table 4.7,
of the two configurations Is provided in Table 4.8 from the results shown in
Figure 4.33 (h=42 feet). It is listed in Table 4.8 that as kJK increases from
0.05 to 0.5 for 0.5 pore volume injected, the amount of recoverable oil
produced increases from 5 1 % to 54% for HW-configuration, whereas the
oil recovered for VW-configuration remains 49%, thereby, showing the
increasing advantage of horizontal well injector as kv/kh increases.
In the following text the performance of horizontal well injector and
the performance of vertical well injector will be referred to as the two
performances. To further see the comparison of the two performances in
different formation thicknesses, oil recovery is plotted against pore volume
injected for different gridblock heights. Figure 4.34 compares the two
performances for a gridblock height 7 feet and 14 feet, where kv/kh value is
0.05. Figure 4.35 shows a comparison of the two performances for a kv/kh
value of 0.5.
A careful examination of Figure 4.34 and Figure 4.35 clarifies that as
the formation thickness increases the advantage of horizontal well Injector
over the vertical well injector decreases. This decrease in advantage of
horizontal well injector is more pronounced in Figure 4.36 through Figure
4.39, where the two perfomnances for a gridblock height of 7 feet are

72
compared with the two performances for gridblock height of 28 feet and 42
feet.
To explain the decrease in advantage of a horizontal well injector
with the increase in formation thickness in more detail, the two extreme
cases of formation thickness, used in this study, can be examined. Figure
4.38 and Figure 4.39 compare the oil recovery for a gridblock height of 7
feet with the oil recovery for a gridblock height of 42 feet for kv/kh values of
0.05 and 0.5, respectively. The oil recovery for the HW-configuration and
VW-configuration, for a gridblock height of 7 feet, kv/kh of 0.05, and pore
volume injected of 1.0, are about 0.58 and 0.54, respectively (Figure 4.38),
which means that HW-configuration has an advantage of 0.04 in temns of
fractional oil recovery. This advantage decreases to about 0.016 in temns
of fractional oil recovery when the gridblock height is increased to 42 feet
(Figure 4.38). Furthermore, the advantage in oil recovery from HW-
configuration as compared to the oil recovery from VW-configuration for a
gridblock height of 7 feet and kv/kh of 0.5 is about 0.07 (for 1.0 pore volume
injected), which decreases to about 0.04 when the gridblock height is
increased to 42 feet. This explanation also clarifies the conclusion made
eariier that as kv/kh increases the advantage of horizontal well injector over
a vertical well injector Increases.
Summarizing the discussion in the last few paragraphs, the
performance of horizontal well injector decreased with the increase in
formation thickness. Furthermore, the advantage of a horizontal well
injector over a vertical well injector increases with the increase in
permeability ratio (vertical permeability).

73
4.3.2 Results Based on Water Cut
Another parameter used to compare the performance of a horizontal
well injector with the performance of a vertical well injector is water cut.
The ratio of total water produced to the total fluid (oil plus water) produced
is known as water cut. Water cut as a function of pore volume injected for
a range of gridblock heights Is plotted in Figure 4.40 through Figure 4.45.
Let us first examine the water cut response for kv/kh of 0.05 in Figure
4.40. It can be seen in this figure that water cut in the horizontal well
injector case is less than the water cut for vertical well injector case until
0.5 pore volume is injected. As mentioned previously, about 50% of the
recoverable oil is produced while 0.5 pore volume is being injected. This
can be further explained by showing the oil and water production rates as
a function of pore volume injected. Figure 4.46 shows the decrease in oil
production rate and an increase in water production rates as the amount of
pore volume injected increases, thereby increasing the water cut.
It is also evident from Figure 4.46 that between pore volume injected
values of 0.6 and 0.7 the oil production rate for HW-configuration becomes
equal to the oil production rate for VW-configuration. As the value of pore
volume injected Increases beyond 0.7, there is a crossover and the oil
production rate for HW-configuration becomes slightly less than the oil
production rate for VW-configuration. Conversely, the water production
rate for HW-configuration increases after the pore volume injected value
goes above 0.7. This change in the oil and water production rates also
explains the increase in water cut for HW-configuration as compared to
VW-configuration when the pore volume injected value is above 0.7
(Figure 4.40).
This crossover in the water cut response for HW-configuration and
VW-configuration in the Figure 4.40 through Figure 4.45 suggest that the

74
oil recovery curves in Figure 4.28-4.33 for the two configurations should
also come closer (or show a crossover). Let us take the example of Figure
4.40, where the gridblock height Is 7 feet and HW-configuratlon response
for kv/kh of 0.05 shows a crossover (at pore volume injected=0.7). Now the
similar case (gridblock height =7 feet, kv/kh = 0.05) for oil production
response would be Figure 4.28, which does not show that the oil
recoveries for the two configurations approach each other at pore volume
injected value of 0.7. The reason for this will be discussed next.
Figure 4.46 shows that the oil production rate for HW-configuration
becomes less than the oil production rate for VW-configuration after pore
volume injected value goes above 0.7. The difference in the oil production
rates (for the two configurations) above the pore volume injected value of
0.7 is very small (under 10 STB/Day). Also, the value of oil production
rates has decreased to about 80 STB/Day at a pore volume injected value
of 0.7 as compared to 265 STB/Day at a pore volume injected value of 0.3
and 145 STB/Day at a pore volume injected value of 0.5. The effect of the
small difference (under 10 STB/Day) in the oil production rates of the two
configurations above a pore volume injected value of 0.7 is very small as
compared to the original oil in place. Furthemnore, the oil production rates
have also decreased to about 80 STB/Day when the pore volume injected
is more than 0.7. Therefore, the oil recovery from HW-configuration does
not approach the oil recovery from VW-configuration when the fraction of
original oil in place is plotted as a function of pore volume injected.
The crossover in Figure 4.40 through Figure 4.45 is only present for
lower values of pemneability ratio (kv/kh=0.05, 0.1). It was discussed eariier
that water from the horizontal well injector can invade further vertically into
the reservoir with an increase in kv/kh (means an increase In vertical
permeability). This further invasion would push more oil towards the

75
producer and hence more oil will be produced, and the horizontal well
injector becomes more advantageous. To further explain this increasing
advantage of a horizontal well injector with an increase in kv/kh, the water
cut values from Figure 4.40 (h = 7 feet) are tabulated in Table 4.9. This
table shows that the water cut for the VW-configuration is 0.76 and the
water cut for HW-configuration is 0.74 (kv/kh = 0.05) and 0.705 (kv/kh = 0.5)
when pore volume injected Is 0.5. Also the water cut for VW-configuration
Increases to 0.91 and for HW-configuration increases to 0.92 (kv/kh = 0.05)
and 0.875 (kv/kh = 0.5) when the pore volume injected is 1.0. These water
cut numbers complement the above statement, that the horizontal well
injector becomes advantageous with the increase in kv/kh. A similar
comparison of water cut for the two configurations is provide in Table 4.10
from the results shown in Figure 4.45 (h=42 feet). Table 4.10 also shows
that as kv/kh increases, the HW-configuration produces less water.
Summarizing the results presented in this chapter, the horizontal
well injector is advantageous in waterflooding operations as compared to a
vertical well injector. This advantage of a horizontal well decreases if the
vertical permeability is not very favorable for the injected water (from the
horizontal well injector) to invade the region in the vertical direction of the
reservoir. In other words, this advantage decreases as kv/kh decreases.
Furthermore, in thicker formations the advantage of a horizontal well
injector over a vertical well injector is not very pronounced. Specifically,
the results showed that when one pore volume of water was Injected, the
amount of recoverable oil produced increased from 6 1 % for vertical well
injector case to 65% for horizontal well injector case for kv/kh value of 0.05
and to 69% for kv/kh value of 0.5 for a gridblock height of 7 feet. This
increase in oil produced in the case of horizontal well injector was less
significant when the gridblock height was increased. For a gridblock height

76
of 42 feet, the amount of recoverable oil produced increased from 63% for
vertical well injector case to 64% for horizontal well Injector case for kv/kh
value of 0.05 to 67% for kv/kh value of 0.5, when one pore volume of water
is injected.

77
Table 4.1: Data for Validation of Simulation model by Analytical Solution
Production Rate, STB/Day 500
Injection Rate, STB/Day 500
Porosity, fraction 25
Permeability, md 100
Viscosity, cp 1
Density, Ib/cft 62.4
Formation Volume Factor, RB/STB 1
Length of the Reservoir, ft 2000
Thickness, ft 50
Wellbore Radius, ft 0.5
Initial Pressure, psi 5000

Table 4.2: Data for Validation of Simulation Model by Material Balance


Pressure, psi Bo, RB/STB
3112 1.4235
2800 1.4290
2400 1.4370
2000 1.4446
1725 1 -4509

78
Table 4.3: Grid Dimensions, Well Flow Rates, and Rock Properties Used

Number of gridblocks in x-direction 9


Number of gridblocks in y-direction 9
Number of gridblocks in z-direction 3
Gridblock dimensions in x-direction, ft 70
Gridblock dimensions in y-direction, ft 70
Gridblock dimensions in z-direction (height), ft 28
Permeability in x-direction (kx), md 200
Permeability in y-direction (ky), md 200
Permeability in z-direction (kz), md 60
Pemneability Ratio (kv/kh), md/md 0.3
Porosity ((j)), fraction 0.12
Rock compressibility, psi"^ 3.50E-6.0
Total injection rate, STB/Day 550
Total production rate, STB/Day 550

79
Table 4.4: Pressure Dependent Variables
Pressure, psi Bo, RB/STB ^o. cp
0 1 2.5
1000 0.996 2.5
2000 0.991 2.5
3000 0.987 2.5
4000 0.983 2.5
5000 0.978 2.5
6000 0.973 2.5
7000 0.969 2.5
8000 0.965 2.5

Table 4.5: Fluid Properties


Water Formation Volume Factor (Bw), RB/STB 1
Water compressibility, psi 3.00E-6.0

Water viscosity (pw), cp 0.6

Water density, Ib/ft^ 62.43

Oil density, Ib/ff^ 51.55

80
Table 4.6: Grid Dimensions, Well Flow Rates, and Rock Properties Used
for Simulation Results
Number of gridblocks in x-direction 9
Number of gridblocks in y-direction 9
Number of gridblocks in z-direction 3
Gridblock dimensions in x-direction, ft 70
Gridblock dimensions in y-direction, ft 70
Gridblock dimensions in z-direction (height), ft 7,14,21,28,35,42
Permeability in x-direction (kx), md 200
Permeability in y-direction (ky), md 200
Permeability in z-direction (kz), md 10,20,40,60,80,100
Permeability Ratio (kv/kh), md/md 0.05,0.1,0.2,0.3,0.4,0.5
Porosity (((>), fraction 0.12
Rock compressibility, psi'^ 3.50E-6.0
Total injection rate, STB/Day 550
Total production rate, STB/Day 550

81
Table 4.7: Oil Recovered Versus Pore Volume Injected for h=7 ft.
Well Oil Recovered Oil Recovered
(h = 7 feet) (PV Injected = 0.5) (PV Injected = 1.0)
Vertical Well, kv/kh=0.05-0.5 48% 61%
Horizontal Well, kv/kh=0.05 52% 65%
Horizontal Well, kv/kh=0.5 55% 69%

Table 4.8: Oil Recovered Versus Pore Volume Injected for h=42 ft.
Well Oil Recovered Oil Recovered
(h = 42 feet) (PV Injected = 0.5) (PV Injected = 1.0)
Vertical Well, kv/kh=0.05-0.5 49% 63%
Horizontal Well, kv/kh=0.05 51% 64%
Horizontal Well, kv/kh=0.5 54% 67%

Table 4.9: Water Cut as a Function of Pore Volume Injected for h=7 ft.
Well Water Cut Water Cut
(h = 7 feet) (PV Injected = 0.5) (PV Injected = 1.0)
Vertical Well, kv/kh=0.05-0.5 0.76 0.91
Horizontal Well, kv/kh=0.05 0.74 0.92
Horizontal Well, kv/kh=0.5 0.705 0.875

Table 4.10: Water Cut as a Function of Pore Volume Injected for h=42 ft.
Well Water Cut Water Cut
(h = 42 feet) (PV Injected = 0.5) (PV Injected = 1.0)
Vertical Well, kv/kh=0.05-0.5 0.77 0.91
Horizontal Well, kv/kh=0.05 0.767 0.92
Horizontal Well, kv/kh=0.5 0.73 0.875

82
Closed

Closed Open

Closed

Figure 4.1: Physical Model for the Analytical Solution

o •«
e

Figure 4.2: Grid Model Used for the Simulator


Validation by Analytical Solution

83
d

00 CO
^^
d CQ
G
0)
E
c H
cu o
[^ r^
o

ula
lyti
CO
d yi
(U
E c
<c CO o
'^2

1•
D
CO
13
o
d
s
5;
CD

in < <
c
d S c
CD
CD
c
CO
D

Its
^
D
d (/)
0)
a:
c
q
'•^S

CO CD
T3
CiZZ
d CD
>
CO
"^
0)
CM ^
3
d

I —r~
CO CO CM oo oo CO

CM
od od od 1^ 1^

84
B
"(D
Q
0)
E
1-
0)
CD

o
CO

s
< CD
c
<
c
CD
O)
c
"w
D
0)
0)

CD
•g
ID
>

0)

O)

85
o
o
CO
CO

o
o

0)
o o
o
Oi c
CO

rial Bal
CM
(U
o
c
CD
CD O 0)
CD O
O
h«. CD

etric M
CO CM
L_
"co
0) 3
CD E
2 CO O
E
1
O .-^^ 3
D lO (0
CM C3.
>
o
(D

ing
sur
(O (O
0) 3
O
CO
a. (/)

Validat ion Result


CM

100
CM

O
O
m
gure4

O U-
O
h«.

o
o
—r- —r-
o oo o
o o
o
o
o
O
o oin o o o
O
o
o o m o m
o CM CM
CO
(aiS) peonpojd 110 eA!;B|niuno

86
•> X

t
y

Figure 4.6: HW-configuration (Top View)

87
> X

o- o

Figure 4.7: VW-configuration (Top View)

88
> X

Figure 4.8: HW-configuration (Side View)

> X

t
z

Figure 4.9: VW-configuration (Side View)

89
• X

Figure 4.10: x, y, and z Direction Orientation

90
d
i2
"5
CO
0)
00
a:
d q

E
d CO
0)

CO
T3
d 0)
CO

CO
0)
d CO
O

(D
o (D

0)
Q.
CO 0)
d JD
0)
Q:
CM
d

O)
U-

A}|||qB9Ujjad 9A!;e|9^

91
d

00 CO

d 3
CO
Q)
Q:
c
1^ q
'•^

(0
o 3
E
CO
CO <u
d • ^
k.
o
«*-
•o
0)
in ^ CO
d CO 3
0)
^
3
o
Tj- 0)
u.
3
d CO
CO
<D
k.
CL
CO ^

o i5
*Q.
CD
O
CM CM

o •
"*
<D
k-
3
O)
^. L.

(!sd) »d

92
d
•o
E E
«w> - n O
00

ute emrieab
CO II CO II d
> ^ > -^

" ^ -O" " ^ T3" Q.


•g E? E r^
Eo Eo o
o o o o
^ CO T - CO ^ o
II Ii II II CO
o n
^ ^ JM^ ^ CO 7S <
CD k-
L.
o »^ *^ o
•o CO
0) CO
ts _>»

je
1 m CD

o ^ < c
0) >»
E • >

nsi
"<^. o
d > 0)
CO
O »^

13: Res tso


Q.
CO 3
o

CM Tf
d <D

CJ)

in CO CM
CO
d d d d d d
(uojpBJi) AjaAooay no

93
0)

abi tyR tio of 0. and 0.3.


o

00
^^
o

1^ CO
d
<^^^ ""~~

ion
CO "C 0)
d CD fe
(1)
"O Q.
CD k.
ts
0)
£
m CO
o c • CO

0)
CD
E c

.ti
0) .>
o
0.
CO
c
CO C0)O

esults of
d

CM OL
o

"•" 3
CD
o U-

r^ CO in CO CM
d d d o o d
(uojpBJi) Aj9Aooay no

94
CJ)

alue of 0.09 nd 0.25.


o

00 CO
o
o> m O)
m
O CM o CM 1^ CO
oII oIi d
^^^ ^^^^
> » >>
^•^ ^•^
CO CO
O
c >
k. o k- o >»

oro sit
o
Q. o
Q.
CO T5
CO
o
T3 CL
<D u.
ts 5CO
in 0)

ysi
^
o
(D ^^
CD
E C
^ <

eVo

sit ivity
0.4
k_
o
Q. c
CO C0)
O
t^-

esults 0
o

CM Q:
O
re 4.15:

"*~ 3
C3)
o LL

r^ CO in CO CM
d d d d d d
(uojpBj;) AjaAooay no

95
CO
T - Q.
^~
CO
LU
m
Oi CM
o T3
cD
C
.-:=- CO : : r ^ CO CO
CO Q . (O Q . Q.
^."^
oo ^
o CO
^^co ^^co
LU m LU m
CM CM CM CM
1^ CO

alue
d
CO CO ( O CO
CO CO < 0 CO
c >

a> £ a> E o
'^^

CO ^—
Q. Q. Q . Q. o .Q
o 2
•^- CO
o o o o CO
"O 0)
oooo CD ll

ect
Q.
-^ -^ ^ ^
s^ s^ « ^ in • " ^ ^
E
q o o o o s= Oo
(nc^otia: 0) ^

e Volum

gure 4.16: Res ts Of Se sitivity A alysi for Roc


0.4
k. CO
o
CL
CO
c
d

CM
c
o

^- 3
d

o
CO in CO CM
d d d d o d
(uojpBJi) Aj9Aooay no

96
O)
d , ^
^r—
0)
CO
00 CO
d o^
>*
'>
Vrf
CO
r^
d c
CCD
O

."ti
CO !5
CO
d 0)
§
CD
0-
in ^ CD
d CO >
•4^

CD
CD
a:
"^ o
t^-
d •o
CD
CO
3
CO
CO CD
d ^
3
o
r^
CM •
• M-
o 0)
^-
3
O)

X;inqB8Uuaci aAj^Biay

97
ay
d
CM
03
CO
00
d
• >

CO
d CD
CO

CO .a
o CD
<D

CD
CL

d CO
CD
ct:
k-

d T3
CD
CO
3
CO
CO CD
d 3
o
od
CM

d
3
O)
U-

C3> 00 r^ CO m -^ CO CM T-

O O o o o o o o o
AiniqBauuad 9A!;B|ay

98
oo
-^
h".
^
<Ji Tf

Given i Figures
o

00
c
o
T - CM 'T- CM
CD CD CD CD

s s ss
CO CO ( O CO

11 11 II 11
1^ CO
CD
o Q
^-^ >*
c
o —

cti
CO CO
o 2 0)
«»- ^
T3 0)
CD n
ts CD
CD >
m _C
d CD
(D 0)
b Q:
3 L_

ore Vol

lysis fo
0.4
CL CD

CO < c

ensitivlty
d

CM CO
o
Figur e4.19:R esults of

—r- o
CO
d d o o o o
(uojpBJj) Aj9Aooay no

99
d

oo
d
^.^
• "

1^ CD
CO
d S
CD
^^
CD
CO Q
^
d • >

'•^
CO
(~
CD
CO
d CO CD
k.
3
CO
CO
0
k.
• CL
o ^
i5
'5.
CD
CO O
••
d o
CM
-^
0)
i—
CM 3
d g)

f o
o

(isd) »d

100
d

oo
d

CM
r^ <D

(cas
d

CO
^^^
CD
CO Q
o >s

CO

m ^ CD
d CO (/J
CD
k.
3
CO
CO
CD

0.4
k_
Q.
^
jO
Q.
CO
CO
o
d T~
CM
• ^

(D
CM k.
3
C3)
d

o
o o o o o o
in o o o
CJ) 00 CO CO CM

(isd) =d

101
CM
•^

CJ)
6
CM

gures 4
o

T - CM - « - CM oo Li-

iven in
CD CD CD CD o
CO CO t o CO

s s ss
X " " II
j r cT'cT'c? r^ O
Q- Q. Q. CL

ata
d
^^ Q
c
o £

cti
3
CO CO
d CD CO
^ _ i - ^
^
•o Q.

Cte
CO
m Q)
• ^ ^ ^
• ^^
o C Q.
^^^ CD
<D C)
h
3 O
k_

^^»
O •^
^r > CO
d CO
0)
k.

o CO
CL
<
CO ^

esults 0 Sensitlvi
o

CM H-
o

T -

o
a:
ure 4.22:

o Oi
CO in CO CM
d o o o o o
(uojpBj^) Aj9A009y no

102
'^
11

< A
4-*
•1
^t .1 CD
• >\ CD
CJ)
• 1
m
• 1
d 0
^—
11 •0
n<
a
'\
•! JC _ c
c
CD
*i m JC m 00
° ^ "'A T- o
CM ' i -
•- o
CM <<-
d
CD

°%
^m A1 ji^jUJivJL ^ • ~

l o ^ ' c O * CO CO CM
D< i\ CO CO CO CO i^_
CD CD <D CD 0
rs.
^c Jil
c Jc^ c^
Q4 *\
^< A•\
«4%
CO
CO
o o o o d CD
ss>>
•^ ^-^ .«-- * - ^~v ^
D5 .\ c c c c c g

(fractic
o o o o
•^-> '*-* ^.rf .4.^
CO CD CO CO CO H
E§ g 1 d c
0
'.^
p p 0 0 CO
CD
< 'X
CM '\
m 0
X I > > CD

le Inj
Ll.
d
a <
5
fc
_3
CO
t*. *\ CO

ire Vo
< A

0.4
^< A»\ CD
c
a,< A* \ 0 <
9^. % \ a. >»
'>
CO V-
CO
^L * \ d
^4 •X c
CD
^3 _ %\
^ ^ j ^ »\
CO
«^
<3 __ \^L
<3 ^ %\
0
B1^ * X
CM CO
^li* ^"^
*v d 3CO
CD
^IX
^ <^
Figure 4.23:

<^

1 1 1 1
T \^
r^ CO in "^ CO CM IT- 0
a\ CD CD CD d d <3 •

(uojpBj;) Aj9A009y no

103
CJ)
v_^
a.
o
o
'^
"O
00
c

0.8 cp a
O

1^ «^-
o o

cosi
ion)
CO T3 CO
d 2
•*-
>
«^^
•o o
"^^

CD V-

in
TS
0)
sCO
• •^^^
o CO
c > N

ity nal
0)
E
3 <
'^
§ >
d CD
C ;ti
O CO
Q. c
CO C0)
O
«^
o o
CO

3
CO
CM 0)

d a:
••

CM
(D
k.
3
g)
Li.

(uojpBJi) ^laAooay no

104
f 40 Ib/cft and 58 1 eft
CJ) £1
o

1^??

0.8
^ ^ ^ .Q
o
"^
OO O 00
i n -^ i n
II 11 ii II
h-
•4^ •*^ 4-rf 4>rf o
CO CO CO CO o CD
CO

CD c
c c: c 3
CD CD CD
"O T3 T3 T3
'c' CD
o >
is

(ol
• ^ ^ m^^ a^MB

CO >*
o o o

nsit
CO
§5 d
1 1 T3 CD
CD Q
t) :s
m CD o
o s o
CD •^
CO
E CO
^ > M

re Vo

tivit Anal
0.4

o
Q.
CO CO
o c
CD
CO

CO
CM
3
d CO
CD
01
in
CM
'^
CD
k.
3
CJ)

CO in CO CM

o d o d o o
(uojpBJj) Aj9Aooay nO

105
ooo
<J)

o
o TS
o
o CO
LX.
00
CD
E
o _3
o O
o >
c
q
CO

000
O

CD

Of Oil
o CO
o j^-.
CO
o CI. >»
"*—' CO
p
k. c
3 <

ity
(0
o
o L_ >
o Q. ^^

Sensi
"^

k.
o o
H:::
o
o

4.26: Dat a Used


CO
000

CM

O <D
O k.
O 3
CJ)

in
98

<J) CD CO CM
CJ) CJ) C3) CJ) CJ) CJ) CJ) 5)
o o o o o o o o o
(aiS/Qd) JopBj 9Lun|0A uojiBUiiOd |!0

106
CO

ven in F ures 4.2


CJ) CJ)
o

00
CO
d

Fac rData
^ CM ' r - CM
CD CD CD CD
«2 S2 CO CO
^ CD CD CO
r^ o
0 c3 c5 o o
^ ^ o o c CD
OQ CQ m CQ o E
3
1 I CO TS
d CD O
1 ^ - >

ion
TJ
CD
^^
O CD

UJIO
Inje
0.5
(D u.
^—
E
3 o
O

o
• > a
0) CO

gure 4.27: Re ults of S nsitivity nalysi


Por
CO <
o

CM CD
o

V— CO
d

o
CO in CO CM
d d o o o o
(uojpBJj) Aj9A009y no

107
"*~ CD
CD
«•—
r^
II
.c
CJ) c

oswhe
o

oo "co

ability R
o
oI
m m
o T - CO in
cS CD CD c> o r^ CD

JU JU iU JUd
o §
II ^^^ CD
^ "^ ^ "^ > c U-
^ ^ ^ ^ 2^ o ^-<

cti
CO CD
• CD ^
I I I I o i=
•*-
CD
St:
TJ
n

Inj Cte
p
m CD •^
CD
d CO

ura on Re pon
<D
E
3 CO

re Vol
0.4
o • i *

CL
CO CJ)
**—
o
c

CM
TJ
d c
CD

OO
CM

CD
k_
3
O)
CO in CO CM UL
o d d d o d
(uojpBJj) AjaAooay no

108
CD
CD

•^>

CJ) ¥
d
0)
CO
o
^^
on CO

bility R
o

r^ CO
CD
o E
,.*-v CD
o Q.
•«-^

eren
tract
0.6
^-^ St:
TJ
CD Q
^
TS
in CD
• ^^^
5
o c
• ^ M 0)
CO
<D c

spo
lum
0)
o Q^

0.4
>
CD c
O g

rat
Q.
3
CO CJ)
o

CM
d

CJ)
CM

0)
k-
3
CO CM
g)
1^ CO m Ll.
d d o o o o
(uojpBJj) AjaAooay no

109
oswhe h=21 fee
CJ)
c
d

OO ^^
CO

bility R
o

CO
r^ <D
o
<-~N
0)
c Q-
o ^ k ^

eren
fract
0.6
'"**' St:
TJ
CD Q
k.
TS
in •
CD
^ H ^

CD
o ~~'^c CO
(D
c

spo
lum
CD
P

0.4
> tr
0) c
o q

rat
Q.
3
CO O)

V-confi
o

CM
O
and

o
CO
'^
CD
3
g)
Ll.
(uojpBJi) AjaAooay no

110
CD
S
oo
CM
11
CJ)
d 0)

CO
O
^irf
on
CO

bility R
o

h". CD
CD
o
^s. 0)
c CL
o ^ 1 ^

eren
fract
0.6
^-^ St:
TJ
CD G
k-
TS
m •C^^^
D 5
0)
o c ^ M M
CO
CD
c

spo
lum
o 0)

0.4
> Q:
0)
k. c
o
p

rat
Q.
3
CO CJ)
l^
o

CM
d

CO

2
3
CO in CO CM g)
d d o o o o
(uojpBJi) AjaAooay no

Ill
oswhe h=35 fee
CJ) c
d

00
"co

bility R
d

1^ CD
<D
d
y—^

c a>
o CL
• ^

eren
fract
0.6
^-^ St:
TJ
CD Q
TS o
in CD •^
«^^^
0)
o c
^^ CO
0) c
E o
ex
3 CO
O CD

0.4
> Od
CD
k. c
g
o
Q. CD
3
CO CD

y-confi
o

CM 5
O
and

CM
CO
"^
2
3
g)
ij.
(uojpBJi) AjaAooay no

112
• '•». \ - ^ (D
* 'o\ \
• •m \ CM
• 'Ol \ "^
**o',\ II
• 'o» I
CJ) c
1 d CD
^*
4- • R I m ^
•'Ol \
d1 CO
•'Ol \ in in g
•'o-, \ o -«- CO m o oo "co
•'Ol \
•'Ol \
d CD CD <D d d Od
jJL j i v JJL j ^ II
•'o^ \ f- ^3 ^3 JC f-
•'Ol \ .^ ^^f c J ^
^ ^
^
^ > ^
^
^ ^ ^ —,
^ ISIZ

•'Ol \
"^^
> > > > > ^
••Ol \ ^ ^ J^ ^ .^ CD
r^
•'Ol \
•'Ol \ § § § § § d E

tPer
1111 r
o
CO tj
CO 2
d k- CD
»-
t-V

for Dil
ected
V.

•f
••
in
v.^\ d ]c* 0)
CO
\Vv 0)
E o
3 CO
o CD
d > Od
CD c
•'p\ g
o
0. 2

lure 4.33: HW- and VW-configi


CO 3
% \
d

CM
d

1 r- 1 1 1 1
f- o
1 o;
r^ co m ^ CO CM •- O U.
• • '
• • o o o
d o o
d
(uojpBJi) AjaAooay nO

113
0) it^
o

and 14
1

r^
II F*.
S ^ 00 «i
M
¥ ¥

h=7
^ -C
II o
•^
>< in in in in c
M
o o o o CD
>< dII d dII d c
1
I
M
^' X
II
ji^ ^ ^ ^ ^ ^ ^ - k r^ ^
1 ^ f** f ^Z £1 o in
\
\
X ^
^^
^ ^ -^ o
> "^ > "> ..-^s
1« ^M j ^
^i_i-^
.^ ^
^i__i-^ ^
ji^
1 - ^ CL oII
11 ><t g
§ ^ § § ^

\ ^' X 5 I >
CO
o
TS
2
• > •
Jd
\ ^ H— ^
4 X
TJ •^ o
A ^
CD 0)
TS CO
in CD
o
Q.
o c*
^^" CO
CD 0)
E tr
o

34: HW andVW-co nfi urati


reV(
d
o en
Q.

CO
o

CM 1

^ -^
gure

T
CO in CO CM
d d d d o o
(uojpBjj) AjdAooay no

114
CJ)
d
d

and 14
!**>-
^r
T"
^
1^
II II II II
• C -C -C •C
00 «i
m in m in

5wh enh=7
oII oII oII oII o
y"^^
*^^s
ff~ f-
sz
^
^^^
^ ^
^^^
^
> > > >

0.7
^ ^ ^ ^
^^^ "^^^
§ ^
§ ^
^.^
X r I > oII
c
4 X
o ^

cted ract
0.6
k.
•^
£

nse
in CD
•^^^ o
n
CO
o c 0)
<D
F
3 c
o
'^ >o CO
k.
d CD 3

Por
Figure 4.35: HW- and VV\ config
CO 1

o
0.2

^^~
d

r-
CO CO CM

O o o o o o
(uojpBj;) Aj9Aooay no

115
O) «i
o

and 28
00 c

hen h=7
o

1^ ^
o in
o
<^-iS
o
II
c
CO
p ^
C >.
^
o 2
«^- k_
^"-^ o
TJ •^
CD 0)

ons
TS
in CD
C* C3.
o ^^" CO
<D CD
E Od
3 c
o o

rati
>
d CD
k_ -}
CJ)
o

36: HW andVW onfi


Q.

CO o
1
o

CM 1

f - '^
gure

CO in CO CM
d d d d o o
(uojpBjj) ^l^Aoo^y no

116
CJ)

and 28 ft
d

00 00
CM CM
r^
II r^
II II II oo JC
-C -C
-c o

enh=7
m in m in
oII oII oII oII
^^s»
^^^ f- <" c
x:
.^ ^ ^ ^ r^ .c
^^^ ^^^ d
^
>
^
> >
Jl^ ^
> m
^*^ ^^^
^ l _ _ l - ^

^_^ oII
§ § 5 § c
X ? I ^ o :^

Inj cted ract


0.6
4 X k.
••—
a

nse
m• CD
o
Q.
CO
o CD
Od
E
3 c
O o

config ati
'^
>
d 0 3

Por
CO 1

CD

iX
0.2

••
rCO
^
Tj-
^^
d 2
3
CJ)

T
—r
1^ CO
in CO CM
d d o o d
d
(uojpBJj) Xj9Aooay no

117
CJ) Jti
o

and 42
oo c

h=7
o

c
CD
f~

r^ ^
o in
o
^—V
oII
c
q v?
CO TS > •

o 2 ^
k_
«c O
TJ «»-
CD 0)
TS CO
m CD
o
Q.
o c*
^^ CO
0) CD
E Q:
3
c
O
o

rati
"St >
d <D
^ 3
O CJ)

38: HW andVW-co nfi


QL

CO
o

CM 1

T - TT
Figure

(uojpBJ)) AjaAOoay no

118
CJ) .

and 42 ft
d

r^ r^ "^ "^
II II
00 iC
d

en h=7
>n in in in
d CD CD CD
JL }L. X X
•^ ^ ^ :^
r^
1 1 1B d
m
^.«^
f^
oII
o ^

Inj cted ract


0.6
k.
•»—
a

nse
i n« CD
o
CL
CO
o <D 0)

E
3
c
f)
O

ati
"^
> k.
d P
k_
3
CJ)
o
Q. c
o
o1
CO

4.39: HW- and V\A


o
0.2

^^^
/-\ <D

CJ>

1^ CO in CO CM
d d d o o d
(uojpBJi) AjaAooay no

119
CD

a
II
O)
CD
o
CO
o
OO "CD

ability R
o

h- CD
o §
^_^ CD
c Q.
,^^
o

cti
CO c
CD ^
0)D
L_ C
o

iff
•*—
TJ n

Inj Cte
p
CD •»—
m 0)
d CO

ura on Re pon
CD
E
3 CO

re Vol
0.4
'^
o
Q.
CO CJ)
i^
d

CM
TJ
d C
CD

O
-^
"^
0)
k.
3
^ o g)
CJ) oo h- CO in "^ CO CM Li.
o o o o o o o o
(uojpBJj) ino ja;BM

120
CD
0)

«^

CJ)
d
¥
0)
CO

00 g
d "co

CD
CD
d
'^ CD

CO
1-
d 2 »^
•h: 0)
^ St
"g b
m ^ o
d CD 0)
«CO
^
CD
O
E Q.
CO
_3 CD
O Od
d > c
2 g
o
Q. "co
CO CJ)
d

CM
d

CD
3
g)
ij.

(uojpBJi) ino jaiBM

121
^. 0)
CD
•^-
^
CM
11
^^
CJ) C

bility R ati oswhe


o

00

1^ CD
0)
d
^—K 0)
Q.
o •^

fract
eren
0.6
^"^ st=
TJ
CD Q
k.
ts o
in 0)
a0)
o c CO
CD c
E o
3 CO
O 0)
d >
CD
k. c
o

urati
o
CL
CO ai
o i^

CM

CM

0
i—
3
CJ)

(uojpBJi) ino jaiBM

122
bility R ati oswhe h=28 fee
CJ) c
d

00

1^ CD
CD
o
^—V 0)
o a.
^hrf

fract
eren
0.6
• * — ^
St:
TJ
CD Q

ect
k-
in • ^ ^ M
a
c 0)
o "^" CO
CD c

spo
lum
o 0)

0.4
> 01
CD c
k-
o p

V-confi urat
Q.

CO CJ)
o

CM
O
e 4.43: W-and

• ' "
X
o

CJ)
CJ) 00 r^ CO •^ CO CM
m
o o o o o o o o
(uojpBJi) ino ja;BM

123
bility R ati oswhe h=35 fee
• -

CJ) c
d

00
d

CD
r^ 0)

UJJ
d
.^-^ 0)
o Q.
^ta^

eren
fract
0.6
'"—^ St:
"O
CD Q
ts o
CD •il
m _C 0)
o CO
CD
c
E o
3 CO
C3.
o
d >
a> c

iZ-confi uratio
Por
CO CJ)
o

CM 5
o
and

0)
3
C3)

(uojpBJi) jno jajBM

124
bility R ati oswhe h=42 fee
CJ)
c
d

oo
d

1^ CO
CD
d
^—iK
0)
c CL
o
^^^

eren
tract
0.6
^-^ st=
TJ
CD Q
k.
TS o
in 0)
•i:
o s
• ^^^ 0)
CO
CD
c
E o
3 CO.
Q
O CD
> Od
d
0) c
k-
g
o

V-confi urat
Q.
CO CJ)
o

CM
o
and

in
"^

CD
k_
3
g)
Li.

(uojpBJi) ;no jaiBM

125
M« • • ^

X4 in
Ml
q
CJ) dII
X|

^ 'ir ^- ^
x» d =
i.
Ml
II jL II II i» ^
^
Ml TJ
in S in ^ i»
C
M
^
o q o q »
CO CO
d 9 o o d c
3« • 1^
n II

when
•VW(I

-HW(
-HW(

-VW(
d
« •
« ^^^
6^6^ tt
C
CO
c
4( O« OX Oa O4 A i
o o

0.6
|ected (fracti

lurati
4(

« 4a
••—
4X •a o
ID o
1^
4X «a d
0)

Volurn
4X • a
TJ
4 X « D
c
^j- CO
4 X • a
d 2 §
4 X • a o I
Q.

ure 4.4 6:P rodu ion Rates fo


4 X • a

4 X • a 0.3
« 4 D X

• a 4 X

• a 4 X
CM ts
• a 4 X
d

« a 4 X
0.1

• a 4 X

• a 4 X

• (1 CJ)
Li.
1 1 1 1H o
c) O O o o o c
c
C£) o o o
CO o
CM
o
T—
> m -^
(ABQ/aiS) sajBy Moid

126
CHAPTER 5
CONCLUSIONS

The simulation model developed in this study was used to


investigate the potential of horizontal well injectors in waterflooding. The
conclusions are as follows.
1. The two most Important factors that affect the recovery of oil when a
horizontal well Is used as injector are vertical to horizontal permeability
ratio and formation thickness (the permeability ratio (ky/kh) was varied
from 0.05 to 0.5 and the formation thickness was varied from 21 feet to
126 feet, in this study).
2. The results of this study showed that the use of horizontal well injectors
in waterflooding is advantageous as compared to the vertical well
injectors for the range of kv/kh and formation thickness, considered in
this study.
3. The use of horizontal well injector results in more oil recovered at the
producer as compared to a vertical well injector, in a thinner reservoir
with high kjk^. Specifically, this advantage of horizontal well injector
was more pronounced when the vertical to horizontal permeability ratio
was 0.3 or above and when the formation was 105 feet thick or less.
4. This study showed that the water cut at the producer is less when a
horizontal well injector is used as opposed to when a vertical well
injector is used, below the pore volume injected value of 0.5. This
water cut for horizontal well injector case becomes more than the water
cut for vertical injector case above pore volume injected value of 0.5.
This increase in water cut occun-ed only for ky/kh of 0.05.

127
5. In this study, about 50% of the recoverable oil was produced when 0.5
pore volume was injected and only about 13% of the additional
recoverable oil was produced when pore volume injected was equal to
1.0.
6. The decrease in kv/kh does not have a significant effect on the
performance of a vertical well injector as long as the vertical well
penetrates the entire vertical reservoir thickness.

128
CHAPTER 6
RECOMMENDATIONS

The recommendations for the future work using the simulation


model, developed in this study, are as follows.
1. The horizontal well length, which was kept constant in this study, can
be varied to see if an increase in horizontal well length always
increases horizontal well's advantage over a vertical well or if there is
an optimum horizontal well length beyond which there is no increase in
advantage.
2. A vertical well was used as a producer in this study. Several simulation
runs can be performed to evaluate if the use of a horizontal well
producer would significantly increase the oil recovery.
3. The areal extent of the reservoir was kept constant in this study, while
the formation thickness was varied. An opposite scenario would be to
vary the areal extent, while the formation thickness is kept constant,
and the potential of a horizontal well injector can be evaluated for this
scenario.
4. The oil viscosity was kept constant in this study and it would be
interesting to see the effect, if any, of viscosity varying with pressure, on
the oil recovery.

129
REFERENCES

1. Willhite, G. Paul: Waterflooding. 1986, SPE, Richardson, Texas.

2. Craig, Forrest F. Jr. : Monooraoh Vol.3 : The Reservoir Enoineering


Aspects of Waterflooding. 1971, SPE of AIME, Dallas, Texas.

3. Latll M.: Enhanced Oil Recoverv. 1980, Imprlmerie Louis-Jean, Paris,


France.

4. Ewing, Richard E.: The Mathematics of Reservoir Simulation. 1983,


SIAM series, Philadelphia.

5. Aziz, K. and Settari, A.: Petroleum Reservoir Simulation. 1979,


Applied Science Publishers Limited, Essex, England.

6. Peaceman, D. W.: Fundamentals of Numerical Reservoir Simulation.


1977, Elsevier Scientific Publishing Company, New York.

7. Crichlow, Henry B.: Modern Reservoir Engineering-A Simulation


Approach. 1977, Prentice-Hall, Inc., New Jersey.

8. Mattax, Calvin C. and Dalton, Robert L.: Monograph Vol.13:


Reservoir Simulation. 1990, SPE, Richardson, Texas.

9. Douglas, J. Jr., Peaceman, D. W., and Rachford, H. H. Jr.: "A Method


for Calculating Multi-Dimensional Immiscible Displacement,"
Transactions AIME. 1959.

10. Coats, K. H.: "An Analysis for Simulating reservoir Performance


Under Pressure Maintenance by Gas and/or Water Injection," paper
SPE 2130 presented at the SPE Rocky Mountain Regional Meeting,
Billings, Mont., June 5-7, 1968.

11. Blair, P. M. and Weinaug, C. F.: "Solution of Two-Phase Flow


Problems Using Implicit Difference Equations," SPEJ. December,
1969.

130
12 Coats, K. H.. Nielsen, R. L., Terhune, M. H., and Weber, A. G.:
"Simulation of Three-Dlmensional, Two-Phase Flow in Oil and Gas
Reservoirs." SPEJ. December, 1967.

13. Aziz, K.: "Reservoir Simulation Grids: Opportunities and Problems,"


JPI, July. 1993.

14. Odeh, A. S.: Practical Aspects of Simulation. Lectures presented at


Fourth International Fomm on Reservoir Simulation, Salzburg,
Austria, August 31 - September 4, 1992.

15. Douglas, J. Jr., Blair, P. M., and Wagner, R. J.: "Calculation of Linear
Waterflood Behavior Including the Effects of Capillary Pressure,"
Transactions AIME. 1958.

16. Daltaban, T. S. and Wall, C. G.: "Petroleum Reservoir Management -


Past, Present and Future," paper presented to the Southern Counties
Branch of IMinE, held at the Royal School of Mines, London, U.K.,
January 15, 1996.

17. Cade, R. W. and Joshi, S. D.: "Horizontal and Extended Reach Well
Technology: North Sea Experience," paper presented at the Improved
Oil Recovery Conference, London, U.K., November 10, 1993.

18. Joshi, S. D.: "A Review of Horizontal Well and Drainhole


Technology," paper SPE 16868 presented at the SPE Annual
Meeting, Dallas, Texas, September 27-30, 1987.

19. Joshi, S. D.: "Augmentation of Well Productivity Using Slant and


Horizontal Wells," JPI, June 1988.

20. Taber, J. J. and Seright, R. S.: "Horizontal Injection and Production


Wells for EOR or Waterflooding," paper SPE 23952 presented at the
1992 SPE Pemriain Basin Oil and Gas Recovery Conference, Midland,
Texas, March 18-20, 1992.

21. Kossack, C. A., Kleppe, J., and Aasen, T.: "Oil Production From the
Troll Field: A Comparison of Horizontal and Vertical Wells," paper
SPE 16869 presented at the 62nd Annual Technical Conference and
Exhibition of SPE, Dallas, Texas, September 27-30, 1987.

131
22. Byrom, T. G. and Coulter, G. R.: "Some Mechanical Aspects of
Formation Damage and Removal in Horizontal Wells," paper SPE
31145 presented at the SPE International Symposium on Fomriation
Damage Control, Lafayette, Louisiana, February 14-15, 1996.

23. Jones, A. T. and Davies, D. R.: "Quantifying Acid Placement: The


Key to Understanding Damage Removal in Horizontal Wells," paper
SPE 31146 presented at the SPE International Symposium on
Formation Damage Control, Lafayette, Louisiana, February 14-15,
1996.

24. Clonts, M. D. and Ramey, H. J. Jr.: "Pressure Transient Analysis for


Wells With Horizontal Drainholes," paper SPE 15116 presented at the
SPE California Regional Meeting, Oakland, California, April 2-4, 1986.

25. Goode, P. A. and Thambynayagam, R. K. M.: "Pressure Drawdown


and Buildup Analysis of Horizontal Wells in Anisotropic Media," SPE
Formation Evaluation. December, 1987.

26. Mutalik, P. N., Godbole, S. P., and Joshi, S. D.: "Effect of Drainage
Area Shapes on the Productivity of Horizontal Wells," paper SPE
18301 presented at the SPE Annual Technical Conference, Houston,
Texas, October 2-5, 1988.

27. Babu, D. K. and Odeh, Aziz S.: "Productivity of Horizontal Wells,"


SPE Reservoir Engineering. November 1989.

28. Goode, P. A. and Kuchuk, F. J.: "Inflow Perfomriance of Horizontal


Wells," SPE Reservoir Engineering. August 1991.

29. Peaceman, D. W.: "Interpretation of Wellblock Pressures in


Numerical Reservoir Simulation With Nonsquare Grid Blocks and
Anisotropic Pemfieability," SPEJ. June 1983.

30. Babu, D. K., Odeh, A. S., Al-Khalifa, A. J., and McCann, R. C : "The
Relation Between Wellblock and Wellbore Pressures in Numerical
Simulation of Horizontal Wells," SPE Reservoir Engineering. August
1991.

132
31. Peaceman, D. W.: "Representation of Horizontal Well in Numerical
Reservoir Simulation," SPE Advanced Technology Series. Vol.1.
No.1. 1991.

32. Brigham, W. E.: "Discussion of Productivity of Horizontal Wells," SPE


Reservoir Engineerino. May 1990.

33. Peaceman, D. W. : "Further Discussion of Productivity of Horizontal


Well," SPE Reservoir Engineering. August 1990.

34. Peaceman, D. W. : "Further Discussion of Productivity of Horizontal


Well," SPE Reservoir Engineering. February 1991.

35. Ramey, H. J., Jr., Kumar, A., and Gulati, M. S.: Gas Well Test
Analysis Under Water-Drive Conditions. 1973, AGA, Arlington, Va.

38. Craft, B. C , Hawkins, M. F., Terry R. E.: Applied Reservoir


Engineering. Second Edition, 1991, Prentice Hall, Englewood Cliffs,
NJ.

39. SPE Reprint Series No.11: Numerical Simulation. 1973, SPE, Dallas,
Texas.

40. SPE Reprint Series No.20: Numerical Simulation II. 1986, SPE,
Dallas, Texas.

41. Pedrosa, A. O., Jr.: "Use of Hybrid Grids in Reservoir Simulation,"


Ph.D. Dissertation, Stanford University, 1984.

42. Mattax, C. C. and Dalton, R. L.: Monograph Vol.13 : Reservoir


Simulation. 1990, SPE of AIME, Dallas, Texas.

133
APPENDIX A
DERIVATION OF FLOW EQUATIONS

134
Derivation of the flow equations
Mass balance for the reservoir block can be written as:

mass entering - mass leaving = mass accumulated. (A.I)

The mass entering the block in a time interval At can be written as:

Mass in = Flux*Area*time. (A.2)

In the case shown in the Figure 3.1

Mass in = F, AyAz + F, AxAz + F, AxAy At, (A.3)

and the mass leaving the system is given by:

Mass out = Fx.X+Ax..^yAz + Fv y+Ay AxAz + F, Z + AZ


AxAy At + qAxAyAzAt, (A.4)

where q is the source term and q > 0 (producer), q < 0 (injector).

Accumulation of mass is given by change in concentration of phase p (Cp)


over time At as:

Accumulation of mass = -C. AxAyAz, (A.5)


t+At

135
where p = oil, water, (p as a subscript will be used in the following
derivation to represent oil and water again.)
Now the mass balance can be written as:

F, AyAz + F, AxAz + F, AxAy At

x+Ax
AyAz + F,
y+Ay
AxAz + F,
z+Az
AxAy At - q AxAyAzAt (A.6)

-C. AxAyAz
t+Ai

Dividing equation (A.6) by AxAyAzAt and rearranging:

-F, y+Ay - F , ) . (c,L-Cp|,)


t+At PI
x+Ax * z+Az 2
-q = (A.7)
Ax Ay Az "At

Applying the limit when Ax, Ay, Az, and At tend to zero, the equation (A.7)
becomes:

aF, aFy aF, _^ ac^


(A.8)
ax ~ ay az ^" at

The flux F for the two phases (oil and water) is given by:

Fo=PoVo.
(A.9)

' w ~ Pw^w »
(A. 10)

136
w h e r e Vpare the darcy's velocities are given as:

V = V +V + v (A.11)

kk„^
fp
^ = Vp,-V (A. 12)
^^ \ ' ''l44gj

where g = acceleration due to gravity in ft/sec^


gc = gravitational constant = 32.2 ft/sec^

Pp=^(PpSTc)- (A. 13)

T h e concentration for the two phases can be written as:

Co =<t>P oSTC D ' (A. 14)


B.

0^w = (t)p
YKwSTC
w (A. 15)
Bw

Combining equations (A.8) through (A. 15), the flow equations for oil and
water can be written as:

r
_a_ r PoSTC v
N \
PoSTC_ dp oSTC
Qo
ax B ay V. B '^° azl B, zo
(A. 16)
ar(i)posTcSo'

137
f. \
r. 'wSTC f.
'wSTC 'wSTC
ax B. ay B. yw
J dz \ Bw zw -qw
r <t>PwSTcSwl (A. 17)
at V Bw J

As the densities are constant at the standard conditions, so the above


equations can be reduced by dividing with densities. The new simplified
form of the flow equations is:

1_ As
-V (A. 18)
B, at B. °

-V [ 1 - 1 -qw a r * o1 (A. 19)


8 ^™ "at B ^-
_ w _ w

where q_ = (A.20)
PpSTC

The simplified form of the Darcy's velocities can be formulated as:

Vp= ^VPp-YpVh), (A.21)


JLtp

where Y„p = p
^p
(A.22)
144g,

Combining equations (A. 18) through (A.22) results in the governing


equations for a black oil model with two components (oil and water):

138
vMvPo-y.vh)]-q,=|^As^:^ (A.23)
^ B,

^•K(vp„ -y„vh)]-q, = 1 A s „ |, (A^24)


^ B^

kk„
where x =—^ (A.25)

The auxiliary relations are given as:

S„ - S„ = 1, (A.26)

Pew = Po - P„ . (A.27)

B p = 7 ^ = f(Pp). (A.28)
''p.STC

where p = o, w,
RC = reservoir conditions,
STC = stock tank conditions.

^ = 4>rec[l-C^(Po-Prer)] (A.29)

139
APPENDIX B
DERIVATION OF FINITE DIFFERENCE FORM OF EQUATIONS

140
Derivation of the finite difference fomn of eauations
The flow equation for oil water is given as:

vMvPp-YpVh)]-qp = —s (B.I)
a

The extended for of this equation is:

_a_ app ah app ah ah


^p^
ax Yp ax + \ <
ay ^''ay + az ^P^ az Yp
az
-qc
ax dy
(B.2)
^*Sp^
at v B p y

The detailed finite difference scheme for the left hand side (LHS) of this
equation is given as:

141
app^
> —<
dx •+1/2.j.k
M/2.j.k
LHS=
Ax,

Xjy^ — > —•< Kr


i+1/2.j.k -^axj i-1/2.j.k

Ax,

/^A^ A
^ ,
r /^;v»
5Pp \
—<
. V
ay' ^'Ji.j+I/Zk L \
5y' /Ji,j_1/2,k
+
Ay,

^5hM
^oY
pip
> — • <
Ky
pip
v^y i.j+1/2.k i.j-1/2.k

Ay,

/";vi ^ ^ap>i
p -
v^y i.jJc-1/2
(B.3)
Az,

p'p
> — • <
Ky
P'P

i.j.k+1/2 i.j.k-1/2
qp-uj
'\.\M
Az,

142
LHS= Aw \\Pp.i+1.j.k Pp.i.j.kj Yp.i+i/2.j.kV''i+i,j.k-hij^jj
Ax, /-*^i+1/2,j.k

"^ 7 3 ^ ?Pp.i-1.j.k - Pp.i,j.k j - Yp.i-1 / 2.j,k V\-Uk - hij.k j )


^^i-1/2,j.k

+ •
1
Aw lvPp,i.j+1,k Pp.i,j.k/ Ypj.j+l/2.klhi,j+i.k - h j j i j j )
Ay, ^yi.j+1/2,k
(B.4)
"^ A r ' ' ' ' ^'Pp-'-i-^-'^ " Pp-'-i-"^ ^ " ^P-'-i-i' 2.k (hij-i.k - hi,j,k)}
'^yi.j-1/2.k

1
A^ \vPp.i.j.k+1 Pp.i.j.k/ Yp.l.j.k+1/2V'i.j.k+1 "i.j.kj/
Az, ^yi.j.k+1/2

^ Av ?Pp.U.»t-1 Pp.i.j.k/ Yp,i.j.k-i/2V'i,j.k-1 '^i.j.k/J qp.i.j.l<


^yiJ.k-1/2

Multiplying t h e a b o v e e q u a t i o n by AXj Ayj AZk gives:

LHS-Tpi^i/2j,it|pp,j+ij,i( Pp,i.j.k Yp.i+i/2.j.kvhi+i.j,k hjjkJJ

•'•Tpi_i/2j,|([ppj_ij,|( ~Pp,i.j.k - Yp.i-1/2.j.k\ni-1.j.k - ' ' i . j . k

+Tp.i.j+1/2,k Pp.i.j+1,k Pp.i.j. "Yp.i.j+1/2.k\hi.j+i. -h, .j.lc


(B.5)
•'"'p.i.j-1/2.k|Pp.i.j-1,k - P p . i J . -Yp.i.j-l/2.kVhi.j-1. -h i.j.k
+Tp i j k+i / 2 lPp.i.j.k+1 - Pp.i.j. -Yp,i.j.k+1/2lhi.j.k+ - h i,j.k
'^'p.i.J.I<-1/2|Pp.i.j.k-1 "Pp.i.j - Yp.i,j.k-1/2lhi,j.| -h i.j.k <p.i.j.k

where T is the transmissibility and for positive x-direction it is given by the


following equation:

_ ^p.i+1/2.j.k .
p,i+1/2,jj< ~ Ay ^i+1/2.j.k (B.6)
^^^ i+1/2

143
The transmissibility expressions in the other directions are derived
similarly. Figure B.I helps understand AXi.1/2. The transmissibilities are
calculated using upstream relative pemneabilities and arithmetic average
for the other temris in equation (B.6). The detailed fonm of oil
transmissibility in field units in the positive x-direction is given by:

A'^xj+1,jj('^xj.jj(j 'ro,upstream
Txo.M/2j.k= 0.001127 Ay^Az, (B.7)
(BoHo) k1/2

Similarly, the y-direction oil transmissibility can be defined as:

^ ro.upstream
T,,,^y2M= 0.001 ^21Ax,Az, (B.8)
Ayjky.i,H>+Ayj,,ky,^, .(B„HJ, V2

and the z-direction transmissibility is as follows:

^('^ZAJjc+V2'^zj.j>; ro,upstream
T„ii*.,„= 0.001127Ax,Ay, (B.9)
^k}^zXiM+^ "*• AZk+ik^jjj^
(B^IT k+1/2

The water transmissibility is calculated in the same manner.


The discretization of the RHS (accumulation term) is given by the following
equations presented in Chapter 3:

144
1 ^

y.i.<i>i V"(|) n+ —
At «V^o^oAj.k 1-8 w
2 bl ' A ,tf'o
p >+
At At
LV y (3.25)
1
n+—
-b„^S:(A,p„-A,pJ

1
n+- 1 1
"A 2 n+— n+—
At '^^"""Aik ^, ^-1-
^w w '-•tr
(3.26)
b";2S';(A,p^_A,pJ

This descritization was performed using the following definitions:

'^°+^ = 0. (B.10)
at a

as„ as„ 5 P » . ='w S I ^ = S ; A . ( P „ - P J . (B.11)


a ap_
cw
a a

as (B.12)
^=-s:,A,(p,-pJ,
a
»n+1 _ on
(B.13)
^"^Po(sr)-Pc(s:)'

1 (B.I 4)
•^P

145
^ 1^

b: = (B.I 5)
Sp,
(pr-p;)

146
AXi.1/2 AXj+1/2

o o o o o

Xi-1 Xj Xi+1

Figure B.I: Example of Ax^^ 1/2

147

You might also like