SOAK ALTERNATING GAS: A NEW

APPROACH TO CARBON DIOXIDE FLOODING

by

MALCOLM DAVID MURRAY, B.S.P.E.

A THESIS

IN

PETROLEUM ENGINEERING

Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE

IN

PETROLEUM ENGINEERING

Approved

Chairperson of the Committee

Accepted

Dean of the Graduate School

August, 2002 ACKNOWLEDGEMENTS

"Wisdom is the principle thing; therefore get wisdom and with all your getting get understanding." - Proverbs 4:7

I would like to express my greatest appreciation to those without whom

this thesis would not have been possible: first, to Dr. Scott Frailey, my thesis

advisor whose expectations of his students have continually challenged me to do

my best in both my studies and in this thesis. Although a positive attitude and

strong determination are necessary for a student to achieve his greatest possible

success, the high standards of his teachers are indispensable to the quality and

level of his education: I therefore count it a great privilege to have worked under

Dr. Frailey while at Texas Tech University. I am also deeply indebted to Dr.

Akanni Lawal who offered me the research assistantship that made studying at

Texas Tech University possible: the opportunity to study here and the quality of the friendships that have been formed have made this a rich experience indeed. I would like to thank the Butler family for the funding to make the research for this

project possible. I would also like to express my heartfelt thanks to Michelle Doss who helped with the final transcripts of this thesis, as well as to the professors and staff of the Petroleum Engineering department who helped in so many ways.

In addition, I would like to thank Dr. Lillian Chou and the people of the

International Family Fellowship who became such wonderful friends during my stay in Lubbock, and last but not least, my mother Zennie Fryer who, with my sisters, Carol Hawes and lla Olsen and their families, gave their words and prayers full of love, support and encouragement. Thank you all for what you have given me during this part of my life. TABLE OF CONTENTS

ACKOWLEDGEMENTS ii

ABSTRACT vii

LIST OF FIGURES ix

NOMENCLATURE xii

CHAPTER

1. INTRODUCTION 1

1.1. An Overview of Petroleum Recovery 1

1.2. The Role of Miscibility in Petroleum Recovery 2

1.3. CO2 Recovery Methods 3

1.4. Soak Alternating Gas: An Alternative Method of Mobility Control 6

2. LITERATURE SURVEY 8

2.1. Introduction 8

2.2. Immiscible Displacement 8

2.3. Miscible Displacement 15

2.3.1. Phase Behavior 16

2.3.2. Diffusion and Dispersion 19

2.3.3. Viscous/Capillary Forces 22

2.3.4. Dead-End Pores 24

2.3.5. Miscibility Development with High Water Saturation 25

IV 2.3.6. Areal Displacement Efficiency 27

2.3.7. Water Injectivity 28

2.4. Displacement Efficiency 29

2.4.1. Microscopic Displacement Efficiency 30

2.4.2 Macroscopic Displacement Efficiency 32

2.5 Field Applications 37

2.5.1. Continuous CO2 Flooding 38

2.5.2. Water-Alternating-Gas (WAG) 42

2.5.2.1. WAG Theoretical and Laboratory Models 42

2.5.2.2. WAG Field Implementations 44

2.5.3. Cyclic CO2 Stimulation (HufTn'Puff) 46

2.5.3.1. CO2 HufTn'Puff in Heavy Oil Reservoirs 46

2.5.3.2. CO2 HufTn'Puff in Conventional Oil Reservoirs 49

2.5.3.2.1. Numerical Simulations 49

2.5.3.2.2. Analyses of Laboratory and Field Data 49

2.5.3.2.3. Full-Scale Field Implementation 52

2.5.4. Summary 54

3. DISCUSSION 80

3.1. Introduction 80

3.2. Miscibility 81

3.3. Phase Behavior 81

3.3.1. The Vaporizing Gas Displacement Process 82 3.3.2. The Condensing Gas Displacement Process 82

3.3.3. The Vaporizing/Condensing Gas Displacement Process 83

3.4. CO2 Recovery Mechanisms 84

3.5. CO2 Recovery Processes 85

3.5.1. Continuous CO2 Flooding 85

3.5.2. CO2 Water Alternating Gas 86

3.5.3. Cyclic CO2 Stimulation (HufTn'Puff) 87

3.6. SAG: A New Alternative 88

3.6.1. SAG Parameters 90

3.6.2. Miscibility Conditions for SAG 93

4. CONCLUSIONS 97

5. RECOMMENDATIONS 99

BIBLIOGRPAHY 103

VI ABSTRACT

Carbon dioxide (CO2) flooding as a method of enhanced oil recovery

(EOR) has been used successfully for many years to increase the recovery of the original oil in place (OOIP) of a reservoir. The two most common methods of using CO2 to accomplish this are the continuous CO2 flood and the cyclic CO2 flood, or CO2 hufTn'puff. In the continuous CO2 flood, CO2 is injected continuously into a wellbore while fluids are recovered continuously at the adjacent wellbores. A frequently used variation of this process is the water- alternating-gas (WAG) flood where CO2 and water are injected alternately to combine the solvent properties of CO2 with the mobility properties of water and further optimize the recovery. However, one limitation to the WAG process is that of water shielding where the relatively high water saturation in the pore spaces prevents much of the potential contact between the crude oil and the CO2 from occurring and causes a significant portion of the oil to be bypassed.

In the cyclic CO2 flood, CO2 is injected into the reservoir, shut in for a soak period, then produced from the same wellbore until the economic limits of production are reached and the cycle must be repeated. During the injection phase of a CO2 hufTn'puff, CO2 is distributed over as great an area as possible throughout the reservoir. Next, during the soak period CO2 is allowed to disperse through the water to contact as much oil as possible and make full use of the

CO2 recovery mechanisms, which allows production to be optimized in the production phase of the process.

vii In this work, a new concept in CO2 flooding is introduced as "soak- alternating-gas," or SAG, which incorporates the soak period of a CO2 hufTn'puff into the continuous CO2 flood to provide additional mobility control and a viable alternative to a WAG process in cases where water injectivity is too low to allow

WAG to be feasible. Since SAG does not depend on water injectivity the prospect of greater recovery in such cases could be quite significant. In addition, the mobility control provided by SAG may offer advantages over those of WAG, even where water injectivity is adequate. Thus, the integration of continuous CO2 flooding techniques with those of the CO2 hufTn'puff appears to offer greater recovery potential than those of either method used separately.

The concepts behind SAG appear to be supported by previous literature on the research, testing, and implementation of the continuous CO2, the CO2 hufTn'puff, and the WAG processes. In order to ascertain its potential, it is recommended that the SAG process be investigated with respect to miscibility conditions as well as the parameters of the injection stage, soak period and production stage, then verified experimentally using slim tube experiments, coreflood experiments and pilot tests prior to full-scale field implementation.

VIII LIST OF FIGURES

2.1 Relative permeability characteristics of porous media

in (a) strongly water-wet rock, and (b) strongly oil-wet rock 57

2.2 Theoretical fractional water flow of water versus water saturation 57

2.3 Actual fractional water flow versus water saturation 58

2.4 Water saturation versus dimensionless distance at a dimensionless time (i.e., pore volumes injected) of 0.2 58 2.5 Dimensionless time versus Dimensionless at various degree of water saturation 59

2.6 Saturation history at XD=1 (Outlet of System) 59

2.7 Interfacial forces at an interface between two immiscible fluids

and a solid 60

2.8 Correlation of recoveries of residual phases as a function of Nca 60

2.9 Experimental and simulated densities from multi-contact miscibility 61

2.10 Holm-Josendal correlation of CO2 MMP as a function of temperature 61 2.11 Correlation of CO2 MMP as a function of extractable C5-C30 hydrocarbons present 62

2.12 Correlation of CO2 MMP as a function of the molecular weight of the crude oil 62

2.13 Schematic representation of viscous finger growth in unstable linear miscible displacements 63

2.14 Idealized impact of transverse dispersion on ultimate crude oil recovery by CO flooding 63

2.15 Image obtained from a microvisualization experiment 64

2.16 Miscible displacement of Fluid A by Fluid B, step change in concentration at inlet 65

IX 2.17 Concentration profiles for the injection of a slug of Fluid B to

displace Fluid A 65

2.18 Dispersion resulting from laminar flow in a straight capillary 66

2.19 Dispersion based on porous media being viewed as a series of mixing tanks 66 2.20 Dispersion resulting from bypassing of fluid trapped in

stagnant pockets 67

2.21 Dispersion caused by variation in the flow paths in porous media 67

2.22 Flow regimes for miscible displacement in a vertical cross-section 68

2.23 Flow regimes in a two-dimensional, uniform linear system 69

2.24 Effect of mobile water on oil recovery for CO2 MCM displacements of reservoir crude oil in a water-wet core 69 2.25 Effect of mobile water on oil recovery for CO2 MCM displacements of reservoir crude oil in an oil-wet core 70

2.26 Solvent cut as a function of pore volumes injected with K as a parameter 70

2.27 Claridge correlation for areal sweep efficiency as a function

of the mobility ratio 71

2.28 Schematic of displacement process with multiple phases 72

2.29 Gravity segregation in displacement processes 72

2.30 Standard flooding patterns 73

2.31 Fractional flow of CO2 and crude oil of well 2-26, Joffre Viking field 74

2.32 Plot of CO2PIUS water injection versus recovery 74

2.33 Calculation of fractional water flow in a tertiary WAG process 75

2.34 Production rates of oil, water, and CO2 before and after CO2 injection in a WAG process (Four-Pattern Area) 75 2.35 Production rates before and after CO2 injection in a WAG process (Seventeen-Pattern Area) 76

2.36 Utilization of CO2 with incremental oil production and CO2 production in a WAG process (Seventeen-Pattern Area) 76

2.37 Comparison between cumulative recoveries on a continuous CO2 flood, 1:1 WAG, and Denver Unit WAG processes 11

2.38 Yearly cumulative production versus soak times, West Sak Field. Alaska 77

2.39 Yearly cumulative production versus CO2 slug size. West Sak Field, Alaska 78

2.40 Summary of the effect of repeated cycles on CO2 hufTn'puff incremental oil recovery 78

2.41 Total recovery of cyclic CO2 and cyclic N2 corefloods versus slug size measured in moles and reservoir pore volumes 79

2.42 Increase in the recovery rate for a CO2 hufTn'puff process, South Louisiana 79

3.1 Plot of oil recovery factor versus pressure at a 1.2 PVs of CO2 injected into a sand pack containing Mead-Strawn stock tank oil 95

3.2 Plot of oil recovery factor versus fraction of solvent in lean gas, showing the point of minimum miscibility enrichment (MME) 95

3.3 Pseudoternary diagram of the vaporizing gas process 96

3.4 Pseudoternary diagram of the condensing gas process 96

XI NOMENCLATURE

A cross-sectional area normal to the direction of flow. In Equation 2.14, this also denotes fluid A. B In Equation 2.14, this denotes fluid B. Ce concentration of fluid B D apparent diffusion coefficient Do molecular diffusion coefficient DBA the coefficient of diffusion characteristic of fluid B being injected into fluid A in the direction of fluid flow between the injection and production wellbores DaBA the apparent diffusion coefficient, characteristic of fluid B being injected into fluid A, taken to be in the average direction of 45 degrees to that of fluid flow between the injection and production wellbores dp particle diameter E overall displacement efficiency EA areal displacement efficiency ED microscopic displacement efficiency El volumetric displacement efficiency Ev macroscopic displacement efficiency F formation resistivity factor FR formation resistivity factor f^ the fraction of water flow with respect to total fluid flow fs the fraction of solvent flow with respect to total fluid flow f^2 the fraction of water flow at the producing wellbore with respect to total fluid flow h height of the porous media K Koval's K-factor

XII Ki longitudinal dispersion coefficient Kt transverse dispersion coefficient k absolute permeability of a porous medium to a given fluid l

wellbores

XIII SD average saturation of the displacing phase, i.e., behind the flood front Sj average saturation of the displaced phase, i.e., ahead of the flood front t time to dimensionless time, as a fraction of the time needed to fill the entire pore volume with the injected fluid assuming a constant rate of injection tD2 dimensionless time needed to propagate saturation Sw2 at the producing wellbore, as a fraction of the time needed to fill the entire pore volume with the injected fluid assuming a constant rate of injection Uw velocity of water in the direction from the injection to the producing wellbore after its entering the porous media u fluid velocity in the direction from the injection to the producing wellbore after its entering the porous media Vpi pore volumes of solvent injected V fluid velocity in the direction from the injection to the producing wellbore after its entering the porous media Vw velocity of water in the direction from the injection to the producing wellbore just prior to its entering the porous media X distance from the injection well normal to the direction of flow Xc width of the constriction in a constricted capillary xsw distance of the waterflood front from the injection well normal to the direction of flow

XDSW dimensionless distance of the waterflood front from the injection well normal to the direction of flow, as a fraction of the distance between the injection and the producing wellbore

XIV Greek Letters a an empirical constant proportional to the amount of water blocking or water shielding during a CO2 flood ^ porosity of the formation X fluid mobility, the ratio of its permeability to its viscosity XD fluid mobility of the displacing phase Xd fluid mobility of the displaced phase XiD total mobility of the displacing phases ?itd total mobility of the displaced phases H viscosity of the fluid under consideration //£) viscosity of the displacing phase ^ viscosity of the displaced phase m viscosity of fluid i Ho viscosity of oil //s viscosity of the solvent /iw viscosity of water p density G interfacial tension cr„w interfacial tension between oil and water

XV CHAPTER 1

INTRODUCTION

The use of carbon dioxide (CO2) as a method of enhanced oil recovery has been studied since the early 1950s,' and its use grew significantly in the

1970s and 1980s.^ It is used as an enhanced oil recovery (EOR) process where it is injected following either natural drive or waterflooding in order to recover additional oil. An overview of petroleum recovery follows to provide background on the methods that use C02 to increase oil recovery.

1.1. An Overview of Petroleum Recovery

During the life of an oil reservoir, production is usually carried out by primary recovery, secondary recovery, and enhanced oil recovery (EOR).

Primary recovery uses the natural energy present in the reservoir to displace oil to the wellbore and includes solution-gas drive, gas-cap drive, natural water drive, fluid expansion, rock expansion, and gravity drainage. Secondary recovery processes add energy to the reservoir by injecting water or immiscible gas, which displaces oil to producing wellbores. Secondary recovery is normally carried out with a waterflood rather than an immiscible gas because of the low cost, abundant supply, and more favorable mobility of water. EOR processes are carried out by gas injection, liquid chemical injection and/or the addition of thermal energy.^ Gas injection with light hydrocarbons, CO2, nitrogen, or flue gas is considered to be an EOR process if the significant recovery mechanisms include oil swelling, oil viscosity reduction or favorable phase behavior. Liquid chemical injection includes the injection of polymers, surfactants, and liquid hydrocarbon solvents, while thermal recovery includes methods such as steam drive, cyclic steam flooding, hot water flooding, and in situ combustion.

1.2. The Role of Miscibility in Petroleum Recovery

While waterflooding is normally the most practical method of maintaining reservoir pressure, water-oil interfacial tension (IFT) combined with an adverse mobility ratio can limit the amount of oil recovered. This is due to oil droplets forming inside water-wet pores and becoming trapped or causing the water to channel through oil-wet pores, leading to viscous fingers. In each case, oil is bypassed and a high water-oil ratio (WOR) occurs at the producing well while much oil remains in the reservoir. These problems can be overcome by flooding the reservoir with a miscible fluid such as CO2, which can reduce interfacial tension and mobilize hydrocarbons at reservoir conditions.

In CO2 processes, two types of miscibility can occur: first contact miscibility and multi-contact (or dynamic) miscibility. First contact miscibility

(FCM) occurs if a single phase forms when CO2 is mixed in all proportions with the crude oil at a specific temperature and pressure. Multi-contact miscibility

(MCM) occurs when miscible conditions are developed in situ through composition alteration of the CO2 gas and crude oil as CO2 moves through the reservoir. Under most reservoir conditions CO2 FCM does not occur, but research and experience have shown that MCM processes can be efficient enough to approach 100% oil recovery.

1.3. CO? Recovery Methods

Three types of MCM processes can occur: the vaporizing gas (or lean gas) process, the condensing gas (or enriched gas) process, and the vaporizing/condensing gas process.'' The mass transfer of light and intermediate hydrocarbons from the oil phase into the CO2 phase characterizes the vaporizing gas process, while the mass transfer of light and intermediate hydrocarbons from an enriched (i.e. hydrocarbon enriched) CO2 phase into the oil phase characterizes the condensing gas process. A mass transfer that occurs in both directions between the phases characterizes the vaporizing/condensing gas process.

The mechanisms by which CO2 increases oil recovery includes oil swelling, reduction of oil viscosity, reduction of oil density, reduction of interfacial tension between the two phases, and the acidization of carbonate formations.

Eliminating interfacial tension between the CO2 and crude oil improves the relative permeability of each fluid. The three main CO2 flooding processes are continuous CO2 flooding, water-alternating-gas (WAG) flooding, and CO2 hufTn'puff (or cyclic CO2 stimulation). The process of continuous CO2 flooding is similar to waterflooding, but causes a gradual rather than abrupt change in the concentrations of the displacing and the displaced phases.

Problems with an adverse mobility ratio in continuous CO2 flooding can often be modified by water-alternating-gas (WAG) treatments where injections of

CO2 are followed by injections of water. The economics of WAG processes have been feasible in many cases due to the reduced volume of CO2 required and the more favorable water/oil mobility ratio over the C02/oil mobility ratio. This decreases viscous fingering and gravity segregation and results in an increase in the volumetric sweep efficiency. However, the WAG process generally has a lower microscopic displacement efficiency due to water shielding that inhibits contact between the oil and CO2. WAG processes can also be affected by low water injectivity (compared to waterflood water injectivity) just as waterfloods can, and result in very low recoveries. Where such problems occur, WAG may not be an option.

In the third process, CO2 hufTn'puff, the injection of CO2 and production of oil occur at the same wellbore, where the extraction mechanism predominates significantly more than in the continuous CO2 or WAG processes. The CO2 hufTn'puff is comprised of a CO2 injection stage, a soak period where the well is shut in, and a production stage. During the injection stage, by design the injected

CO2 bypasses the oil and is dispersed throughout the reservoir near the wellbore to allow mass transfer between the CO2 and crude oil to occur during the soak period. The soak period is necessary due to the poorer miscibility conditions, and varies in duration. During the soak period, oil swelling occurs throughout the region contacted by CO2 rather than at a flood front; this increases the fractional flow of oil. Lower oil viscosity and lower IFT between the CO2 and crude oil also allow greater oil migration. In the production stage of a CO2 hufTn'puff, the oil is recovered through both the extraction mechanism and by the physical displacement of the oil phase to the wellbore.

A cyclic CO2 hufTn'puff process can be more difficult to optimize because the crude oil is to be bypassed during injection then displaced during production when the flow is reversed, whereas bypassing is not desired in a forward drive process such as a continuous CO2 flood or WAG. If the pressure is too high or increases too much during the injection stage of a CO2 hufTn'puff, the physical displacement mechanisms will take effect prematurely and displace oil away from the wellbore. The subsequent recovery of this oil will likely be lower since it is displaced further from the wellbore.

When there is poor water injectivity and neither WAG nor waterflooding are viable, the remaining alternatives are CO2 hufTn'puff or the continuous CO2 flood. Although a continuous flood will normally produce a higher oil recovery than a CO2 hufTn'puff, it can also be prone to problems such as viscous fingering due to an unfavorable mobility ratio and the use of large volumes of CO2. Thus, a method of mobility control for a continuous CO2 flood to take the place of the

WAG option would be desirable. 1.4. Soak Alternating Gas: An Alternative Method of Mobility Control

An alternative method of mobility control appears to be possible in CO2 processes by applying the soak period of the CO2 hufTn'puff to the continuous flood. This seems especially well suited for reservoir conditions where water injection is not possible, due to low permeability or water contamination.

The injection cycle of a CO2 hufTn'puff is run at immiscible or near- miscible conditions to disperse CO2 throughout the reservoir, after which a mass transfer occurs between the oil and CO2 phases during the soak period. This allows oil to be displaced efficiently back to the injection wellbore during the production stage. When a continuous flood is run under immiscible conditions, bypassing will also occur and cause the CO2 recovery mechanisms to operate farther behind the flood front. To minimize this delay in recovery, it is suggested that the injection well be shut in to let the recovery mechanisms take full effect before proceeding further with the injection. This concept of optimizing the recovery in a continuous CO2 process by using soak periods has been termed soak alternating gas (SAG). It would appear that a SAG process can be successful over a range of miscibility conditions that include those of the CO2 hufTn'puff and the continuous CO2 flood.

The objective of this thesis is to examine the theory behind current CO2 recovery processes and identify the mechanisms through which those processes mobilize hydrocarbons to reconfigure those mechanisms into the SAG process. This includes the range of conditions under which the SAG process will be a viable means of enhanced oil recovery. CHAPTER 2

LITERATURE SURVEY

2.1. Introduction

Fluid flow through porous media is governed by Darcy's equation, which is written for single-phase flow through a horizontal, linear system as follows:

. = -^^. (2.1) H dx

Where two fluids flow simultaneously, such as in a secondary recovery process, equation 2.1 is modified to include the concept of relative permeability:

1,-^^- (2.2) Mi dx

The relative permeability of porous media to oil and water can be modeled as functions of water saturation (Figure 2.1) and generally have a nonlinear relationship. This is due to factors that include capillary forces, interfacial tension, wettability of the matrix, and fluid viscosities.

2.2. Immiscible Displacement

As the practice of waterflooding grew, so did the need to model binary fluid systems. A widely accepted method of doing this was developed by Buckley

8 and Leverett,^ who based their model on the fractional flow of water, defined as the ratio of water flow to total fluid flow:

/.=-^^. (2.3)

The fractional flow equation can also be expressed in terms of relative permeability and fluid viscosity by substituting the right hand side of equation 2.2 for the flow terms in equation 2.3 and then reducing:

A= r-v-- (2.4) k^.+ ' M. ^ \i

When the viscosities of the fluids are known and constant, equation 2.4 can be modeled as shown in Figure 2.2. The Buckley and Leverett equation in turn models the position of the water flood front, defined for linear systems as follows:

^s, ^ <1 C (2 5) dt (jAdS^'

By separating the variables in the left side of equation 2.5 then integrating and dividing both sides by the length of the system, a dimensionless form results:

X _3^ = J1L.^1}^qt 5/-^ (2.6) L (j)AL dS,

Expressed in dimensionless terms, equation 2.6 takes the following form:

x =t -^ (2 7) Since the dimensionless time in equation 2.7 is also equivalent to pore volumes of water injected at a constant rate of flow, the differential term for fractional flow is equivalent to dimensionless velocity:

(2.8) dS^ t^

Since the slope of the plot of Figure 2.2 is equal to the dimensionless velocity at any given saturation at or behind the flood front, the velocity at the flood front, which is constant over the range of saturations at the flood front, must appear as a straight line. This straight line is drawn from the point of initial water saturation to the point where it is tangent with the fractional flow curve, to arrive at the plot shown in Figure 2.3. The derivative of the plot in Figure 2.3 is one of dimensionless velocity as a function of water saturation; if this derivative is plotted, the plot's axes exchanged, and the dimensionless velocity is multiplied by the dimensionless time, a plot of water saturation versus dimensionless distance for a specific dimensionless time (or pore volumes injected) is the result, shown in Figure 2.4.

Welge® used the concepts of Buckley and Leverett to develop a method to calculate the average water saturation in the reservoir after water breakthrough.

His equation for a linear system is given as follows:

S. = S^,+t,,{\-fJ. (2.9)

Through Welge's derivation the average water saturation, S„, at breakthrough can be found graphically from Figure 2.3 by extending the tangent

10 line to the fractional water flow of one. Further plots based on the Buckley-

Leverett model include those of Figures 2.5 and 2.6. Figure 2.5 is a plot of dimensionless distance versus dimensionless time at various water saturations, where it can be seen how progressively higher water saturations correlate with progressively lower slopes, or water velocities. This shows that the velocity of the water increases through the swept zone for a given injection rate as the fraction of bypassed oil increases. Figure 2.6 is a plot of the water saturation at the outiet over the given span of dimensionless time.

The flood front saturation shown in Figure 2.4 only shows the water saturation from the swept zone rather than the entire reservoir, but not all areas of the reservoir are swept. Thus, the determining factors in the secondary recovery process are the macroscopic and microscopic displacement efficiencies. Macroscopic displacement efficiency is reflected by the volumetric sweep efficiency, which is in turn a product of the areal and vertical sweep efficiencies. Microscopic displacement efficiency is reflected by the saturation of the swept zone, i.e., behind the flood front. The relationship between relative permeability and saturation shown involves viscous and capillary forces. Viscous forces are reflected by the magnitude of the pressure gradient that is forcing the fluids to flow, while capillary forces are determined by interfacial tension (IFT) and rock wettability that contribute to the resistance to fluid flow. Interfacial tension acts at the interface between two different liquid phases, and is defined as the force acting in the plane of the surface per unit length of the surface where

11 it applies to two different liquids or a liquid and a solid.^ Wettability acts at fluid/solid interfaces and is defined as the tendency of one fluid to spread on or adhere to a solid surface in the presence of a second fluid^ (Figure 2.7).

Amyx, Bass, and Whiting^ studied wettability and concluded it to be a function of adhesion tension, which was in turn a function of IFT. Wettability is determined by the contact angle through the water phase between the oil/water interface and the face of the solid. Green and Willhite^ defined four types of surfaces with respect to wettability: water-wet, oil-wet, intermediate-wet, and mixed-wet. A water-wet surface has a greater preference to adhere to water than oil, whereas the opposite is true for an oil-wet surface. An example of water- wetness is the concave-upward meniscus that occurs in a hollow upright tube that is partly filled with water and the remaining portion filled with oil. An intermediate-wet surface has no significant preference to adhere to one fluid over another, and a mixed-wet surface has some areas that are water-wet, other areas that are oil-wet, as well as other areas that are intermediate-wet. Moore and Slobod® found that when the capillary forces are greater than the viscous forces, the larger capillaries are bypassed. Taber^ found that the oil recovery factor increases with the ratio Ap/La, which has subsequently been referred to as the Taber number, and that there was a minimum Taber number below which no residual oil could be recovered. Stegemeier'° studied oil entrapment and mobilization and found that trapping behavior is controlled by pore geometry, fluid-rock properties (particularly wettability), and fluid-fluid properties such as the

12 viscosity ratio and the density difference. In addition, he provided correlations of recoveries of residual oil as a function of the capillary number, a dimensionless variation of the Taber number (Figure 2.8). Stegemeier" also developed a method of predicting the residual oil saturation as a function of rock type and fluid properties, and developed the following capillary number:

N,,=^ (2.10) where the righthand side is the equivalent, through Darcy's equation, of the

Taber number multiplied by the absolute permeability of the porous media.

Melrose and Bradner^^ studied residual oil saturation as a function of viscous and capillary forces, as well as the point at which the dominant displacement mechanism shifted between these two forces by correlating the microscopic displacement efficiency to the capillary number. They concluded that the residual fluid distribution in a homogeneous porous media was a function of the capillary forces, and that the recovery decreased as the pore size distribution increased, and proposed the following capillary number:

K.=^- (2-11)

Here, it can be seen that the term Uw/0 is the equivalent of the Darcy velocity, Vw, and the difference between equation 2.11 and equation 2.10 is that Stegemeier included the term for the relative permeability of water, which changes as the water saturation changes.

13 Mungan'^ found that altering core conditions from oil-wet to water-wet yielded an additional 2.6-3% oil recovery in a waterflood, but altering the conditions from oil-wet to intermediate-wet did not increase the recovery factor.

Wagner and Leach''' found that volumetric sweep efficiency increases when the

IFT is reduced below 0.07 dynes/cm. Gunn and Slattery'^ showed that for a given IFT, the residual oil saturation is more easily removed from smaller pores in a water-wet or intermediate-wet rock than from an oil-wet rock, and estimated a critical capillary number of 2.23x10"^ for the recovery of residual oil from pores of a Berea sandstone whose pore neck radii are larger than the mean pore neck radius. Their definition for the critical capillary number was that of Eriich, Hasiba, and Raimondi'® as follows:

NJ^. (2.12)

Here the term |V/)|/cr is the Taber number for three-dimensions, and the absolute permeability, /c, is used, but through Darcy's equation it can also be seen that this capillary number is simply the product of Darcy velocity and viscosity divided by the interfacial tension between the fluids, or the equivalent of equation 2.11.

Arriola, Willhite, and Green'^ investigated the trapping and mobilization of oil drops through a constricted square capillary through which water was flowing, and found trapping to occur when the viscous forces equaled the capillary forces, i.e., hydrodynamic stability was achieved. Their apparatus consisted of a square

14 lOOmmxIOOmm tube with a square constriction and used the following capillary number that would correlate with the capillary number used by Gunn and

Slattery'^ for porous media:

N„-^ • (2.13)

a

Here, Ap was the pressure drop across the capillary, and Xc was the width of the constriction.

2.3. Miscible Displacement

The mass transfer of components from one fluid into another fluid characterizes miscible fluid behavior. In a C02/crude oil system first contact miscibility (FCM) occurs if a single phase forms when CO2 is mixed in all proportions with the crude oil.^ Multi-contact miscibility (MCM) occurs when the miscible conditions are developed in situ through composition alteration of the

CO2 and/or crude oil phase as CO2 moves through the reservoir.^ Thus, CO2 is advantageous because either first-contact or multi-contact miscibility can occur.

Although the reservoir pressure is not usually high enough to create conditions where FCM can occur, experience has proven that MCM processes can approach 100% oil recovery.

15 2.3.1. Phase Behavior

Hutchinson and Braun'° studied the condensing gas drive mechanism in a methane/crude oil system and felt that it dominated the displacement process once hydrocarbon components of the crude oil began to mix with the injected methane. Zick proposed the vaporizing/condensing gas drive to account for density shifts (Figure 2.9) observed during the recovery of crude oil in an MCM process, characterizing it as a mass transfer of both the CO2 into the crude oil phase as well as the light and intermediate hydrocarbons into the CO2 phase.

Holm and JosendaP^ studied CO2 displacement mechanisms and concluded that solution gas drive, oil swelling, viscosity reduction of oil, and a hydrocarbon extraction from the oil to the CO2 phase promoted microscopic displacement efficiencies close to 100%. Another advantage they observed in

CO2 over miscible hydrocarbon gases is that the displacement of oil did not depend on the presence of light hydrocarbons and therefore could apply to reservoirs depleted of light hydrocarbons. After further investigation^^ they concluded that the CO2 miscibility with crude oil was a function of CO2 density, which needed to be at least 0.42 gm/cm^, close to the critical CO2 density of

0.468 gm/cm^ to achieve an oil recovery of 94% or greater. Minimum miscibility pressure (MMP) has been defined in terms of ternary diagrams as the minimum pressure at which the limiting tie line just passes through the reservoir oil composition point.^ Holm and Josendal^ also showed MMP as a function of temperature (Figure 2.10) as well as the amount of extractable C5-C30

16 hydrocarbons present (Figure 2.11), the molecular weight of the C5+ hydrocarbons in the oil (Figure 2.12), and the molecular structure of hydrocarbons (e.g., aromatic rings) present in the oil.^'

Metcalfe and Yarborough^ studied miscibility processes of CO2 and concluded that the displacement mechanisms of both vaporization and condensation occurred in C02/hydrocarbon systems, and that reservoir temperature as well as reservoir pressure determined which mechanism would control the displacement. Metcalfe^ studied the effect of specific gases on the

MMP of a C02/crude oil system and concluded that the presence of H2S and C2+ lowered the MMP while Ci raised it. He also observed that an increase in temperature raised the MMP of the system.

Gardner and Ypma^^ observed the behavior of viscous fingering in MCM phase behavior that yielded low recoveries in unstable vaporizing-gas drive processes. They concluded that this was mainly due to transverse mixing of oil into regions of viscous fingers that were depleted of the components necessary for effective oil displacement, resulting in high residual oil saturation where bypassing occurred. They also noted that the regions of highest residual oil saturation occurred in the regions of greatest throughput, and that due to lateral boundary effects, recovery predictions based on coreflood results would be optimistic.

Figure 2.13^'' shows a schematic representation of viscous finger growth in unstable displacements where the greater mobility of the displacing fluid drives

17 it ahead of the displaced fluid via the viscous fingers. Figure 2.14 shows the effect that the reciprocal Peclet number, a dimensionless indicator of transverse dispersion comprised of the ratio of the time, L/v^, for a particle of fluid to be convected through the length of a core (i.e., residence time), to the time d^/Dr, to disperse across the diameter of the core, has on the ultimate oil recovery in a continuous CO2 flood. The implications of Figure 2.14 are that when viscous fingering dominates the CO2 flooding process, that is, when the displacement process is unstable, the ultimate recovery of a continuous CO2 flood will be substantially lower than when the displacement is stable and viscous fingers do not dominate the process.

Shyeh-Yung^^ investigated continuous CO2 EOR corefloods both at and below the MMP, and among his conclusions were (1) residual oil saturation increases linearly, and CO2 mobility decreases as pressure decreases; (2) both the physical displacement and extraction mechanisms increase as pressure increases, where the displacement mechanism appears to dominate the process during the first pore volume (PV) of injection while the extraction process appears to dominate thereafter; (3) CO2 mobility in an MCM system is solely a function of

CO2 saturation and can be reduced up to two orders of magnitude by the presence of residual oil; (4) oil recovery is not only the result of low-IFT physical displacement but is influenced significantly by mass transfer and water shielding crude oil in the reservoir pore spaces from the CO2 to the point where secondary

18 C02 processes yield higher recoveries than tertiary CO2 processes (i.e., following a waterflood).

Stern^ investigated the effects of pore-level fluid distribution, flow rate, core length, oil viscosity, wettability, WAG ratio, and initial water saturation on displacement mechanisms. He concluded that the high residual oil saturations often seen in continuous CO2 corefloods were the result of pore-level bypassing although the lighter hydrocarbons are subsequently recovered from bypassed pores, and that in tertiary floods the bypassing appeared to be the result of capillary pressure effects and pore-level dispersion. He also concluded that miscibility develops over relatively short distances and that in mixed-wet rock waterblocking is a small effect, while in water-wet rock at high WAG ratios waterblocking is sufficient to prevent bypassed oil from being extracted, resulting in a lower oil recovery. His work includes microvisualization experiments that show pore-level bypassing in thin sections (Figure 2.15).

2.3.2. Diffusion and Dispersion

Dispersion refers to mixing that occurs between the displacing and the displaced fluids during the recovery process.^ On the molecular level the fluids are mixed by diffusion, which is the mass transfer of fluid components between the two phases. Diffusion is characterized by a concentration gradient of one of the fluids across the phase boundary between the fluids. Figure 2.16 shows the concentration profiles of the displacing and displaced fluids changing with time,

19 where the concentration shifts abruptly at the earlier time but shifts more and more gradually as time progresses. During a CO2 injection where the fluids are moving, the concentration profile of the fluids will change with both time and distance, as shown in Figure 2.17.

The process of diffusion is modeled in Pick's first law^ shown in the following equations from Perkins and Johnston:^^

m,,=-D,,A(dC,ldx). (2.14)

This equation states that the rate of diffusion of fluid B into fluid A is proportional to the concentration gradient of fluid B. To allow for tortuosity in porous media, an apparent diffusion coefficient can be modeled that moves at an average of 45 degrees to the direction of flow:

D^, = 0.707 D,,. (2.15)

Perkins and Johnston^^ also modeled the relationship between the diffusion coefficient and the apparent diffusion coefficient using the formation resistivity factor as follows:

^^ = — . (2.16)

Taylor^® showed how in laminar capillary flow, the displacing fluid disperses longitudinally into the displaced fluid and that in the velocity profile of the system at breakthrough the volume of the injected fluid is half the total volume of the capillary. Because the width of the displacing phase is smaller at greater distances from the capillary inlet, there is a large area of contact where

20 diffusion predominates in the lateral direction. This phenomenon is called the Taylor effect (Figure 2.18).

Aris and Amundson^ modeled porous media as a series of mixing cells where perfect mixing occured within each cell (Figure 2.19), which allowed the dispersion process to be modeled as a process that occurred between the adjacent cells as the displacement process progressed through the system.

Aris also introduced a model of the diffusion process between the oil in dead­ end pores and the injected CO2 (Figure 2.20), where no mechanical mixing occurred. Raimondi et al.^' studied dispersion as a result of the variation in flow paths within porous media where fluid particles that are side-by-side at one point arrive downstream at different times due to the varying tortuosity in the different flow paths (Figure 2.21).

Perkins and Johnston^^ studied fluid miscibility as a function of both the diffusion process and a mechanical mixing process that was purely non- diffusional. They defined dispersion coefficients for longitudinal flow (KI) and transverse flow (Kt) to account for these phenomena when the ratio vdpD/Do\s greater than 50:

K,=£l. + 0.5vd^a . (2.17)

K=^ + 0.0\57vdn (2.18) ' F(l> ' The first term in both equations 2.16 and 2.17 are common and represent the apparent diffusion coefficient, D, in terms of the molecular diffusion

21 coefficient, Do, and the formation electrical resistivity factor with porosity, 1/Fn.

The second term of each equation represents mechanical mixing by convection as a function of the average longitudinal interstitial velocity, v, the molecular diameter of the fluid, dp, and a factor for the inhomogeneity of the media, q then shows the longitudinal mixing to be greater than the lateral mixing by a factor of approximately 32.

The coefficient of longitudinal dispersion, Ki, is used to determine the concentration, CB, of a second fluid introduced into a first fluid at point x in a porous media by the following equation:

(x-vt) C, = yj\-erf 2V^r_ (2.19)

Equation 2.18 has the following initial and boundary conditions:

f = 0: X > 0, Ce = 0.0 , (2.20)

t >0:x = 0, CB= 1.0 , (2.21)

f >0:;c^Qo, CB = 0.0 . (2.22)

2.3.3 Viscous/Capillary Forces

Simon, Rosman, and Zana^^ applied the concept of the capillary number to a CO2 flood to study low-IFT effects and showed that the reduced IFT could cause up to a 39% greater reduction in the residual oil saturation beyond that of a waterflood. Stalkup^^ also developed a viscous/gravity ratio, formulated for consistent units as follows:

22 Craig et al. studied the effect of gravity segregation on miscible gas drive systems, concluding that up to 80% bypassing could occur in a linear system and that the amount of segregation was influenced more by the average injection rate than by variations in the injection rate. Moore and Slobod^^ also found that when viscous forces dominate in a miscible process the flow rate is proportional to the square of the radius of the channel.

Crane, Kendall, and Gardner^ identified four flow regimes where viscous fingering and gravity segregation occur. Stalkup showed their characteristic profiles (Figure 2.22) and estimated their volumetric displacement efficiencies at breakthrough for mobility ratios of 1.35, 6.5 and 27 over viscous/gravity ratios ranging from unity to 100,000 (Figure 2.23). In Region I of Figure 2.22, the geometry of the gravity tongue changes and the vertical displacement efficiency increases as a function of the viscous/gravity ratio, while in Region II the geometry of the gravity tongue changes but the vertical displacement efficiency does not change with the viscous/gravity ratio. In Region III secondary viscous fingers begin to form on the gravity tongue and the vertical displacement efficiency increases as the viscous/gravity ratio increases, while in Region IV viscous fingers dominate the displacement process and the vertical displacement efficiency is again independent of the viscous/gravity ratio. An analysis of a CO2 flood in terms of these flow regimes can be a useful predictive tool for the

23 recovery and to determine the extent of macroscopic bypassing. Since the viscous/gravity ratio can be controlled to an extent by the flow rate, such an analysis could also determine to what extent the operator can influence the

recovery by adjusting the flow rate, especially for single-digit mobility ratios

(Figure 2.23).

Perkins, Johnston, and Hoffman^^ studied viscous fingering and observed that (1) transverse flow occurred only near the front ends of the viscous fingers,

(2) finger widths were about half of their lengths, (3) fingers could be suppressed

by transverse dispersion, and (4) the widths of the extending viscous fingers were close to the minimum length of viscous finger that could grow at any stage

of the displacement. Their findings included that there existed an initial region

devoid of viscous fingers due to suppression by longitudinal dispersion, the

lengths of the viscous fingers were proportional to the mean displacement, and

the widths of the fingers were proportional to the square root of the mean displacement. They also found that in radial systems where the mean displacement was large compared to the wellbore radius, the number of viscous fingers was constant.

2.3.4. Dead-End Pores

Coats and Smith^ studied the effects of dead-end pore volume (DEPV)

(Figure 2.20), concluded that dispersion coefficients based on a diffusion model would be too large to account for the DEPV dispersion mechanism, and

24 proposed an alternate differential capacitance model to arrive at a more accurate dispersion coefficient. BakeP^ later modified this capacitance model and developed a method of rapidly matching this model to experimental data to model the unit-viscosity ratio, laboratory-miscible tests over a wide range of velocities. He found that the dispersion coefficients and fluid transfer coefficients were dependent on velocity while the fractional flow of oil was independent of velocity, and concluded that the capacitance effects due to DEPV might produce longer mixing-zone lengths and increase the solvent requirements.

2.3.5. Miscibility Development with High Water Saturation

Stalkup''^ studied the displacement of oil under conditions of high water saturation using a propane solvent and concluded that water blocked a portion of the oil from the propane, which rendered the oil immobile, but that the hydrocarbons would still diffuse into the solvent and produce a significant oil recovery. He also concluded that oil trapping by water saturation may be less pronounced in pores that were more intermediate-wet than in ones that were more water-wet, and that the longitudinal dispersion coefficient as well as the amount of bypassed oil increased as water saturation increased.

Shelton and Schneider'" also observed that in oil displacement under conditions of high water saturation in water-wet rock the amount of trapped oil correlated with the hysteresis envelope between the drainage and imbibition relative-permeability curves, and in oil-wet rock the amount of recoverable oil

25 was greater than in water-wet rock. They concluded that the transition zone between crude oil and a displacing solvent phase in an MCM process was very short and that an increase in the displacement rate decreases oil recovery within the swept region. In addition, they also found that the flooding response in long cores for an MCM process using CO2 or rich gas was equivalent to that of liquid solvents in an FCM process.

Tiffin and Yellig''^ studied water-alternating-gas (WAG) processes in 8 foot cores and concluded that the overall oil recovery did not change between a secondary CO2 flood to a tertiary flood. In water-wet tertiary tests where water and CO2 were simultaneously injected, there was a significant amount of oil trapping and interference with the miscibility process that resulted in a lower oil recovery, while in oil-wet tertiary tests the simultaneous injection of water and

CO2 did not affect either the recovery or the development of miscibility (Figures

2.24 and 2.25).

Koval*^ developed a method to predict the oil recovery and solvent cut as a function of pore volumes injected (Vpi) and what he termed the K-factor (K) for unstable miscible displacements in heterogeneous media, based in part on the

Buckley-Leverett method.^ For Koval's method, the fractional flow of the solvent, fs, is a function of the solvent saturation and the K-factor:

26 The K-factor is in turn a product of the reservoir heterogeneity factor, H, and the viscosity ratio factor, E. Where the reservoir heterogeneity is equal to 1, the K- factor is expressed using the quarter-power mixing rule as follows:

K=[0.7% + 0.22(Mjiu,f't. (2.25)

The recovery between breakthrough and total recovery is then calculated as follows:

^^ 2(i:/F,)"'-l-F,

Figure 2.26 is similar to the fractional flow curve used in the Buckley-Leverett method but shows the fractional flow of the solvent, fs, as a function of the number of pore volumes of solvent injected Vpi, rather than the saturation of the solvent.

2.3.6. Areal Displacement Efficiency

Claridge''^ later developed a correlation for estimating oil recovery that combined the existing methods of predicting areal displacement efficiency and applied them tofive-spot patterns . His correlation is shown in Figure 2.27 where areal displacement efficiency, EA, is given in displaceable pore volumes (PVs) of oil produced as a function of displaceable PVs of fluid injected, Fi, and the mobility ratio, M. The lower "base" curve labeled FJBT versus M shows the areal sweep efficiency at breakthrough as a function of the mobility ratio, regardless of

27 the amount of fluid injected, while the remaining curves show the areal sweep efficiency at a set number of PVs injected as functions of the mobility ratio.

2.3.7. Water Injectivity

Shelton and Yarborough"^ studied phase behavior in MCM processes to address problems associated with asphaltene deposits and water injectivity during CO2 and rich-gas floods. Based on the work of Harvey et al.,"^ it was understood that injectivity reduction could be caused by a few percent residual oil saturation behind the injection flood front. This small amount of residual oil was characterized by an unusually high trapped rich-gas saturation and a subsequent decrease in water relative permeability. Shelton and Yarborough concluded that multiple liquid phases were likely to form when CO2 was injected into reservoirs of relatively low temperature, and that the precipitation of hydrocarbon solids as well as an oil-rich liquid phase was probable when the oil was ashphaltic. They also concluded that the amount of solid dropout in a WAG process was proportional to the growth of a transition zone between the oil and CO2 at the flood front and that the residual oil saturation could produce three-phase flow effects and reduce the relative permeability to water (Figure 2.28). Mungan'*'' concurred, citing the problem as a major consideration in designing a CO2 flood and compared the precipitation process to that of the deasphalting of crude oil by

LPG or other solvents. He noted that this could be especially problematic in multi-porosity systems that occur in limestone reservoirs and recommended

28 various test that could be performed before a field performance be carried out.

These tests include a comparison of filtering residue from a representative crude oil with that of a C02-contacted crude oil, conducting a PVT analysis on a

C02/crude oil mixture, or a slim tube test at reservoir conditions.

Low water injectivity can also be the result of reservoir and/or water problems. Leone and Scott''^ cited clay mineralogy to be the cause of fines migration once a critical injection velocity in a coreflood had been reached.

Cusack et al.''^ observed that a biomass film could build up in the vicinity of the wellbore as a result of nutrients in the injected water and create significant injectivity losses. Kumar^ also noted that fine particles in the injection water supply could be "strained" through the reservoir and result in permeability losses.

2.4. Displacement Efficiency

The overall displacement efficiency,^ £, also referred to as simply the displacement efficiency, is the product of the microscopic and macroscopic displacement efficiencies, ED and Ev, respectively:

E = E^xEy . (2.27)

Microscopic displacement efficiency is the fraction of the oil that has been mobilized in a recovery process and pertains to the swept volume only, while macroscopic displacement efficiency is the fraction of the total hydrocarbon- bearing volume of the reservoir that has been contacted by the displacing fluid.

Gardner, Orr and Patel^' stated that bypassing, relative permeability, phase

29 behavior, and dispersion influence the displacement efficiency of a CO2 flood, and that bypassing can occur on both the microscopic scale and the macroscopic scale.

2.4.1. Microscopic Displacement Efficiency

Within a swept volume, microscopic displacement efficiency is mathematically defined as follows:

£0 = ^%^"' • (2.28)

Conceptually, Green and Willhite^ defined microscopic displacement efficiency as

"a measure of the effectiveness of the displacing fluid in moving (mobilizing) the oil at those places in the rock where the displacing fluid contacts the oil" (2). They relate it to the displacement or mobilization of oil at the pore scale with the added condition that it is "reflected in the magnitude of the residual oil saturation, Son in the regions contacted by the displacingfluid." Their use of the term "regions" allows not only the pore spaces that were contacted by the displacing fluid to be included, but also neighboring pore spaces that were not contacted. Thus, where bypassing has occurred amongst a local number of pores, i.e., where one pore was bypassed while another was not due to a difference in the size of their pore throat openings, this would still be treated as a microscopic phenomenon while viscousfingering an d gravity segregation would be considered to be macroscopic phenomena.

30 Although bypassing by CO2 can occur whenever there is interfacial tension between the oil, water and CO2 phases, hydrocarbon mobilization may

still be possible. Within an individual pore that is water-wet, immiscible (i.e.,

physical) displacement will no longer occur once the oil forms into a droplet within the pore and does not deform enough to squeeze through the pore throat

opening. Within a pore that is oil-wet, a limit will also be reached when the

displacing fluid can no longer deform the oil that coats the pore wall and cause it to move out of the pore. At this point, however, miscible gas drive processes can

still cause the displacement of hydrocarbons to occur by extracting them into the

more mobile CO2 phase.

Bypassing between neighboring pores can be caused by pore structure

characteristics such as dead-end pores or a difference in the size of pore throat openings between neighboring pores. Where interfacial tension is high, the capillary forces dominate the recovery process and the displacingfluid flows preferentially through the smaller pore openings, bypassing the larger ones.

Where the reservoir conditions are miscible enough for interfacial tension to be lowered to the point where viscous forces can dominate, and the displacing fluid will flow preferentially through the larger pore openings, bypassing the smaller ones.

31 2.4.2. Macroscopic Displacement Efficiency

Macroscopic displacement efficiency, Ev, is defined mathematically in terms of the areal and vertical sweep efficiencies, EA and E/, respectively, as follows:

Ey=E^xE, . (2.29)

Conceptually, the macroscopic displacement efficiency is that portion of the reservoir volume that has been contacted by the displacing fluid as a fraction of the total reservoir volume. The areal sweep efficiency, EA, is conceptually defined as the area of the reservoir contacted by the displacing fluid when observing it from the plan view, as a fraction of the total area of the reservoir. This means that at whatever position in the reservoir theflood front has reached, it is assumed that variations in the shape of theflood front may be in the horizontal plane only, while the flood front is vertical at every point. Vertical sweep efficiency, E/, in turn can be defined conceptually as the volume that has been contacted by the displacing phase divided by of the product of the area of the reservoir contacted by the displacing fiuid when observing it from the plan view and the height of the reservoir.

Macroscopic bypassing can occur by viscousfingering, gravity segregation, and channeling. Viscousfingering occur s when there is a high mobility ratio, that is, when the mobility of the displacing fluid. Do, is significantly greater than the mobility of the displaced fluid, Dd- The basic mobility ratio is defined as follows:

32 M = ^. (2.30) Xj"d

For cases where there is no interfacial tension the two phases have equal permeability, and equation 2.26 reduces to the viscosity ratio between the displaced and displacing phases:

M = ^. (2.31)

MD

For a waterflood with piston-like displacement, only water is flowing behind the flood front, and only oil is flowing ahead of the flood front. Thus, the relative permeability of the formation to water is the relative permeability to water behind the flood front at the residual oil saturation, while the relative permeability of the formation to oil is the relative permeability to oil ahead of the flood front at the interstitial water saturation. In this case, the equation for mobility is written as the following:

M = (2.32)

Craig^^ defined a mobility ratio for the case where there is mobile water initially in the formation as well as mobile oil behind the flood front. Craig's mobility ratio is as follows:

M,=^ (2.33) fSd

33 where (A^)^-^ is the mobility of the water at its average saturation behind the flood front, and (A^)^^ is the mobility of the oil at its average saturation ahead of the flood front. Stalkup^ added a further refinement to this by basing the definition of mobility ratio on the total mobility of the oil, water and gas phases ahead of the flood front and the total mobility of the oil, water and gas phases behind the flood front: ^'-jt where

(k ^ M-s.='L -^ (2.35) and

^K^ (^Ji.=S (2.36)

Stalkup*..3^3 recommended this definition for "...characterizing the mobility ratio between an oil bank and the solvent displacing the oil bank when mobile water is present in either region" (32). Stalkup also pointed out that where multiple injection slugs are used, the mobility ratio at any particular displacement front depends not only on the mobilities of all the phases both ahead of and behind the displacement front, but the mobilities of all phases surrounding the other fronts as well. No unique method of defining the mobility ratio in such a case has been made. Green and Willhite^ caution that when using correlations based on the

34 mobility ratio, the particular definition for the mobility ratio used must be consistent with the definition upon which the original correlations were based.

A mobility ratio of more than one indicates, via Darcy's equation, that the velocity of the displacing phase is greater than that of the displaced phase. This difference takes effect when a perturbation, or pore level heterogeneity, in the reservoir is encountered, and the displacing phase begins to surge ahead of the displacement front. Such a perturbation could consist of a sufficient inhomogeneity in the permeability in the direction of flow, which would cause the velocity to increase according to Darcy's equation. As more displacing fluid passes through the perturbation it would continue to cause the viscous finger to grow.

Bypassing on the macroscopic scale can also occur due to gravity segregation, which is the result of the difference in density between the displaced and the displacing fluids. When the displacing fluid is less dense than the displaced fluid, the displacing fluid will tend to override the displaced fluid, and when the displacing fluid is more dense than the displaced fluid, the displacing fluid will tend to underride the displaced fluid. These phenomena are known as gravity override and gravity underride, respectively (Figure 2.29). Changing the injection rate can cause the viscous forces to dominate the process and control gravity segregation. The optimum injection rate is determined by the viscous- gravity ratio:

35 f uu V/'i ^'^=\TT-\I:\ (2.37) where the two controlling parameters are u and p. The denominator of the viscous/gravity ratio contains the pressure differential between the two fluids,

Apgh, while the remaining terms are equivalent to the pressure term on the righthand side of Darcy's equation in the numerator.

Since Darcy's equation shows how the rate of flow varies directly with permeability, rate of flow in a CO2 flood is greater in anisotropic regions where the permeability is greater. A reservoir can have an uneven vertical permeability distribution due to the deposition cycle of the grains that comprise the rock matrix. This can result in a horizontal channel of high permeability lying adjacent to at least one horizontal channel of low permeability. When the pressure gradient is horizontal, the displacing fluid will therefore flow faster through a channel of high permeability and bypass the displaced fluid in the channel of low permeability. If a vertical pressure gradient can be created at some point in fime by shutting in the producing wells and continuing to inject into the injection wells, the displacing fluid will be able to contact the oil in the low permeability channels.

Well spacing is also significant to the areal displacement efficiency, but is not discussed at length here. The most common well pattern is the five-spot, which has a producer/injector ratio of 1:1, and is characterized by producing wells at the nodes of a square grid pattern and injectors at the center of each of the squares (Figure 2.30).

36 2.5. Field Applications

Hadlow^ reported on the state of the U.S. petroleum industry experience with CO2 processes in 1992 and summarized the innovations being used to improve recovery in CO2 processes. At the time of his writing, the industry was injecting about 2.4 BCF/D of CO2 into 45 active projects, 40% of which was produced and reinjected. The overall incremental response of 142,000 BOPD resulted in a CO2 utility of about 17 MSCF CO2 per incremental barrel of oil.

Experience to the time of his writing included the following observations:

• Significant production response was attainable through improved reservoir

management prior to CO2 injection.

• The response in most CO2 recovery programs exceeded expectations.

• CO2 retention (lack of premature breakthrough) was excellent and in

agreement with predictions.

• Estimates of CO2 utility had been accurate.

• The average injectivity loss in WAG projects was 20%.

• The ultimate recovery would likely exceed expectations.

The most frequently implemented methods of CO2 EOR are the WAG and the continuous flooding processes, A number of successful CO2 hufTn'puff processes have also been documented, although comparatively little literature has been written on them. One feature of the CO2 hufTn'puff process that appears to be very significant is that it has achieved greater recoveries when it is

37 implemented at reservoir conditions that are below the MMP, unlike the continuous CO2 flood or WAG processes. However, the CO2 huff'n'puff process is also used as a method of stimulating production wells prior to a continuous

CO2 flood and is not used solely as an EOR method.

2.5.1. Continuous CO2 Flooding

Continuous CO2 flooding is often excluded from use in favor of WAG, since WAG offers more mobility control and can increase the volumetric sweep efficiency substantially while lowering the volume of CO2 required for the process. However, in regions where the injection water is incompatible with the reservoir or formation water WAG may not be feasible, leaving continuous CO2 flooding as the optimal recovery method. Continuous flooding is also studied extensively in the lab as a means of understanding the mechanisms involved in the CO2 injection phase of the WAG process. Table 2.1^ shows a comparison between a number of field performances of successful continuous CO2 floods and WAG processes that have been documented where significantly lower recoveries have occurred in the WAG processes.

Pontious and Tham^^ reported on a secondary CO2 continuous flood in the

Crosset Field in Crane and Upton counties of West Texas where the reservoir permeability was too low to be feasible for water injection. The response was considered "definite and encouraging" although not as good as expected. The process was multi-contact miscible rather than first-contact miscible, and it was

38 concluded that in order to predict the performance accurately, a simulation was needed to predict phase behavior, compositional changes, and the essential properties of the resulting phases.

Yellig^ conducted an extensive laboratory study of CO2 displacement of

Levelland oil to support CO2 pilots being carried out in that area. He found the

MCM process to dominate above the MMP, where a vaporizing gas displacement occurred. It was also found that the greatest length required for miscible displacement efficiency to develop took place where single contact mixtures of

CO2 and oil formed two liquid phases.

In 1992, Mizenko^^ reported on a continuous CO2 flood in the North Cross

Devonian Unit CO2fiood tha t began twenty years eariier. A core was acquired from the swept portion of the reservoir, which indicated that the residual oil saturation was 3% of the pore volume, and a high recovery through an MCM process was reported from areas of the reservoir whose initial pressure was significanfly below the MMP. It was estimated that 43% of the OOIP was recoverable via the continuous CO2 process to bring the total recovery to 63%.

Stephenson, Graham, and Luhning^ reported on the performance of

Canada's first miscible CO2flood, which took place in the Joffre Viking field. The field was abandoned in the mid-1960s after the economic limit of primary recovery had been reached at about 42% of the OOIP. In the mid-1980s after analyses with lab tests and simulations were performed, the field was determined to be suitable for a miscible CO2 recovery program. However, conditions leading

39 to viscous fingering and gravity override were present: C02/oil and C02/water mobility ratios were 22 and 23, respectively, at reservoir conditions, and the CO2 density was 60% and 80% of water and oil, respectively. Although mobility control was needed it was also known that a WAG program could reduce the recovery via water shielding. A series of experimental pilots were therefore conducted to compare the performance of continuous CO2 flooding, WAG, simultaneous C02/water injection, and the effect of foam-generating surfactants on CO2 mobility. The reservoir oil saturation was comparable to that of a post- secondary residual oil saturation, and the continuous CO2 flooding program started in mid-1985 (Figure 2.31). The fractional flow of the oil was between 5% and 7% of the total fluid flow until late 1986 when the oil bank reached the producing wells, and the oil fraction rose steeply to peak at over 60% in early

1987. The CO2 fraction began to rise when the oil fraction declined accordingly, until the CO2 fraction peaked at over 70% in late 1988 while the oil fraction troughed at less than 20%. In June 1988, the program was switched to a waterflood, and the CO2 fraction declined very steeply to less than 5% by eariy

1989, then continued to decline slightly. This steep decline in the CO2 fraction caused the oil fraction to increase correspondingly and peak at almost 40% in eariy 1989, after which it declined steadily to about 10% in 1990. By the end of

1991 the oil recovery was 13% OOIP with 41% HCPV CO2 injected, 40% which had been recovered. The ultimate recovery was estimated at 18% OOIP, and it was concluded that C02-slug injections establish a high-C02-saturation flow path

40 at the top of the formation. The CO2 remains virtually immobile during water injection, causing a sharp reduction in produced CO2, but when CO2 injection is restarted, flow through the C02-saturated region is quickly reestablished.

Kumar and Eibeck^^ reported on a CO2 tertiary pilot in the Garber field,

Garfield County, Oklahoma, which had been previously waterflooded. Water injection preceded the CO2flood i n order to raise the reservoir pressure, then a

35% HCPV slug of CO2 followed as a straight slug. At the time of their writing

70,000 STB of oil had been produced, and an estimated 14% OOIP recovery was expected from the EOR treatment (Figure 2.32). The success of the project led the authors to conclude that CO2 recovery projects were possible at depths of less than 2000 feet, although they did not identify the recovery mechanisms.

Perry^ reported on a gravity stable miscible CO2 project conducted in a reservoir at a depth of over 12,000 feet. In thisflood, a CO2 slug was injected at the gas-oil contact and moved downward through the watered-out sand by the production of downdip water. It was concluded that since a substantial amount of oil remained from the primary natural water drive stage of production, a substantial amount of oil could be recovered from this type of displacement process.

Bellavance^' reported on the Dollarhide Devonian CO2flood that was started TO years eariier in 1986, where water injection was limited to areas of severe CO2 breakthrough due to the detrimental effects of WAG cycles on injectivity. After 11.2% HCPVof CO2 injected, only 15% had been produced; thus

41 C02 breakthrough was not severe. While primary and secondary recovery had together produced 43% OOIP, tertiary recovery was expected to produce 14%

OOIP with infill drilling producing an additional 5% OOIP. It was concluded that the continuous CO2flood was preferred over the WAG process, which was detrimental to productivity for this field and that the overall tertiary production rates were expected to be about 250% of the pre-flood production rates.

2.5.2. Water-Alternating-Gas (WAG)

Most of the EOR projects that use CO2 recovery methods are of the water- alternating-gas (WAG) type. Although a WAG process offers mobility control that a continuous CO2flood does not, the microscopic displacement efficiency of a

WAG process is usually less than that of a CO2 continuous flood due to water shielding. However, the more favorable mobility ratio of the WAG process appears to be sufficient in most cases to increase volumetric displacement efficiency of the injectedfluids to compensate for the losses in microscopic displacement efficiency.

2.5.2.1. WAG Theoretical and Laboratory Models

Tiffin and Yellig"^ investigated the effect of water blocking by mobile water on oil recovery where a C02/water mixture was injected in both water-wet and oil- wet cores. They determined that a pronounced difference between the two types

42 of cores existed due to the greater exposure of the oil phase to CO2 in oil-wet

cores than in water-wet cores over a range of residual oil saturations (Figures

2.24 and 2.25). Huang and Holm.^ Stalkup,''^ Raimondi and Torcaso,®^ as well as Lin and Huang®^ determined similarfindings. Raimond i and Torcaso^^

introduced a correlation for the phenomenon of oil trapping before and after a solvent flood in the presence of mobile water in water-wet rock. Oil trapping occurs when interfacial tension causes oil droplets to form inside the pores rather than deform and move through the pore throat openings. Raimondi and

Torcaso's correlation is based on the following equation:

^-'^ = l + akilk (^-^^^ ro rw where Sor is the residual oil saturation following waterflooding, Sor.wb is the residual oil saturation following the CO2flood in the presence of mobile water and a is an empirical constant that ranges from 1.0 in cases of strong water blocking and thus oil trapping, to values in the order of 100 in cases of relatively little water blocking. Chase and Todd^ used this correlation in numerical simulations, and Lin and Huang^ modifled it as well to model oil-wet and mixed- wet rock in addition to water-wet rock.

The injection of a single C02/water phase has also been modeled because the calculations are simpler than with alternate CO2 and water slugs.

Stalkup^ developed equations that could optimize this process, where if too much water was injected, the water moved faster than the solvent, causing a

43 high water saturation at the solvent/oil interface and resultant oil trapping. If too little water was injected, a solvent bank would form ahead of the water front, leading to viscousfingering. Gree n and Willhite^ derived an equation based on the work of Walsh^^ that would allow the water/solvent injection ratio to be calculated for equal water/solvent velocities in linear systems whose initial water is immobile. The solution for the fractionalflow o f water, fw, is as follows:

J- S„-S^ (2.39)

The solutions for fw and Sw can be found graphically (Figure 2.33) where the fractional flow curves for the oil/water and solvent/water systems are plotted on the same graph and a line is drawn from the upperright hand corner where fw

= Sw = 1 to the ordinate axis, passing through the point tangent to the oil/water fractional flow curve just above the ordinate axis. The point where the line passes through the solvent/water fractional flow curve yields the values of fw and Sw, where fw is the fraction of injected water, thus determining the water/solvent injection ratio. This solution also provides a basis forfinding the new mobility ratio.

2.5.2.2. WAG Field Implementations

WAG projects have been quite successful in the field, but the resulting recoveries have generally been less than that of continuous CO2 floods.

Brockmeyer et al.^ reported on a CO2 WAG project at the Lost Soldier Tensleep

44 field in Wyoming where an estimated 13% OOIP would be produced by tertiary recovery. He concluded:

• reservoir management was critical to optimizing the project

• the evaluation of all the components of the production system

could provide significant improvements in productivity, and

• the reservoir could be re-pressured to achieve MMP

In 1988, Langston et al.°^ reported on the CO2 flooding response in the

SACROC unit at the Kelly field in Scurry County, Texas in two multi-pattern areas that included the 600-acre Four Pattern Area (4PA) and the 2700-acre

Seventeen Pattern Area (17PA). A CO2 WAG program for each pattern areas

(PA) commenced in 1981, and the estimated recovery in the 4PA was 9% OOIP, while that of the 17PA was 5% OOIP (Figures 2.34, 2.35 and 2.36).

The results of several successful WAG and continuous CO2 flooding projects are summarized in Table 2.1 ,^ where in all cases the slug sizes were greater than 15% HCPV. The average recovery in the single-slug projects was

15% OOIP whereas the average recovery in the WAG projects was 8.2% HCPV, which reflects, to a significant extent, the reduction in microscopic displacement efficiency for WAG processes as compared to continuous CO2 floods. Tanner et al.^ studied the performances of a 1:1 WAG process and a single-slug CO2 displacement in the Denver Unit of the Wesson field of West Texas. These were carried out in side-by-side areas for total injection of 40% HCPV in both cases.

The cumulative response to the continuous CO2 injection was higher at first, but

45 high GOR's in some wells forced them to be shut in, and the WAG treatment ultimately yielded a higher recovery. As a result, the operators considered implementing a hybrid process called the Denver Unit WAG (DUWAG) consisting of a continuous CO2 injection for 4 to 6 years followed by a 1:1 WAG. A comparison between the three methods is shown in Figure 2.37.

2.5.3. Cyclic CO2 Stimulation (Huff'n'puff)

The CO2 hufTn'puff process was originally developed to provide an alternative to the hufTn'puff steamflood in heavy oil reservoirs, where the viscosity must be lowered sufficiently to flow to the wellbore. Both of the hufTn'puff processes were used as alternatives to continuous drive processes, which required at least two adjacent wellbores as opposed to a single wellbore.

Thus, one of the principal advantages of a hufTn'puff process is that a lower capital investment is required before returns are realized. Another significant benefit of the CO2 hufTn'puff process, with respect to this study, has been that the mechanisms of recovery that dominate a successful CO2 hufTn'puff process show potential to help optimize a continuous CO2 flood.

2.5.3.1. CO? Huff'n'puff in Heavy Oil Reservoirs

In1979 Patton. Coats, and Spence^^ performed numerical CO2 hufTn'puff simulafions to determine the volume of CO2 to use per cycle and the optimal

46 number of hufTn'puff cycles for a heavy oil reservoir. They concluded that the adverse C02/heavy oil mobility ratio and the small mobile water saturation present would assist the distribution of CO2 in the reservoir, that after a soak period the oil around the wellbore would be mobilized due to reduced viscosity, that the free CO2 around the wellbore would result in a low WOR during production, that three cycles would optimize production, and that a CO2 hufTn'puff process would produce comparable results in a conventional oil reservoir in terms of CO2 utility, i.e., MSCF CO2 injected/STB oil produced and

MMSCF CO2 injected/foot of pay. In 1980, Patton, Sigmund, Evans, Ghose and

Weinbrandt' designed a cyclic CO2 process for a heavy oil reservoir based on these principles, hypothesizing that the mechanisms of displacement would be oil swelling, viscosity reduction, and a solution gas drive from the C02-saturated oil.

The CO2 utility ranged from 4 to 8 MSCF/STB on the first cycle with two treatments per well appearing to be optimal.

Khatib, Eariougher and Kantar^° studied the cyclic CO2 hufTn'puff process in heavy oil reservoirs and described the process as CO2 dissolving in the oil as the reservoir pressure, i.e., miscibility, built up, which allowed the mechanisms of oil swelling, viscosity reduction, and pressurized gas drive to develop during this stage as well as the soak period. They concluded that two hufTn'puff cycles were optimal and that enough CO2 should be used to saturate the reservoir oil and water. Sankur and Emanuel^' also conducted a laboratory study of heavy oil recovery and concluded that a CO2 huff'n'puff process would be effective where

47 the pressure increase during injection aided the process, that smaller CO2 slugs reduced the CO2 utilization, and that to add nitrogen to the CO2 in a heavy oil project would be detrimental.

Reid and Robinson''^ studied a four-phase heavy oil project that consisted of (1) cyclic CO2 stimulation at the producing wells, (2) continuous CO2 injection.

(3) WAG, and (4) water injection. They concluded that a hufTn'puff stimulation on the producing wells improved the oil rate, reduced the WOR, contacted otherwise unavailable residual oil, raised reservoir pressure and further reduced the oil viscosity, utilized production mechanisms not offered through continuous flooding, and allowed wells to be produce more oil by natural lift. In 1992,

Bakashi, Ogbe, Kamath and Hatzignatiou^^ studied the feasibility of cyclic CO2, both in the laboratory and in the field, for resen/oir crudes ranging from 14° API to 22.5° API and determined that a optimal bottom hole pressure existed above which production decreased due to a low drawdown pressure and a rise in miscibility with CO2. They did not find varying the soak period to appreciably change the recovery, but raising the slug size increased the recovery and lowered the CO2 utility as well as the incremental recovery on each added unit of

CO2 used. They also found that over a period of one year, the same amount of

CO2 injected in a single-cycle stimulation obtained a better yield than a two-cycle stimulation (Figures 2.38 and 2.39).

48 2.5.3.2. CO? HufTn'Puff in Cnnvfintional Oil Reservoirs

2.5.3.2.1. Numerical Simulations

in 1986 Hsu and Brugman^'' performed numerical simulations of a CO2 hufTn'puff process to determine the optimal number of cycles, the timing of each stage of the process, and CO2 quality, or purity. They concluded that the major miscibility process consisted of the vaporization of intermediate hydrocarbons into the CO2 phase, that the soak times which varied from 5 to 40 days had little effect on the recovery, that up to a 20-mol % nitrogen content had little effect on the recovery, and that the most important parameter of the process was the quantity of injected CO2. They also found that the incremental recovery dropped between the first two cycles, and a third cycle did not appear to be feasible. Their conclusions on soak periods and CO2 quality are significant in that they do not appear to correlate with coreflood andfleld data analyses and raise questions as to the capability of numerical simulators, which were designed to simulate a continuous CO2 process, to simulate a CO2 hufTn'puff process.

2.5.3.2.2. Analyses of Laboratory and Field Data

In 1986 Monger and Coma'^^ analyzed coreflood andfield tests on CO2 hufTn'puff processes for light crude reservoirs and concluded that lower oil viscosities, viscousfingering and near-miscible pressures were desirable. They found that the recovery factor increased with the amount of CO2 injected and that

49 C02 utility on the first and second cycles was less than 2 MSCF CO2 per STB oil.

They did not find the durations of the soak periods to be significant but found that when water was injected following the CO2 injection the recovery increased, possibly because excess CO2 was able to contact oil deeper in the cores. Figure

2.40 shows the changes in saturation they observed over three CO2 hufTn'puff cycles plus a water injection cycle, while Figure 2.41 shows a linear relationship between the ultimate incremental oil recovery and the CO2 slug size.

In 1988. Monger, Ramos and Thomas^® conducted a laboratory and field investigation for a light-crude oil recovery in a pressure depleted reservoir, evaluating sixty-five single-well cyclic CO2field tests. They concluded that the

CO2 hufTn'puff was feasible with an average CO2 utilization of less than 1

MSCF/BBL, and that the process reduced the WOR. They concluded that oil recovery improved with low quality CO2, evenly distributed initial gas saturation, and a low reservoir pressure that would allow more reservoir space to be occupied by a unit mass of CO2. Thomas and Monger-McClure^ concluded in a later study that the recovery factor in a CO2 huff'n'puff was proportional to the mass of CO2 injected, that a treatment radius up to 150 feet would allow the recovery to be optimized, and that shallow reservoirs, sub-MMP pressures, thicker pay zones, as well as mobile water or free gas saturations were desirable.

They also concluded that high water cuts were not necessary and that the ultimate recovery was improved with an extended soak period and backpressure during eariy production.

50 In 1990, Thomas, Berzins, Monger, and Bassiouni^° performed experimental and numerical tests on cyclic CO2 stimulation and concluded that gas caps, high residual oil saturation, and gravity segregation, which helped the

CO2 penetrate deeper into the reservoir, improved recoveries. In 1992 Karim,

Berzins, Schenewerk, Bassiouni, and Wolcott^^ analyzed the influence of nitrogen drive (or chase) gas, injection rate, and reservoir dip on cyclic CO2 stimulation core tests and determined that moderate injection rates were optimal, dip angle was significant and downdip injection was preferable to horizontal or updip injection. They also found that when a CO2 injection was followed by a nitrogen injection, twice as much oil was recovered on the first cycle and three times as much oil on the second cycle than would have been recovered if CO2 alone was injected. In addition, they found that low injection rates inhibit bypassing and lower the recovery, and concluded that near-miscible conditions were preferable to miscible conditions. In 1996 Shayegi, Jin, Schenewerk and

Wolcott°° also found that certain mixtures of CO2, methane, and nitrogen in cyclic stimulation could yield up to three times better recoveries than any pure component by itself.

Hara and Christman^' investigated CO2 hufT'n'puff recoveries in diatomite cores and found that when the rate of diffusion is significant compared to the rate of flow, bypassing was diminished and more of the reservoir could be contacted, and concluded that the process could displace a significant amount of oil under immiscible conditions. In 1986, Haskin and Alston^^ studied 28 field tests of a

51 C02 huff'n'puff process and found that soak periods longer than 10 to 17 days did not significantly improve performance but that greater incremental recoveries were achieved with larger injections of CO2. Similar observations as to the response to soak periods have been observed by other authors, but explanations in many cases have not been offered. Haskin and Alston found the CO2 hufTn'puff process to be relatively fast, relatively inexpensive, and effective under conditions of poor interwell communication. They also concluded that viscosity reduction and oil swelling dominated the recovery process, and developed a predictive method based on these two mechanisms. In 1991, Adamache,

Kantzas, Mclntyre, and Sigmund^^ investigated "cyclic interruptions" in a vertical

(downward) miscible displacement project where shutting in the production wells during plant turnarounds resulted in reversals of declining oil production and increasing GORs. Numerical simulations and corefloods also verified this, showing incremental oil recoveries of 8-14% pore volume and significantly lower

GOR's compared to continuous miscibleflooding condifions .

2.5.3.2.3. Full-Scale Field Implementation

In 1986, Palmer, Landry, and Bou-Mikael^'' implemented a CO2 huff'n'puff program for eleven wells infive fields i n south Louisiana. In most cases 500 tons

(8.6 MMSCF) were injected over a 24-hour period, as fast as possible to promote fingering and increase C02/oil contact. The wells were then shut in for two to four weeks. In some cases the oil rate increased tenfold after injection, and two years

52 later incremental increases were still recorded. They hypothesized that the main displacement mechanisms were oil swelling, viscosity reduction, and IFT reduction. Figure 2.42 shows the increase in the recovery rate in this project from

1984 to 1986. In 1989, Bou-Mikael and Palmer®^ concluded that the single most important criteria to be met for the economical formation of an oil bank in a CO2 hufTn'puff was a minimum displaceable oil saturation (MDOS). They also concluded that the cyclic stimulation of the producing wellbores improved recovery efficiency, response time, cashflow, an d that gas channels as well as

C02/nitrogen/methane gas mixtures were beneficial to a CO2 hufTn'puff process in depleted reservoirs because they allowed deeper reservoir penetration prior to phase changes under higher post-injection pressure conditions. They found that after sufficient injection, the reservoir pressure increased to where an oil bank could form and that a 45-day soak period was sufficient. At the time of their writing, a minimum of three CO2 huff'n'puff cycles had been planned.

In 1990. Miller^^ investigated a CO2 hufTn'puff process that was being performed on 200 light stripper wells and found that the oil production inaeased in a cost-efficient manner that was more favorable than other CO2 methods. His approach to optimizing the process was to create bypassing via fingering, channeling, and that initial gas saturation and fractured formations were desirable since the process was immiscible and a large volume of oil needed to be contacted. He found that the fractional fiow characteristics and relative permeability hysteresis curves favored oil mobilization. In 1991, Bardon, Coriay,

53 Longeron, and Miller°^ also found unexpected production increases in the CO2

huff'n'puff process was not explained by oil swelling and viscosity reduction

alone, and it was concluded that solution COagas drive, changes in IFT and wettability, and the presence of C02-saturated water contributed to the favorable

production. In 1994, Miller^ reported that the project had expanded to 390 CO2 hufTn'puff cycles on 240 wells, and concluded that the economics of a CO2 hufTn'puff program will be the most favorable if the treatments are initiated well in advance of reaching economic production limits. In 1998, Miller^^ provided another update in which, due to the escalating price of CO2 the treatments were now being performed with flue gas and nitrogen-enriched gas, which were both found to be economically feasible even though longer soak times were needed.

2.5.4. Summary

In 1989, Brock and Bryan^ summarized the results of cyclic CO2 flooding between 1972 and 1987. They cited many of the observations given thus far including some apparent contradictions and concluded that good data management was necessary for accurate assessments to be made. It is evident that while the same recovery mechanisms are used in the hufTn'puff process that are used in the continuousflooding an d WAG processes, they are applied in a manner that allows different mechanisms to dominate the process. It would appear that the mechanisms that cause extraction to occur from the oil to the

CO2 phase are more prominent in the CO2 huff n'puff process than in the

54 continuous CO2 process, although it can be seen that the physical displacement of the oil operates in this process as well.

55 Table 2.1 - Incremental Oil Recovery from Field CO2 Projects

Lease or Unit LIthology %OOIP Single-Slug 002 Little Creek sandstone 18 Twofreds sandstone 10 Garber sandstone 14 Mead-Strawn sandstone 15 Shannon, West Sussex sandstone 13 Maljamar dolomitic sandstone 18

Average 15

WAG 002 SACROC (Phase 111) limestone 7.6 SACROC (Total) limestone 7.0 SACROC (Pilot) limestone 6.0 Slaughter Estate dolomite 21.0 South Welch dolomite 5.2 Shannon sandstone 2.0 North Meadow Creek Levelland dolomite 8.9

Average 8.2

56 1.0 1.0

0.8 h 0.8

ro.6 •0.6

I 0.4 0.4

I 0.2 0.2

0 20 40 60 80 100 20 40 60 80 100 Water Saturation, %PV Water Saturation, %PV

(a) Wotor-Wet Rock (b) Oil-Wet Rock

Figure 2.1 Relative permeability characteristics of porous media in (a) strongly water-wet rock, and (b) strongly oil-wet rock

Fractional Flow (fv,) of Water Versus Water Saturation (S^) (Theoretical)

0.8 - X^ 0.6 - 0.4 - /

0.2 - n U 1 y 1 1 1 0 0.2 0.4 0.6 0.8 1

Sw

Figure 2.2 Theoretical fractional water flow of water versus water saturation

57 Plot of the Fractional Flow (fj of Water Versus Water Saturation (Sw) (Actual)

1 0.8 0.6 ** 0.4 0.2 0 0.8

Figure 2.3 Actual fractional water flow versus water saturation

Plot of Water Saturation (S^) Versus Dimensionless Distance (XQ) at a Dimensionless Time (to) of 0.2

•i ^ 1 0.8 - 0.6 - 0.4 -

0.2 - 0 - C) 0.2 0.4 0.6 0.8 1

XD

Figure 2.4 Water saturation versus dimensionless distance at a dimensionless time (i.e., pore volumes injected) of 0.2

58 Dimensionless Distance (XQ) versus Dimensionless Time (to) or Pore Volumes Injected

0.5 1.5

Figure 2.5 Dimensionless time versus dimensionless at various degrees of water saturation

Plot of Water Saturation (S^) Versus Dimensionless Time (to) at XD=1

1 n

0.8 -

0.6 - (0 0.4 -

, ^ —1 1 0 - 1 0.2 0.4 0.6 0.8 to

Figure 2.6 Saturation history at XD=1 (Outlet of System)

59 Figure 2.7 Interfacial forces at an interface between two immiscible fluids and a solid

Wagner & Leach 1.0

\^^ Dombrowski - Moore & o . Slobod Du Prey \ * ^'°^"®" Abrams CO 0.5 \ \ Taber )u Prey\

— Nonwetting Residual - ---Wetting Residual 1 I i L. J L 10 •8 10" 10-^ 10r' 2 10^ Nca = up/a

Figure 2.8 Correlation of recoveries of residual phases as a function of Nca

60 2 3 4 5 6 Number of Contacts

Figure 2.9 Experimental and simulated densities from multi-contact miscibility experiments on crude oil

*090

( ) MOl.WT. C,* > (]4II SOOO • (IM)

O Of) 3 4000 •I tnn 5" •(>m (110) as• 9000

3000

YIILIO SPIM77 ris.« 1000 ONIW DAT* • ooi loi noami upoti N.. It . I*" -L J_ X 120 140 lAO UO 200 220 240 «0 •0 100 TIMPIIATUH.f

Figure 2.10 Holm-Josendal correlation of CO2 MMP as a function of temperature

61 4S0O • rARNSWONTH o S 4000- su 3500- • WILMINQTON w FORD ZONC

3000

•noenciTi'F)* 9 2S00

\t9 Ul» • CUT MCAO-«flUM 2000 'CKia CUT 50 M 70 eo 90 100 Cg-C)o COMTEMT OF OIUCJ-CM, WT.%

Figure 2.11 Correlation of CO2 MMP as a function of extractable C5-C30 hydrocarbons present

4500

FARNIWORTH

a. 4000 e « (A 3500

WILMINQTON FORD ZONE

3000 NORTH OUNOAt WEtT POKON a •mOER 3 (in-F) s 2500 z

2000|-«/fci-CMC0T '^MEAO-tTRAWN I I 100 150 200 250 300

MOLECULAR WEIGHT Cj*

Figure 2.12 Correlation of CO2 MMP as a function of the molecular weight of the crude oil

62 o z

z SOLVENT

I}}}})} iyrTrV////^// k •AL

AL«t UNTIL X~d THEN AL-s/t

Figure 2.13 Schematic representation of viscous finger growth in unstable linear miscible displacements

>

Stable Unstable 5 Displacement Displacement

E

Convection Time/Transverse Dispersion Time Ratio. Ki L/vd^

Figure 2.14 Idealized impact of transverse dispersion on ultimate crude oil recovery by CO2 flooding

63 T7TTTT—r- . .

Water [^Oil Solvent [:• /'I Rock I I 100^m

Figure 2.15 Image obtained from a microvisualization experiment

64 Fluid B

1.0 r Downstream Concentration '^'"'" at Fixed X CB 0.5 . step Response Input-* with Mixing

Time, t

Figure 2.16 Miscible displacement of Fluid A by Fluid B, step change in concentration at inlet

Inlet Slug of B

Distance from Inlet. X Figure 2.17 Concentration profiles for the injection of a slug of Fluid B to displace Fluid A

65 Figure 2.18 Dispersion resulting from laminar flow in a straight capillary

, Perfect Mixing —, 1 B—^ —•

vsag^' —• —>-A

Figure 2.19 Dispersion based on porous media being viewed as a series of mixing tanks

66 Figure 2.20 Dispersion resulting from bypassing of fluid trapped in stagnant pockets

Figure 2.21 Dispersion caused by variations in the flow paths in porous media

67 (c) REGION IV

Figure 2.22 Flow regimes for miscible displacement in a vertical cross-section

68 u LEGEND z u •FLUX. BO FT" o f.i • VISCOSITY. CP it 100 L •LENGTH. FT h - HEIGHT. FT k •PERMEABILITY. MO .^i> -SOLVENT-OIL DENSITY DIFFERENCE GMCC

IV • Mc65

M-27

^- IV-

I I l|ll| I I I I IIM| I 1 I I llll| I I I I I III too 1000 tOOOO 100000

VISC0US<5RAVITY FORCE RATIO = 2050 "''^'f''' k 'Aii^i Figure 2.23 Flow regimes in a two-dimensional, uniform linear system

Plot of Oil Recovery Versus Percentage Water in Injected Water and CO2 (Water-Wet Core)

100 ^•^ d 80 ^ 0) > HO 0 u 40 Qa>£ in 20 0 0 10 20 30 40 50 60 70 80 90 100 Percentage Water in Injected Water and 002 (%)

Figure 2.24 Effect of mobile water on oil recovery for CO2 MCM displacements of reservoir crude oil in a water-wet core

69 Plot of Oil Recovery Versus Percentage Water in Injected Water and CO2 (Oil-Wet Core)

100 2 80 I* 60-j o g 40

0 -1 1 1 r T r 0 10 20 30 40 50 60 70 80 90 100 Percentage Water in Injected Water and 002 (%)

Figure 2.25 Effect of mobile water on oil recovery for CO2 MCM displacements of reservoir crude oil in an oil-wet core

Figure 2.26 Solvent cut as a function of pore volumes injected with Kasa parameter

70 1.0 ^JI TTTT 1 1 TTII 1 '' 1 III -

0.8

^ 0.6

i - 5- 5.0 (O 4.0 Ta 0.4 3.0 ^\: "^ s^ I X'2. 0 ^ •^ sN^ i:::^ 1.5' 1:? 0^ ^ ^^^ 0.8 0.6" •i^^* ).4 ^^^** 0.3 - 0.2. 0.1

1 1 Ml) 1 1 INI 1 1 nil 11 Mil 0.1 1.0 10 100 1000 Mobility Ratio. M

Figure 2.27 Claridge correlation for areal sweep efficiency as a funcfion of the mobility ratio

71 Oil-Rtch Liquid Aspfialtic Solid

Figure 2.28 Schematic of displacement process with multiple phases

Displacing Displaced Phase Phase

Displacing Displaced Phase Phase

(a) Gravity Override pQ P(j

Figure 2.29 Gravity segregation in displacement processes

72 T*0-f»OT T>«»fr-MOT

A HucenoM ntu • MoewcTiM vtu MTTtM MUMMm

A «"?.-/*> O ! O

.»1 VTv" A © f O f

HEflUtAR POUR-SPOT tWtWtO TOUW-SFOT 1 1 A O A O A NOmikL NINE'SPOT mVUTCD NINE-SPOT

0-4.

^ o V o «^ 0

rive-SPOT I II I I I I I lit;• I . • J J

A-—^ o ^—A 0»—O A J^--O rt-ttt / » » o V—1( o A >—< A / \ Ir—ik A—A-~i -A— ^.^ A )>—O DIRECT LINE DRIVE STACCERED LINE DRIVE t \ f o ^—^ o A >..-<( A \—rf o 5*—A 0-"-(f A V—O SCVEN-SPOT INVERTED SEVEN-SPOT

Figure 2.30 Standardflooding pattern s

73 Figure 2.31FractionalflowofCO2andcrudoiwel2-26,JoffrVikingfield Figure 2.32PlotofCO 2 pluswaterinjectionversurecovery -S U. >- 0 M u. 8 0. 0 4 U 3 3 < .et p > FRACTION OF TOTAL FLOW REG It 0.2 0.4 0.6 0.8 _^;C—1 ' 0 7 — f — OILC02 1 t\J^ 85 16 ••••• COaINJPERIOD CUMULATIVE INJECTIONASFRACTIOOFHCPV 7 J^ <|f 1- 2 w > i K - 87 7 74 1.0 I.I 8 7 I 1]r 1 89 7 l.t i. 1111111 ) 90 1.0

0.8

0.6 Oil/Water ^ 0.4

0.2

0.0 0.3

Figure 2.33 Calculation of fractional water flow in a tertiary WAG process

40.000 i^55*sr Water Production (bbl/D) •***»/ '^JfJ '^^^V-^

to.000 C02 I Production I (Mft3/D)

1.000: COj injection Begins June 1981

300 -III |iiiiiiiiiii|iiiiiiiiiin niiWiiiiii'iiiiiiiiii|iiiii |iiiiiiiiiTi|*iii I iiiiii|iiiiiiiiirniiiiiiiiii 77 78 79 80 81 82 83 84 86 86 87 88 Year

Figure 2.34 Production rates of oil, water, and CO2 before and after CO2 injection in a WAG process (Four-Pattern Area)

75 100.000

Wator Production (bbl/D) •*•* rf*"^'< 1

A ProduProductioc n 10,000 (Mft^/D)

(bbl/D)

1.000:! P -* ^(bbl/D)

600 '"""""1"" I """|""'li"ll|"Tt>llllll[ll»ll»Tll|IIIIIIUIII|IIIHIIIIII] [IHIIl'ifcui I I 77 78 78 80 81 82 83 84 86 86 87 88 Year

Figure 2.35 Production rates before and after CO2 injection in a WAG process (Seventeen-Pattern Area)

20 CO2 Utilization (Mft»/bbl) 15

10

Incremental _ Oil Recovery (% OOIP) COj Production (% HPV) I t I—f—i—I—i—r 15 20 25 30

Cumulative CO2 Injection (% HPV)

Figure 2.36 Utilization of CO2 with incremental oil production and CO2 production in a WAG process (Seventeen-Pattern Area)

76 CUMULATIVE INCREMENTAL EOR RECOVERY V8 TIME

o 1:1 DUWAG ^^

Figure 2.37 Comparison between cumulative recoveries on a continuous CO2 flood, 1:1 WAG, and Denver Unit WAG processes

O I

is A WtSTtAKniOT«S s o wtsTs*Knior«« I' •s.Zi

i. 1 I • I I I I I I I I I I I I I I I • I • I • I I I •• I I I I I I 8 10 IS 20 29 30 3S 40 SOAK TIME, DAYS

Figure 2.38 Yeariy cumulative production versus soak times. West Sak Field, Alaska

77 A >WtSTMKnU>T#S O WE$TMCniOT#9

'—t-"—- 100 ISO 200 250 SSO 400 SLUG SIZE. UUSCf

Figure 2.39 Yeariy cumulative production versus CO2 slug size. West Sak Field, Alaska

Figure 2.40 Summary of the effect of repeated cycles on CO2 hufTn'puff incremental oil recovery

78 SLUO Size I, mot) 10 OT J^ 3,0 "f so —r- -ih o COt(Rtt«i«Cir ft% VM«mt t « m*l| wOt • N|IRttif«*ir r*i* VflwMt) > Ntlt«*l)

J as 0.S 0.4 0.S OS o.r -»t SLUG SIZE (Re*er«eir Pore Volume)

Figure 2.41 Total recovery of cyclic CO2 and cyclic N2 corefloods versus slug size measured in moles and reservoir pore volumes

10.000 n

INCREMENTAL OIL

1000 o GO

COo INJECTION^ 100 1982 1983 1984 1965 1986

Figure 2.42 Increase in the recovery rate for a CO2 huff'n'puff' process, South Louisiana

79 CHAPTER 3

DISCUSSION

3.1. Introduction

To develop an understanding of how CO2 can be used to mobilize hydrocarbons and maximize oil recovery, similarities and differences in miscibility development and phase behavior in existing processes, i.e., the continuous CO2 flood, the WAG process, and the CO2 hufTn'puff process are studied. Two common models of phase behavior that are currently in use are the vaporizing gas displacement process and the condensing gas displacement process, which are commonly studied using pseudoternary diagrams based on the unrealistic assumption that each of the CO2, the C2-C6 and the C7+ groups of pseudocomponents each mix equally and in all proportions with each of the others. A third model called the vaporizing/condensing gas displacement process does not rely on this assumption and shows miscible processes to be more dynamic than simpler models would suggest.

A comparison between the continuous CO2 flood and the CO2 hufTn'puff processes suggests that under conditions of low miscibility, oil recovery can be increased if the soak period of a CO2 hufTn'puff process is applied to a continuous C02 flood. This proposed soak alternating gas, or SAG, technique appears to have the potential to control mobility by minimizing viscous fingering and maximizing mass transfer between the oil and CO2 phases, and appears to compare favorably with the WAG process, which also controls mobility and raises

80 the maximum volumetric displacement efficiency but at the expense of a significantly lower microscopic displacement efficiency due to water shielding.

3.2. Miscibilitv

The concept of minimum miscibility pressure (MMP) is one that has developed from a practical need to correlate reservoir pressure to a practical recovery level in a continuous CO2flood. Figure 3.1 shows a plot of oil recovery factor versus pressure at a CO2 injection of 1.2 pore volumes (PVs), where the

MMP occurs at the 'break' in the curve and is characterized by a recovery of greater than about 95%. The value of the MMP can also be identifled as the pressure beyond which incremental recovery is no longer economically feasible in a continuous CO2 flood, and is a function of reservoir temperature, crude oil composition (Figure 2.10), and CO2 quality, or minimum miscibility enrichment

(MME) (Figure 3.2).

3.3. Phase Behavior

The phase behavior of C02/crude oil systems can be studied using pseudoternary diagrams, which illustrate the vaporizing gas displacement process and condensing gas displacement process. These two models explain how hydrocarbons transfer into the CO2 phase and CO2 transfers into the crude oil phase, respectively. However, pseudoternary diagrams are limited in that it must be assumed that each of the pseudocomponents mixes equally and in all

81 proportions with the other pseudocomponents. A third model of phase behavior, the vaporizing/condensing gas displacement process, is characterized by a two- way mass transfer between the CO2 and crude oil phases until miscibility is achieved.

3.3.1. The Vaporizing Gas Displacement Process

The vaporizing gas displacement process is characterized by a mass transfer of light hydrocarbons, followed by progressively heavier hydrocarbons from the crude oil into the CO2 until the CO2 becomes sufficiently hydrocarbon- enriched to be miscible with the crude oil. Here, the CO2 phase is initially pure or neariy pure. This process is shown on the pseudoternary diagram in Figure 3.3, where the composition path extends from the composition point of the crude oil at point C to a point tangent to the binodal curve, then follows the binodal curve to the point where it is closest to the composition point of the CO2 phase represented by point A.

3.3.2. The Condensing Gas Displacement Process

The condensing gas displacement process is characterized by the mass transfer of light hydrocarbons from a light hydrocarbon-enriched CO2 phase into the crude oil until the crude oil becomes sufficiently light hydrocarbon-enriched to

be miscible with the light hydrocarbon-enriched CO2. Here, the light hydrocarbon-

enriched CO2 is represented on a pseudoternary diagram as a point between the

82 upper and lower-right vertices. This process is shown in Figure 3.4, where the composition path extends from the composition point of the CO2 phase at point A to a point tangent to the binodal curve, then follows the binodal curve to the point where it is closest to the composition point of the crude oil represented by point

C. Since hydrocarbon-enriched CO2 is more expensive than pure CO2, the composition of the displacing phase is normally designed to be as lean as possible and still be to theright o f the limiting tie line.

3.3.3. The Vaporizing/Condensing Gas Displacement Process

The vaporizing/condensing process is comprised of a two-way mass transfer between the CO2 and crude oil phases where the displacing phase is composed of light intermediate hydrocarbon-enriched CO2, which contacts and transfers its light hydrocarbons into the crude oil. The CO2 phase at the flood front, now leaned-out and more mobile, moves ahead while the light intermediate hydrocarbon-enriched CO2 behind theflood fron t continues to give up its light intermediate hydrocarbons to the oil. The density of the oil both at and behind the front lowers as a result, while at the same time middle intermediate hydrocarbons transfer from the oil into the injected gas, which does not contain these components. Thus, the crude oil close to theflood fron t becomes rich in light intermediates and depleted in middle intermediates, which makes it less miscible with the enriched CO2 phase at theflood front .

83 Further behind the flood front, however, the crude oil becomes saturated in light intermediates and the transfer of these components from the light intermediate hydrocarbon-enriched phase eventually ceases. The CO2 phase at that point is rich in both light and middle intermediate hydrocarbons, significantly more so than at the flood front and almost completely miscible with the reservoir oil. This model was originally developed to explain why the density of the produced oil becomes lower than that of the reservoir oil then gradually rises to that of the reservoir oil as more oil is produced.'' The concept of a two-way mass transfer between the CO2 and crude oil phases provides more insight into the dynamics of miscibility and helps explain how miscibility can occur even when the miscibility conditions are poor.

3.4. CO? Recovery Mechanisms

The mechanisms of CO2 recovery such as oil swelling, the reduction of oil viscosity, and the reduction of interfacial tension between the CO2 and the crude oil, result from the transfer of CO2 into the crude oil. Other mechanisms of CO2 recovery include the extraction of hydrocarbons from the crude oil phase into the

CO2 phase, solution gas drive caused by the liberation of CO2 from the crude oil under low pressure conditions near the wellbore, and the acidization of carbonate formations, which appear to be able to cause an increase in permeability.

When oil swells, its saturation, i.e., relative volume within the pore spaces, increases and ceteris paribus creates more movable oil that can subsequently be

84 recovered. This can be seen on a fractional flow curve where an increase in oil saturation is accompanied by a corresponding increase in the fractional flow of the oil. Additionally, the displacement of mobile water by CO2 inaeases the relative permeability of the oil, and consequently the fractional flow of oil behind the CO2 flood front. In a CO2 hufTn'puff' process this also results in a low WOR. A reduction in oil viscosity also occurs, which according to Darcy's equation allows oil to flow faster, while an accompanying reduction in interfacial tension between the oil and CO2 phases allows oil within the pore spaces to deform and move through pore throat openings. The extraction mechanism involves the mass transfer of hydrocarbons from the crude oil phase to the CO2 phase, and continues to operate once the movable oil has been produced. Thus, once the mechanisms of miscibility have created more movable oil and physically displaced it to the producing wellbore, they extract light and intermediate hydrocarbons from bypassed oil into the CO2 and continue to contribute to the oil recovery process.

3.5. CO? Recovery Processes

3.5.1. Continuous CO2 Flooding

Continuous flooding with CO2 offers up to 15% OOIP additional recovery beyond the wateri'looding stage, and has been performed successfully in the fleld. One of the strengths of the continuous CO2 flood is that it has a high microscopic displacement eff'iciency, but some of its drawbacks include that large

85 volumes of CO2 are required, and an unfavorable mobility ratio or difference in density between the CO2 and the crude oil can cause viscousfingers o r gravity segregation, respectively, to develop and bypass oil which can also result in premature breakthrough. Capillary forces acting on various sized pore-throat openings can also cause bypassing at the pore level. This bypassed oil, which gradually gets stripped, is more miscible with CO2 that is hydrocarbon-enriched from the extraction process rather than CO2 that is relatively as pure as when it was injected. An uneven vertical permeability distribution between the injecting and producing wellbores can also cause channeling, which results in a poor sweep efficiency, and asphaltene deposits can occur when light and intermediate hydrocarbons are extracted from the oil, resulting in poor injectivity. Of all these potential drawbacks, one of the most substantial is that of the unfavorable mobility ratio between the CO2 and the crude oil, which can result in a low volumetric sweep efficiency.

3.5.2. CO2 Water Alternating Gas

WAG treatments were developed to alleviate the poor volumetric sweep efficiencies and premature CO2 breakthroughs associated with the continuous

CO2flood, where one WAG cycle consists of a slug of CO2 followed by a slug of water. The economics of this recovery method are generally more favorable due to the reduced volume of CO2 required and the more favorable mobility ratio between water and crude oil than in a continuous CO2 process. This improves

86 the volumetric sweep efficiency and results in less viscousfingering and gravity segregation. A significant drawback to the WAG process, though, is that the microscopic displacement efficiency is lower due to water blocking, or water shielding, where the exposure of the CO2 to the crude oil can be much lower, especially in water-wet conditions.

3.5.3. Cyclic CO2 Stimulation (HufTn'PufO

A CO2 hufTn'puff' is comprised of a CO2 injection, a soak period, and a production stage. During the injection stage, CO2 is injected into the reservoir under immiscible (or near-miscible) conditions so as to bypass oil and contact as much of the reservoir crude oil as possible. Injecting at low reservoir pressure

(usually at high injection rates) and adding impurities to the CO2 such as nitrogen or methane are normally the methods used to delay the miscibility. Once CO2 is dispersed away from the injection wellbore and the reservoir pressure has reached near-miscible levels, the well is shut in to begin the soak period.

In a CO2 hufTn'puff process, most of the mass transfer of CO2 into the crude oil and vice versa is delayed until the soak period because the CO2 is injected under immiscible conditions, either under low pressure or by using nitrogen- or methane-diluted COa.^'' Since most of the crude oil contacted by CO2 is bypassed during the injection, most of the mass transfer of CO2 occurs throughout the entireflooded regio n during the soak period. Mechanisms of miscibility such as oil swelling, lowering of viscosity and lowering of interfacial

87 tension also occur throughout the contacted region rather than predominate at the injection flood front as in a continuous CO2 flood. During the production stage, a relative permeability shift can be caused by the displacement of any mobile water by CO2, and promotes oil production.

The CO2 huff'n'puff process also offers a viable means of oil recovery when there is no communication between adjacent wells and continuous methods are not possible. However, CO2 hufTn'puff' can be more difflcult to optimize as it must operate within a range of reservoir conditions that are preferably less miscible in the injection stage and more miscible after the soak period. Thus, if the pressure is too high or increases too much in the injection stage, the CO2 will no longer bypass the oil but displace it away from the wellbore during the injection stage. Similarly, if the miscibility conditions are less than optimal during the production stage, which can occur when reservoir pressure is too low or with impurities in the CO2 that impede the swelling mechanism, lower oil recoveries will result.

3.6. SAG: A New Alternative

Of the conventional CO2 recovery methods, WAG treatments off'er better mobility control, lower CO2 expenditure and adaptability to multiwell patterns, but reduce the microscopic displacement efl'iciency. In addition, WAG recoveries may not always be viable due to poor water injectivity. Where WAG treatments are not feasible, an alternate means of mobility control is desirable, and if the

88 mobility control can be achieved without reducing the microscopic displacement

efficiency, it may be desirable even when WAG processes are viable.

This appears to be possible by combining the soak period of the CO2

hufTn'puff process to that of the continuous flood to establish mobility control.

By applying a soak period to a continuous CO2 flood, the mass transfer between

the CO2 and crude oil can maximize the effectiveness of both the extraction

mechanism and the physical displacement mechanisms. On the macroscopic

level, a soak period can allow the viscous fingers to dissipate more effectively

and cause a more even oil saturation to occur in the direction transverse to the

length of the fingers. On the microscopic level, water blocking will not occur as it would in a WAG process, and the high microscopic displacement efficiency of a continuous CO2 process can be retained. When the conditions in a continuous

CO2 flood become less and less miscible, the recovery mechanisms will operate progressively farther behind the injection flood front as they do in a CO2 hufTn'puff process, and the application of a soak period to the continuous CO2 flood appear to aid miscibility and be of progressively greater benefit.

Thus, if the injection and production wells shut in periodically, a SAG process would allow the transifion zone at the flood front to be more consistent and less prone to viscous fingering while retaining the microscopic displacement efficiency of a continuous CO2 flood. In addition, the extraction mechanism in a

SAG process would have initial mass transfer of hydrocarbons from the crude oil into the pure CO2 phase, with subsequent mass transfer of hydrocarbons into a

89 hydrocarbon-enriched CO2 phase. When this occurs in a CO2 hufTn'puff under non-miscible conditions, the results have been satisfactory even without the benefit of an oil bank at the injection flood front of a forward-drive process.

Therefore, a SAG process appears to be progressively more viable as conditions become progressively less miscible.

3.6.1. SAG Parameters

The parameters of the SAG process will govern injection, soak, and production aspects as well as the number of cycles. Prior to injection, reservoir pressure and CO2 quality help determine the miscibility of the system. If a SAG process that is carried out under fully miscible conditions, soak periods can be applied to optimize mobility control and minimize viscous fingering while retaining a high microscopic displacement efficiency. If the conditions are near-miscible or immiscible, the soak periods of a SAG process will help optimize the mass transfer between the two phases behind the flood front as well as minimize viscous fingering. Since a mass transfer under non-miscible conditions would likely take longer, a potential advantage to a rapid CO2 injection followed by a soak period over a slow flow rate and no soak period is the prospect of recovering the oil in a shorter amount of time. One reason this could occur is because the entire target area could be contacted with CO2 in as short a time frame as possible, which would allow the mechanisms of miscibility to operate over a maximum reservoir volume in the minimum amount of time. A second

90 reason this could occur is that when the mass transfer of hydrocarbons into the

CO2 phase begins while the CO2 is not moving, the CO2 and crude oil phases oil will become more compatible where they contact each other, whereas if the CO2 phase is moving through the reservoir there will be a much more abrupt change in the concentration of the two phases and the miscibility will not be optimal.

Parameters in the injection stage include the injection flow rate and the injection volume. A rapid injection flow rate followed by a soak period would need to be compared with progressively slower injection rates accompanied by progressively shorter soak periods over the same time frame to compare recoveries under similar time constraints. Under miscible conditions, a slow injection flow rate should help control viscous fingering but under near-miscible or immiscible condifions a mass transfer between the phases behind the flood front, more time might be necessary to maximize the mass transfer between the phases.

Since the purpose of the soak period is to provide mobility control and to give the mechanisms of miscibility more time to operate than in a confinuous process, three of the more important parameters are the duration of the soak period, the reservoir pressure and the CO2 quality. Miscibility can be raised during injection by shutting in the producing wells and allowing the reservoir pressure to build, whereas if the reservoir pressure is sufficient before injection, injecting and producing simultaneously can maintain the miscibility.

91 For economic reasons, it is attractive to maximize the flow rate during producfion, but once it is determined whether a slower flow rate accompanied by a shorter soak period would provide the same or greater recovery, injection and production flow rates can be determined to optimize the recovery. An opfimal number of cycles must also be determined for the SAG process, as well as its relationship to the volume injected on each cycle, the rate of injection, the rate of production, and the durafion of the soak period.

It is also practical to extend the concept of residence time, defined by

Gardner and Ypma^'' as the time for a "particle" offluid to be convected through the length of a core, from that of continuous processes to the SAG process.

Residence time in continuous CO2 coreflood is the core length divided by the average interstitial velocity of the fluid. A similar concept for a SAG process can also be deflned, whereby keeping the injection flow rates constant and the duration of each cycle equal, an analogous model to Gardner and Ypma's can be written for a linear system:

t^=l + nts (3.1) V

92 where there are n soak periods between the injection and production of any given particle of CO2. This can also be extended to radial systems where the particle velocity, v, is equal to the injection rate, q, divided by the aoss-secfional area of the reservoir, 2nrh, at any distance, r. from the injection wellbore. Thus, the residence fime in a radial system will be as follows:

^R= ^^ + nts . (3.2)

This will allow calculations to be made based on Gardner and Ypma's concepts.

3.6.2. Miscibility Condifions for SAG

A continuous CO2 flood produces optimal recoveries when it is at miscible conditions, while a CO2 hufTn'puff' produces optimal recoveries when it is at near- miscible conditions. Where condifions are immiscible, the recovery for either process will not be as high, although it may still be economically feasible. If a continuous CO2 flood is considered to be a special case of SAG where the durafion of the soak fime is zero, a similar considerafion can also be made for a

CO2 hufTn'puff that has been converted to a fonvard drive process, i.e., a SAG process, or analogously, a CO2 "hufTn'hufT'. It would seem that by converting the

CO2 hufTn'puff to a forward drive process, there would be an added advantage of an oil bank ahead of the injection flood front, which can add to the recovery expected from a CO2 huff^n'puff' where the recovered oil comes from behind the injection flood front. Thus, by locating the opfimal continuous CO2 flood and the

93 C02 hufTn'puff on the spectrum of miscibility, SAG parameters can optimize the recovery process across the entire spectrum of miscibility, including the two special cases of the continuous CO2flood an d the forward-drive CO2 hufTn'puff.

Addifionally, a soak period applied to a WAG process following the CO2 injection may yield significant results if it is found that the effectiveness of a soak period is related to the water saturafion. If a soak period compensates for the effects of water shielding, where portions of the oil are not contacted by CO2 due to the presence of injected water, then the mechanisms of miscibility may become more effective if a soak period is applied to a WAG process as well as to a confinuous CO2 flood.

94 )00 uiTiMAre

80- o

S?60 > O W lu A O M£AO STKAWN STO o A • STO WITHOUT Cj -C4

20

_L _L _L J, 1200 1600 2000 2400 FIOOO PRESSUItE, PSIG Figure 3.1 Plot of oil recovery factor versus pressure at 1.2 PVs of CO2 injected into a sand pack containing Mead-Strawn stock tank oil

>• S > E

-t- -t- -t- 0 0.1 0.4 0.« 0.1 1 FRACTION OP SOLVENT IN LEAN QAS

Figure 3.2 Plot of oil recovery factor versus fraction of solvent in lean gas, showing the point of minimum miscibility enrichment (MME)

95 Figure 3.3 Pseudoternary diagram of the vaporizing gas process

C.c«

Figure 3.4 Pseudoternary diagram of the condensing gas process

96 CHAPTER 4

CONCLUSIONS

The following conclusions can be drawn from the research:

• A new concept in CO2 enhanced oil recovery (EOR) called SAG, or soak

alternating gas, has been introduced that combines the soak period of a

CO2 hufTn'puff process to that of a continuous CO2flood. The purpose of

this is to provide an alternate means of mobility control to the water

alternafing gas (WAG) process and a means of increasing and/or

accelerafing oil recovery beyond currently practiced EOR methods. An

additional advantage to the SAG process is that it can be implemented in

areas of poor water injectivity.

e The SAG process consists of continuous CO2flood with intermittent soak

periods, as are pracficed in a CO2 hufTn'puff to opfimize the miscible

process, but unlike a CO2 hufTn'puff, SAG is a multiwell process where the

production flow occurs in the same direction as the injection flow.

• Although a WAG process improves the volumetric sweep over that of a

continuous CO2flood, wate r shielding, which occurs when water contacts

much of the crude oil rather than CO2, diminishes much of the

effectiveness of the injected CO2 (Table 2.1). Therefore, if a means of

mobility control can be found that does not have the properties of the

WAG process that are detrimental to oil recovery, the improvements will

be significant.

97 • The hypothesis that the SAG process can improve recoveries in CO2

applicafions appears to be reasonable. If the CO2 hufTn'puff process were

converted to a fonvard drive, similar if not better recoveries should be

experienced due to the recovery of fluids from oil banking ahead of the

injection flood front, which would not likely be recovered from a CO2

hufTn'puff. Similariy, if a soak period were performed in a continuous CO2

flood, the diffusion process can cause a more constant concentration

gradient to occur lateral to the flow and alleviate problems due to viscous

fingering.

• The effectiveness of the components of the SAG process appears to be

supported in the literature on research, tesfing, and implementation of

confinuous CO2 flooding, CO2 hufTn'puff, and WAG processes. An

additional aspect to the SAG process is that it may also be applicable to

WAG process where the soak period could alleviate problems of low

recovery that result from water shielding.

98 CHAPTER 5

RECOMMENDATIONS

Based on the research and conclusions, the author makes the following recommendations:

• The SAG process appears to be a viable alternative to the confinuous CO2

flood, WAG and the CO2 huff'n'puff processes; its validity needs to be

verified. This could be done with numerical simulations whose initial

operafing parameters are based on those of continuous CO2floods an d

CO2 hufTn'puff processes to provide a direct comparison under a range of

miscibility condifions. These comparisons should be able to determine the

optimal conditions are under which a SAG process can be run, as well as

the trends in its response that develop as the parameters are varied.

e It is of particular interest to learn the condifions a SAG treatment could be

applied where the recovery could be greater than by using other

conventional recovery methods. The identification of trends in the

response to parameter variations should establish qualitative indicators of

responses that will be encountered in the field.

• The capability of the soak period to control mobility in a SAG process can

be assessed to determine how SAG compares to convenfional CO2

flooding that is ideally run under miscible condifions. The miscibility can

then be lowered progressively unfil the conditions are immiscible.

Similariy, a comparison under immiscible condifions between a

99 convenfional CO2 hufTn'puff and a CO2 SAG process can also be made, followed by further comparisons under condifions that are progressively

more miscible unfil full miscibility is reached. Such an analysis should yield trends that compare SAG with the CO2 huff'n'puff and a confinuous CO2 flood over a full range of miscibility conditions. Two significant properties would be the oil recovery and the production rate, which could be the

basis of a full economic analysis prior to a field test.

The performance of SAG, compared to the continuous CO2 flood and the

WAG process, can be invesfigated with respect to variafions in the following parameters:

o Conditions of miscibility: oil composition and properties, C02

quality, pressure variations over the injection cycle,

o Reservoir condifions: pressure, temperature, porosity, permeability,

permeability distribution, natural fractures, lithology (carbonate

versus silicate), geological trend, water leg, gas cap.

o Injecfion stage: injecfion volume, durafion of injecfion, pressure

changes during injection, injection flow rate, effect of producfion

during the injection cycle,

o Soak period: duration of soak period, conditions of miscibility during

soak period.

100 o Production stage: rate of producfion, composition and

characterisfics of producedfluids versu s those of the reservoir and injected fluids.

o Number of SAG cycles.

• Optimal SAG cases can then be identifled then compared to the WAG

process, particularly with respect to the recovery, rate of recovery, and

profitability of each method.

• The techniques of SAG and WAG should be combined in a cycle

consisting of a water injecfion, a CO2 injection, and then a soak period, to

investigate the extent to which water shielding could be alleviated in the

soak period. Prior to afield pilot test, an analysis can also be conducted

with slim tube and/or coreflood experiments. e The slim tube, because of its length, would be useful in analyzing the

effect of varying the times and distances between soak periods which

could be used to determine CO2 injection volumes and the number of SAG

cycles to be performed in the field. It would also be useful to compare

such tests to continuous CO2floods ru n under the same conditions, which

would allow direct comparisons to be made between the two methods.

• An analysis of SAG performed in corefloods should allow it to be studied

under condifions close to those encountered in situ, where again,

comparisons between the SAG, continuous CO2flooding an d CO2

hufTn'puff would be desirable.

101 Comparisons between SAG and WAG from slim tube and/or coreflood experiments would also be useful, as would a study as to what extent the effects of water shielding in WAG processes can be alleviated by soak periods following the CO2 injections. Once such analyses have been

performed, strategies for field implementation can then be designed and

simulafions run to opfimize them.

102 BIBLIOGRAPHY

1 Patton, J.T., Sigmund, P., Evans, B., Ghose, B. and Weinbrandt, D.: "Design of a CO2 Stimulation Process for Heavy Oil Reservoirs," paper SPE 8897 presented at the 1980 SPE/AIME California Regional Meeting, Pasadena, CA, April 9-11.

2 Brock, W.R. and Bryan, L.A.: "Summary Results of CO2 EOR Field Tests, 1972-1987," paper SPE 18977 presented at the 1989 SPE Joint Rocky Mountain Regional/Low Permeability Reservoirs Symposium and Exhibifion, Denver, CO, March 6-8.

3 Green, D.W. and Willhite, G.P.: Enhanced Oil Recovery, SPE, Richardson, TX, 1998.

4 Zick, A.A.: "A Combined CondensingA/aporizing Mechanism in the Displacement of Oil by Enriched Gases," paper SPE 15493 presented at the 1986 SPE Annual Technical Conference and Exhibition, New Orieans, LA, October 5-8.

5 Buckley, S.E. and Leverett, M.C.: "Mechanism of Fluid Displacement in Sands," Trans., AIME (1942) 146, 107-116.

6 Welge, H.J.: "A Simplified Method for Compufing Oil Recoveries by Gas or Water Drive," Trans. AIME (1952) 195, 91-98.

7 Amyx, J.W., Bass, D.M. and Whifing, R.L.: Petroleum Reservoir Engineering, McGraw-Hill, Inc., New York (1998 Reprint).

8 Moore, T.F. and Slobod, R.L.: "Displacement of Oil by Water-Effect of Wettability, Rate and Viscosity on Recovery," paper SPE 502-G presented at the 1955 AIME 30th Annual Fall Meeting, Petroleum Branch, New Orleans, October 2-5.

9 Taber, J.J.: "Dynamic and Stafic Forces Required to Remove a Disconfinuous Oil Phase from Porous Media Containing Both Oil and Water," paper SPE 2098 presented at the 1968 SPE European Regional Meeting, Milano, Italy, April 4-6.

10 Stegemeier, G.L.: "Mechanisms of Entrapment and Mobilizafion of Oil in Porous Media," Improved Oil Recovery by Surfactant and Polymer Flooding, Academic Press Inc., New York (1977) 55-91.

103 11 Stegemeier, G.L.: "Relafionship of Trapped Oil Saturafion to Petrophysical Properties of Porous Media," paper SPE 4754 presented at the 1974 SPE- AIME Improved Oil Recovery Symposium, Tulsa, OK, April 22-24.

12 Melrose, J.C. and Bradner, C.F.: "Role of Capillary Forces in Determining Microscopic Displacement Efficiency for Oil Recovery by Waterflooding," Journal of Canadian Petroleum Technology (October-December 1974), 54-62.

13 Mungan, N.: "Role of Wettability and Interfacial Tension in Waterflooding," paper SPE 705 presented at the 1963 SPE Annual Meefing, New Orieans, October 6-9.

14 Wagner, O.R. and Leach, R.O.: "Effect of Interfacial Tension on Displacement Efl'iciency," paper SPE 1564 presented at the 1966 SPE Annual Meefing, Dallas, TX, October 2-5.

15 Gunn, S.O. and Slattery, J.C: "Interfacial Tension Required for Significant Displacement of Residual Oil," paper SPE 5992, SPEJ (April 1979) 83-95.

16 Eriich, R., Hasiba, H.H. and Raimondi, P.:"Alkaline Waterilooding for Wettability Alteration - Evaluating a Potential Field Application," paper SPE 4905 presented at the 1974 SPE Annual California Regional Meeting, San Francisco, CA, April 4-5.

17 Arriola, A., Willhite, G.P. and Green, D.W.: "Trapping of Oil Drops in a Noncircular Pore Throat," paper SPE 9404 presented at the 1980 SPE- AIME Annual Technical Conference and Exhibifion, Dallas, TX, September 21-24.

18 Hutchinson, CA. and Braun, P.H.: "Phase Relafions of Miscible Displacement in Oil Recovery," A.I.Ch.E Journal, (March 1961) 64-72.

19 Holm, L.W. and Josendal, V.A.: "Mechanisms of Oil Displacement By Carbon Dioxide," paper SPE 4736 presented at the 1974 SPE-AIME Improved Oil Recovery Symposium, Tulsa, OK, April 22-24.

20 Holm, L.W. and Josendal, V.A.: "Effect of Oil Composifion on Miscible- Type Displacement by Carbon Dioxide," paper SPE 8814 presented at the 1980 SPE-DOE Enhanced-Oil Recovery Symposium, Tulsa, OK, April 20- 23.

104 21 Hagedorn, K.D. and Orr Jr., P.M.: "Component Partitioning in C02/Crude Oil Systems: Effects of Oil Composifion on CO2 Displacement Performance," paper SPE 25169 presented at the 1993 SPE Internafional Symposium on Oilfield Chemistry, New Orieans, March 2-5.

22 Metcalfe, R.S. and Yarborough, L.: "The Effect of Phase Equilibria on the CO2 Displacement Mechanism," paper SPE 7061 presented at the 1978 SPE-AIME Fifth Symposium on Improved Methods for Oil Recovery Symposium, Tulsa, OK, April 16-18.

23 Metcalfe, R.S.: "Effects of Impurities on Minimum Miscibility Pressures and Minimum Enrichment Levels for CO2 and Rich-Gas Displacements," paper SPE 9230 presented at the 1980 SPE-AIME Annual Technical Conference and Exhibifion, Dallas, TX, September 21-24.

24 Gardner, J.W. and Ypma, J.G.J.: "An Investigation of Phase Behavior/Macroscopic-Bypassing Interacfion in CO2 Flooding," paper SPE 10686 presented at the 1982 SPE/DOE Enhanced-Oil Recovery Symposium, Tulsa, OK, April 4-7.

25 Shyeh-Yung, J-G.J.: "Mechanisms of Miscible Oil Recovery: Effects of Pressure on Miscible and Near-Miscible Displacements of Oil by Carbon Dioxide," paper SPE 22651 presented at the 1991 SPE Annual Conference and Exhibifion, Dallas, TX, October 6-9.

26 Stern, D.: "Mechanisms of Miscible Oil Recovery: Effects of Pore-Level Fluid Distribufion," paper SPE 22652 presented at the 1991 SPE Annual Conference and Exhibifion, Dallas, TX, October 6-9.

27 Perkins, T.K. and Johnston, O.C: "A Review of Diffusion and Dispersion in Porous Media," SPEJ (March 1963), 70-81; Trans. AIME (1963) 228, 70-81.

28 Taylor, G. I.: "Dispersion of Soluble Water in Solvent Flowing Slowly Through a Tube," Proc. Royal Society, London (1953) 219,186.

29 Aris, R. and Amundson, N.R.: "Some Remarks on Longitudinal Mixing and Diffusion in Fixed Beds," A.I.Ch.E. Journal{1951) 280-282.

30 Aris, R.: "The Longitudinal Diffusion Coefficient in Flow Through A Tube With Stagnant Pockets," Chemical Engineering Soc/efy (1959) 194-198.

105 31 Raimondi, P., Gardner, G.H.F. and Patrick, C.B.: "Effects of Pore Structure and Molecular Diffusion on the Mixing of Miscible Liquids in Porous Media," preprint 43 presented at the 1959 SPE/A.I.Ch.E Joint Symposium, San Francisco, CA, Dec 6-9.

32 Simon, R., Rosman, A. and Zana, E.: "Phase-Behavior Properties of CO2- Reservoir Oil Systems," SPEJ (February 1978) 20-26.

33 Stalkup, F.I. Jr.: Miscible Displacement, Monograph Series, SPE, Richardson, TX (1992) 8, 32.

34 Craig, F.F. Jr. et al.: "A Laboratory Study of Gravity Segregation in Frontal Drives," Trans., AIME (1957) 210, 275-282.

35 Moore, T.F. and Slobod, R.C: "The Effect of Viscosity and Capillarity on the Displacement of Oil by Water," Producers Monthly (August 1956) 20- 30.

36 Crane, F.E., Kendall, H.A. and Gardner, G.H.F.: "Some Experiments of the Flow of Miscible Fluids of Unequal Density Through Porous Media," SPEJ (December 1963) 277-80.

37 Perkins, T.K., Johnston, O.C. and Hoffman, R.N.: "Mechanics of Viscous Fingering in Miscible Systems," paper SPE 1229 presented at the 1965 SPE Meeting, Denver, CO, October 3-6.

38 Coats, K.H. and Smith, B.D.: "Dead-End Pore Volume and Dispersion in Porous Media," SPEJ (March 1974) 73-84.

39 Baker, L.E.: "Effects of Dispersion and Dead-End Pore Volume in Miscible Flooding," paper SPE 5632 presented at the 1975 SPE-AIME Annual Technical Conference and Exhibition, Dallas, September 28-October 1.

40 Stalkup, F.I. Jr.: "Displacement of Oil by Solvent at High Water Saturation," paper SPE 2419 presented at the 1969 SPE Improved Oil Recovery Symposium, Tulsa, OK, April 13-15.

41 Shelton, J.L. and Schneider, F.N.: "The Effects of Water Injection on Miscible Flooding Methods Using Hydrocarbons and Carbon Dioxide," paper SPE 4580 presented at the 1973 SPE-AIME Annual Meeting, Las Vegas, NV, September 30-October 3.

106 42 Tiffin, D.L. and Yellig, W.F.: "Effects of Mobile Water on Mulfiple-Contact Miscible Gas Displacements," paper SPE 10687 presented at the 1982 SPE/DOE Enhanced Oil Recovery Symposium, Tulsa, OK, April 4-7.

43 Koval, E.J.: "A Method of Predicting the Performance of Unstable Miscible Displacement in Heterogeneous Media," SPEJ (June 1963) 145-154.

44 Claridge, E.L.: "Predicfion of Recovery in Unstable Miscible Flooding," paper SPE 2930 presented at the 1970 SPE Annual Meeting, Houston, TX, October 4-7.

45 Shelton, J.L. and Yarborough, L.: "Multiple Phase Behavior in Porous Media During CO2 or Rich-Gas Flooding," paper SPE 5827 presented at the 1976 SPE-AIME Symposium on Improved Oil Recovery, Tulsa, OK, March 22-24.

46 Harvey, M.T. Jr., Shelton, J.L. and Kelm, C.H.: "Field Injectivity Experiences With Miscible Recovery Projects Using Alternate Rich-Gas and Water Injection," JP7 (September 1977), 1051-1055.

47 Mungan, N., "Improved Oil Recovery," Interstate Oil Compact Commission, Oklahoma City, OK (1983), 113-172.

48 Leone, J.A. and Scott, E.M.: "Characterizafion and Control of Formafion Damage During Waterflooding of a High Clay-Content Reservoir," paper SPE 16234 presented at the 1987 SPE Producfion Operafions Symposium, Oklahoma City, OK, March 8-10.

49 Cusak, F., et al.: "Diagnosis and Removal of Microbial /Fines Plugging in Water Injection Wells," paper SPE 16907 presented at the 1987 Annual SPE Technical Conference and Exhibition, Dallas, TX, September 27-30.

50 Kumar, T.: "Formafion Damage Resulfing from Particle Invasion," paper SPE 23561, 1991 unsolicited.

51 Gardner, J.W., Orr, F.M. and Patel, P.D.: "The Effect of Phase Behavior on C02-Flood Displacement Efficiency," paper SPE 8367 presented at the 1979 SPE Annual Technical Conference and Exhibifion, Las Vegas, NV, September 23-26.

52 Craig, F.F. Jr.: The Reservoir Engineering Aspects of Waterflooding, Monograph Series, SPE, Richardson, TX (1971) 3.

107 53 Hadlow, R.E.: "Update of Industry Experience With CO2 Injection," paper SPE 24928 presented at the 1992 SPE Annual Technical Conference and Exhibition, Washington, DC, October 4-7.

54 Huang, E.T.S. and Holm, L.W.: "Effect of WAG Injecfion and Rock Wettability on Oil Recovery During CO2 Flooding," paper SPE 15491 presented at the 1986 SPE Annual Technical Conference and Exhibifion, New Orieans, LA, October 5-8.

55 Ponfious, S.B. and Tham, M.J.: "North Cross (Devonian) Unit CO2 Flood - Review of Flood Performance and Numerical Simulation Model," paper SPE 6390 presented at the 1977 SPE-AIME Permian Basin Oil and Gas Recovery Conference, Midland, TX, March 10-11.

56 Yellig, W.F.: "Carbon Dioxide Displacement of a West Texas Reservoir Oil," paper SPE 9785 presented at the 1981 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, OK, April 5-8.

57 Mizenko, G.J.: "North Cross (Devonian) Unit CO2 Flood: Status Report," paper SPE 24210 presented at the 1992 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, OK, April 22-24.

58 Stephenson, D.J., Graham, A.G. and Luhning, R.W.: "Mobility Control Experience in the Joffre Viking Miscible CO2 Flood," paper SPE 23958 published with permission of AOSTRA, JVTOU, and Westcoast Petroleum Ltd. (May 20, 1992).

59 Kumar, R. and Eibeck, J.N.: "CO2 Flooding a Waterflooded Shallow Pennsylvanian Sand in Oklahoma: A Case History," paper SPE 12668 presented at the 1984 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, OK, April 15-18.

60 Perry, G.E.: "Weeks Island 'S' Sand Reservoir B Gravity-Stable Miscible CO2 Displacement," paper SPE 10695 presented at the 1982 SPE/DOE Enhanced Oil Recovery Symposium, Tulsa, OK, April 4-7.

61 Bellavance, J.F.R.: "Dollarhide Devonian CO2 Flood: Project performance review 10 years later," paper SPE 35190 presented at the 1996 SPE Permian Basin Oil & Gas Recovery Conference, Midland, TX, March 27- 29.

62 Raimondi, P. and Torcaso, M.A.: "Distribufion of the Oil Phase Obtained Upon Imbibifion of Water," SPEJ (March 1964) 49-55.

108 63 Lin, E.C and Huang, E.T.S.: "The Effect of Rock Wettability on Water Blocking During Miscible Displacement," paper SPE 17375 presented at the 1988 SPE/DOE Enhanced Oil Recovery Symposium, Tulsa, OK, April 17-20, 1988. J' y K . . K

64 Chase, CA. Jr. and Todd. M.R.: "Numerical Simulafion of CO2 Flood Performance," paper SPE 10514 presented at the 1984 SPE Reservoir Simulafion Symposium, New Orieans, January 31-February 3.

65 Walsh, M.P.: "An Analysis of Miscible Flooding Chase-Fluid Strategies," paper SPE 19866 presented at the 1989 SPE Annual Technical Conference and Exhibifion, San Antonio, October 8-11.

66 Brokmeyer, R.J. et al.: "Lost Soldier Tensleep CO2 Tertiary Project, Performance Case History; Baroil, Wyoming," paper SPE 35191 presented at the 1996 SPE Permian Basin Oil and Gas Recovery Conference, Midland, TX, March 27-29.

67 Langston, S.F. et al.: "Definifive CO2 Flooding Response in the SACROC Unit," paper SPE 17321 presented at the 1988 SPE/DOE Enhanced Oil Recovery Conference, Tulsa, OK, April 17-20.

68 Tanner, C.S. et al.: "Production Performance of the Wesson Denver Unit CO2 Flood," paper SPE 24156 presented at the 1992 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, Oklahoma, April 22-24.

69 Patton, J.T., Coats. K.H. and Spence. K: "A Parametric Study of the CO2 Huf-n-Puf Process," paper SPE 9228 presented at the 1979 SPE Annual Technical Conference and Exhibition, Las Vegas, NV, September 23-26.

70 Khatib, A.K. and Eariougher, R.C and Kantar, K.: "CO2 Injection as an Immiscible Applicafion for Enhanced Oil Recovery in Heavy Oil Reservoirs," paper SPE 9928 presented at the 1981 SPE California Regional Meefing, Bakersfield, CA, March 25-26.

71 Sankur, V. and Emanuel, A.S.: "A Laboratory Study of Heavy Oil Recovery With CO2 Injection," paper SPE 11692 presented at the 1983 SPE California Regional Meeting, Ventura, CA, March 23-25.

72 Reid, T.B. and Robinson, H.J.: "Lick Creek Meakin Sand Unit Immiscible C02/Waterfiood Project," paper SPE 9795 presented at the 1981 SPE/DOE Enhanced Oil Recovery Symposium, Tulsa, April 6-7.

109 73 Bakashi, A.K., Ogbe, D.O., Kamath, V.A. and Hatzignafiou, D.G.: "Feasibility Study of CO2 Sfimulafion in the West Sak Field, Alaska." paper SPE 24038 presented at the 1992 SPE Western Regional Meefing, Bakersfield, CA, March 30-April 1.

74 Hsu, H.H. and Brugman, R.J.: "CO2 Huff-Puff Simulation Using a Composifional Reservoir Simulator," paper SPE 15503 presented at the 1986 SPE Annual Technical Conference and Exhibition, New Orieans, LA, October 5-8.

75 Monger, T.G. and Coma, J.M.: "A Laboratory and Field Evaluation of the CO2 Huff 'n Puff Process for Light Oil Recovery," paper SPE 15501 presented at the 1986 SPE Annual Technical Conference and Exhibifion, New Orieans, LA, October 5-8.

76 Monger, T.G., Ramos, J.C. and Thomas, G.A.: "Light Oil Recovery From Cyclic CO2 Injection: Influence of Low Pressures, Impure CO2, and Reservoir Gas," paper SPE 18084 presented at the 1988 SPE Annual Technical Conference and Exhibifion, Houston, TX, October 2-5.

11 Thomas, G.A. and Monger-McClure, T.G.: "Feasibility of Cyclic CO2 Injection for Light-Oil Recovery," paper SPE 20208 presented at the 1990 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, OK, April 22-25.

78 Thomas, G.A., Berzins, T.V., Monger, T.G. and Bassiouni, Z.: "Light Oil Recovery From Cyclic CO2 Injection: Influence of Gravity Segregation and Remaining Oil." paper SPE 20531 presented at the 1990 SPE Annual Technical Conference and Exhibition, New Orieans, LA, September 23-26.

79 Karim, F., Berzins, T.V., Schenewerk, P.A., Bassiouni, Z.A. and Wolcott, J.M.: "Light Oil Recovery From Cyclic CO2 Injecfion: Influence of Drive Gas, CO2 Injecfion Rate, and Reservoir Dip," paper SPE 24336 presented at the 1992 SPE Rocky Mountain Regional Meefing, Casper, WY, May 18- 21.

80 Shayegi, S., Jin, Z., Schenewerk, P., Wolcott, J.: "Improved Cyclic Stimulation Using Gas Mixtures," paper SPE 36687 presented at the 1996 SPE Annual Technical Conference and Exhibition, Denver, CO, October 6-9.

110 81 Hara, S.K. and Christman, P.G.: "Invesfigafion of a Cyclic Countercun-ent Light-0il/C02 Immiscible Process," paper SPE 20207 presented at the 1990 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, April 22- 25.

82 Haskin, H.K. and Alston, R.B.: "An Evaluation of the CO2 Huff 'n' Puff Field Tests in Texas," paper SPE 15502 presented at the 1986 SPE Annual Technical Conference and Exhibifion, New Orleans, LA, October 5-8.

83 Adamache, I., Kantzas, A., Mclntyre, F.I. and Sigmund, P.M.: "The Effect of Cyclic Flow Interruptions on the Performance of Vertically Directed Miscible Floods," paper SPE 22935 presented at the 1991 SPE Annual Technical Conference and Exhibifion, Dallas, TX, October 6-9.

84 Palmer, F.S., Landry, R.W. and Bou-Mikael, S.: "Design and Implementation of Immiscible Carbon Dioxide Displacement Projects (CO2 Huff-Puff) in South Louisiana," paper SPE 15497 presented at the 1986 SPE Annual Technical Conference and Exhibition, New Orieans, LA, October 5-8.

85 Bou-Mikael, S. and Palmer, F.S.: "Field-Derived Comparison of Tertiary Recovery Mechanisms for Miscible CO2 Flooding of Waterdrive and Pressure-Depleted Reservoirs in South Louisiana," paper SPE 19636 presented at the 1989 SPE Annual Technical Conference and Exhibifion, San Antonio, TX, October 8-11.

86 Miller, B.J.: "Design and Results of a Shallow, Light Oilfield-Wide Applicafion of CO2 Huff 'n' Puff Process," paper SPE 20268 presented at the 1990 SPE/DOE Symposium on Enhanced Oil Recovery, Tulsa, April 22-25.

87 Bardon, C, Coriay, P., Longeron, D. and Miller, B.: "Interpretafion of a CO2 Huff 'n' Puff Field Case in a Light-Oil-Depleted Reservoir," paper SPE 22650 presented at the 1991 SPE Annual Technical Conference and Exhibifion, Dallas, TX, October 6-9.

88 Miller, B.J.: "CO2 Huff 'n' Puff Field Case: Five-Year Program Update," paper SPE 27677 presented at the 1994 SPE Permian Basin Oil and Gas Recovery Conference, Midland, TX, March 16-18.

Ill 89 Miller, B.: "Field Case, Cyclic Gas Recovery for Light Oil-Using Carbon Dioxide/Nitrogen/Natural Gas," paper SPE 49169 presented at the 1998 SPE Annual Technical Conference and Exhibition, New Orieans, LA, September 27-30.

90 Brock, W.R. and Bryan, L.A.: "Summary Results of CO2 EOR Field Tests, 1972-1987," paper SPE 18977 presented at the 1989 SPE Joint Rocky Mountain Regional/Low Permeability Reservoirs Symposium and Exhibifion, Denver, CO. March 6-8.

112 PERMISSION TO COPY

In presenting this thesis in panial fiilfUlment of the requirements for a master's degree at Texas Tech University or Texas Tech University Health Sciences Center, I agree that the Library and my major department shall make it freely available for research purposes. Permission to copy this thesis for scholarly purposes may be granted by the Director of the Library or my major professor. It is understood that any copying or publication of this thesis for financial gain shall not be allowed without my further written permission and that any user may be liable for copyright infringement.

Agree (Permission is granted.)

Student Signature Date

Disagree (Permission is not granted.)

Student Signature Date