You are on page 1of 10

International Journal of Greenhouse Gas Control 28 (2014) 2534

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control


journal homepage: www.elsevier.com/locate/ijggc

Comparison of different methods for determining key parameters


affecting CO2 storage capacity in oil reservoirs
Changlin Liao a,b, , Xinwei Liao a , Xiaoliang Zhao a , Hongna Ding a ,
Xiaopeng Liu c , Yongge Liu a , Jing Chen a , Ning Lu a
a

Petroleum Engineering Department, China University of Petroleum, Beijing, Peoples Republic of China
PetroChina Research Institute of Petroleum Exploration and Development, Beijing, Peoples Republic of China
c
Geological Exploration and Development Research Institute of CNPC, Chuanqing Drilling Engineering Company Limited, Chengdu, Sichuan,
Peoples Republic of China
b

a r t i c l e

i n f o

Article history:
Received 9 November 2012
Received in revised form 9 June 2014
Accepted 10 June 2014
Available online 28 June 2014
Keywords:
Storage capacity
Determination methods
Oil reservoirs
Stepwise regression
Carbon sequestration

a b s t r a c t
Storing CO2 in oil reservoirs is not only an effective method for reducing CO2 emissions and greenhouse
effects, but also a means to be more economical by enhancing oil recovery. The evaluation of CO2 storage
capacity in oil reservoirs which is very important for the implementation of CO2 storage includes the
evaluation of theoretical, effective, practical and matched storage capacities. Based on the volumetric
balance theory, considering CO2 dissolved in remaining oil and water, sweep efciency and displacement
efciency, this paper utilizes three methods to calculate theoretical and effective CO2 storage capacity
in oil reservoirs, in which CO2 volumetric sweep efciency, oil recovery factor and sequestration factor
are key parameters. This work presents a reservoir numerical simulation method, an empirical formula
method, and a stepwise regression method. The feasibility, superiority and limitations of the methods
for calculating these three key parameters and storage capacities including theoretical and actual CO2
storage capacities were analyzed through simulated applications in three reservoirs of the Xinjiang
Oileld of China. The results indicated that the assessment results of stepwise regression has a high level
of accuracy, and that this oileld can provide a large storage capacity and is thereby worthy of further
study.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
The underground storage of carbon dioxide (CO2 ), i.e., carbon sequestration, is a promising method to reduce greenhouse
gas emission. Many international institutions have been investing
resources to evaluate geological CO2 storage potential. There are
ve major geological systems that have been identied as the most
promising targets for CO2 storage: oil and gas reservoirs, saline formations, un-minable coal areas, shale, and basalt formations (IPCC,
2005). After signicant exploration and development, oil reservoirs
have been better characterized, which makes the estimation of CO2
storage capacity in oil reservoirs simpler and more straightforward
than in other geologic formations in normal situations.

Corresponding author at: China University of Petroleum-Beijing, Mail Box 269,


Changping, Beijing 102249, Peoples Republic of China. Tel.: +86 010 8973 3223;
fax: +86 010 8973 3223.
E-mail address: liaocup@163.com (C. Liao).
http://dx.doi.org/10.1016/j.ijggc.2014.06.010
1750-5836/ 2014 Elsevier Ltd. All rights reserved.

Based on the volumetric balance theory, the fundamental


assumption made in storage capacity calculations is that the volume previously vacated by liquid recovery is fully available for CO2
storage (Stevens et al., 2001; Bachu and Shaw, 2003). The Carbon
Sequestration Leadership Forum (CSLF) presented a method to calculate theoretical storage capacity based on this assumption. The
calculation is shown by Eq. (1) (CSLF, 2007). The volume of injected
water, aquifer inux and produced water were considered in this
equation.
MCO2 t = CO2 r [Rfl A h  (1 Swi ) Viw + Vpw ]

(1)

The United States Department of Energy applied another


method to calculate CO2 theoretical storage capacity, based on CO2
enhanced oil recovery theory and volumetric balance theory. The
general form of the volumetric equation to calculate CO2 theoretical storage capacity in oil reservoirs is shown by Eq. (2) (Goodman
et al., 2011). Table 1 summarizes the terms shown in Eqs. (1) and
(2).
MCO2 t = CO2 s A h  (1 Swi ) Bo Eoil

(2)

26

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

Table 1
Parameters for theoretical storage capacity estimation published in references.
Parameter

Units

Description

MCO2 t
CO2 r

kg
kg/m3

m3 /m3

A
h

Swi
Viw

m2
m
m3 /m3
m3 /m3
m3

Vpw

m3

CO2 s

kg/m3

Bo
Eoil

standard m3 /reservoir m3
m3 -CO2 /m3 -OOIP

Theoretical storage capacity


CO2 density under reservoir
conditions
Total oil recovery factor on
reservoir volume basis
Oil reservoir area
Oil reservoir thickness
Oil reservoir porosity
Initial water saturation
Cumulative volume of injected
water and aquifer inux
Cumulative volume of
produced water
CO2 density under standard
conditions
Oil formation volume factor
Storage efciency factor that is
derived from local CO2
enhanced oil recovery (EOR)
experience or reservoir
simulation, as the ratio of
standard cumulative volume of
CO2 to standard volume of
original oil-in-place

There are four types in the pyramid theoretical, effective, practical and matched storage capacities, and each of them is a subset of
the previous one except theoretical capacity. The theoretical storage capacity (Bachu et al., 2007) represents the physical limit that
the geologic system can accept, and it occupies the entire resource
pyramid. The effective storage capacity (Bachu et al., 2007) can be
acquired by applying a range of technical (geological and engineering) cutoff limits in a storage capacity assessment. This paper will
concentrate on the theoretical and effective storage capacities with
some introduction to their calculation methods. The calculations
are based on the assumption that reservoir pressure is above the
bubble point pressure of crude oil, and only oil and water in the
pore space without free gas.
The volumetric balance theory is commonly adopted to calculate storage capacity (Bachu and Shaw, 2003, 2005; Bachu et al.,
2007). However, the effect of CO2 dissolved in the remaining oil and
water of the storage capacity, which is considerable and essential
in the storage capacity calculation (Enick and Klara, 1990; Kovscek,
2002), was not been taken into consideration in their studies. In this
study, the CO2 storage capacities of oil reservoirs were calculated
by using the volumetric balance theory, and the CO2 dissolved in
the remaining oil and water was considered.
2.1. Calculation of theoretical storage capacity

Others (Shaw and Bachu, 2006; Shen and Liao, 2009; Brennan
et al., 2010) presented a number of various methods according
to different reservoir development characteristics, but all of them
were based on the volumetric balance theory.
In this paper, CO2 dissolved in remaining oil and water,
volumetric sweep efciency and displacement efciency were
implemented according to the development characteristics of Xinjiang oileld in China, while using theoretical and effective storage
capacity calculations. The three key parameters Rf , Ef and SCO2 were
calculated with three different methods, which have been applied
to the Xinjiang oileld for storage capacity calculations. The work
provides a reference for the feasibility study of CO2 stored in the
Xinjiang oil reservoirs.
2. Calculation of storage capacity
The techno-economic resource-reserve pyramid for CO2 storage capacity (CSLF, 2005; Bachu et al., 2007) is shown in Fig. 1.

The theoretical storage capacity in oil reservoirs can be divided


into three parts: free-phase CO2 , CO2 dissolved in the remaining oil,
and CO2 dissolved in the remaining water. The rectangular sketches
(shown in Fig. 2-12-3) are adopted to present the components of
the total pore volume (entire rectangle) in the theoretical storage
capacity. After CO2 ooding, the total pore volume is lled with
free-phase CO2 , remaining oil and water. And the volume of the
remaining oil and water are composed of two parts: swept and unswept parts. In this way, the CO2 dissolved in the remaining oil and
water that had been swept by CO2 was considered.
The calculation model is shown by Eqs. (3)(6). The terms that
were shown in Eqs. (3)(6) are summarized in Table 2. Eq. (3) is the
sum of the three parts of the storage capacity. The storage capacity
of the free-phase CO2 (Eq. (4)) is a derivation of Eq. (1). Most oil
reservoirs in the Xinjiang oileld have been developed by water
ooding over a long period of time. The historical data such as
the cumulative volume of injected water, produced water and oil,
and original oil-in-place are incomplete, which signicantly affects
the accuracy of the CO2 storage capacity calculation. Therefore,
the storage capacity is estimated based on the present situation of
oil reservoirs. Therefore, the present oil-in-place is adopted in Eq.
(4), and the present water-in-place is adopted in Eq. (5). Eqs. (5)
and (6) are the storage capacities that CO2 dissolved in the swept
remaining oil and water respectively. The sweep efciency denominators of remaining oil and remaining water are POIP (1 Rf )
and PWIP + Viw Vpw respectively. However, there are no appropriate methods to obtain the sweep efciencies of the remaining
oil and water. The two sweep efciency may have the same values,
and they are near to the overall volumetric sweep efciency. Therefore, it is assumed that the values of the sweep efciencies of the
remaining oil and water are equal to that of the overall volumetric
sweep efciency.
MCO2 t = Mfree-phase
Mfree-phase

CO2

CO2

+ MCO2 in sweptoil + MCO2 in sweptwater

= CO2 r [Rf POIP Viw + Vpw ]

MCO2 in sweptoil = CO2 r Ef POIP (1 Rf ) mCO2 in oil

Fig. 1. Techno-economic resourcereserve pyramid for CO2 storage capacity (CSLF,


2007).

(3)
(4)
(5)

MCO2 in sweptwater = CO2 r Ef (PWIP + Viw Vpw ) mCO2 in water


(6)

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

27

Fig. 2. Rectangular sketches for showing the components of the total pore volume. Fig. 2-1 shows the components of pore volume before CO2 ooding, Fig. 2-2 shows the
components of pore volume after CO2 ooding, Fig. 2-3 shows the components of the swept pore volume after CO2 ooding, Fig. 2-4 shows the components of the pore
volume for effective storage capacity after CO2 ooding. The entire outer rectangle represents the total pore volume, including POIP and PWIP before CO2 ooding and
free-phase CO2 , remaining oil and water after CO2 ooding. Both the remaining oil and water were composed of swept and un-swept parts. The pore volume for effective
storage capacity is part of the volume of the free-phase CO2 , the swept remaining oil and the swept remaining water.

mCO2 in oil is calculated by Eqs. (7)(9) (Xue et al., 2005).


mCO2 in oil =

KP
44xT
1 + KP Mave

ln K = 773.247

(7)

1
18.5898
T 273.15

= 0.00008286(T 273.15) 0.7287

(8)
(9)

The mCO2 in water can be determined by the estimation model


(Duan and Sun, 2003). This model was based on a specic particle
Table 2
Parameters for theoretical storage capacity estimation.
Parameter

Units

Description

Mfree-phase CO2
MCO2 in sweptoil

kg
kg

MCO2 in sweptwater

kg

Rf

m3 /m3

POIP
PWIP
Ef

m3
m3
m3 /m3

mCO2 in oil

m3 -CO2 /m3 -oil

mCO2 in water

m3 -CO2 /m3 -H2 O

K
P
os
xT

106 Pa
(kg/m3
kg/kg

kg/kg

Mass of free-phase CO2


Mass of CO2 dissolved in the swept
remaining oil
Mass of CO2 dissolved in the swept
remaining water
Oil recovery factor from CO2
ooding, dened as the ratio of
cumulative volume of produced oil
to present oil-in-place (POIP)
Present oil-in-place
Present water-in-place
Overall volumetric sweep
efciency, dened as the ratio of
swept pore volume to total pore
volume
CO2 solubility in oil at reservoir
pressure and temperature
CO2 solubility in water at reservoir
pressure and temperature
Equilibrium constant
Oil reservoir pressure
Oil density at standard condition
Weight fraction of hydrocarbon
components in oil
The secondary dissolving
coefcient that indicates the
dissolving capacity of CO2 in the
mixture composed of hydrocarbon
components and CO2
Average molecular weight of
hydrocarbon in oil
Oil reservoir temperature
Present average water saturation
over entire pore volume
Water recovery factor which is the
ratio of the volume of water
produced after CO2 ooding to
volume of the present
water-in-place

Mave

g/mol

T
Swp

K
m3 /m3

Rw

m3 -H2 O/m3 -PWIP

interaction theory for the liquid phase and a highly accurate


equation of state for the vapor phase, and has a broad application
with temperature ranges from 0 C to 260 C and pressure ranges
from 0 to 200 MPa.
It was assumed that there was no aquifer beneath the oil reservoir in this paper. The volume of injected and invaded water before
CO2 ooding is not considered, and the method is limited to continuous CO2 ooding, so Viw is zero. Thus, Eqs. (4) and (6) can be
substituted by Eqs. (10) and (11) respectively.
Mfree-phase

CO2

= CO2 r POIP Rf + PWIP Rw

(10)

MCO2 in sweptwater = CO2 r Ef (PWIP Vpw ) mCO2 in water


(11)

2.2. Estimation of effective storage capacity


Many parameters out of considered in the theoretical storage
capacity, such as buoyancy, mobility ratio and heterogeneity, could
reduce the pore volume available for CO2 storage (IPCC, 2005). It
would make the effective storage capacity inevitably less than the
theoretical storage capacity. The rectangular sketches (shown in
Fig. 2-3 and 2-4) are adopted to illustrate the components of the
total pore volume in the effective storage capacity. As shown in
Fig. 2-4, the pore volume for effective storage capacity is part of
all three volumes: the free-phase CO2 , the swept remaining oil and
water. A calculation of effective storage capacity was introduced by
CSLF (Bachu et al., 2007):
MCO2 e = Cm Cb Ch Cw Ca MCO2 t = Ce MCO2 t

(12)

A calculation method for effective storage capacity is generated


after substituting the results of Eqs. (3), (5), (10) and (11) into Eq.
(12). Detailed derived steps for Eqs. (13)(17) are shown in the
Appendix.
MCO2 e = CO2 r A h  (1 Swp ) SCO2
SCO2 = Ce (Sfree-phase

CO2

(13)

+ SCO2 in sweptoil + SCO2 in sweptwater )


(14)

Sfree-phase

CO2

= Rf +

Swp
Rw
1 Swp

SCO2 in sweptoil = Ef (1 Rf ) mCO2 in oil


SCO2 in sweptwater = Ef

Swp
(1 Rw ) mCO2 in water
1 Swp

(15)
(16)
(17)

28

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

Table 3
Parameters for effective storage capacity estimation.
Parameter

Units

Description

MCO2 e
Cm
Cb
Ch
Cw
Ca
Ce

kg
(kg-CO2 -effective)/(kg-CO2 -theoretical)
(kg-CO2 -effective)/(kg-CO2 -theoretical)
(kg-CO2 -effective)/(kg-CO2 -theoretical)
(kg-CO2 -effective)/(kg-CO2 -theoretical)
(kg-CO2 -effective)/(kg-CO2 -theoretical)
(kg-CO2 -effective)/(kg-CO2 -theoretical)

SCO2
SCO2 displace
SCO2 in oil
SCO
2 in water

m3 -CO2 /m3 -POIP


m3 -CO2 /m3 -POIP
m3 -CO2 /m3 -POIP
m3 -CO2 /m3 -POIP

Effective CO2 storage capacity


Capacity coefcient effected by mobility
Capacity coefcient effected by buoyancy
Capacity coefcient effected by heterogeneity
Capacity coefcient effected by water saturation
Capacity coefcient effected by aquifer strength
Effective capacity coefcient that incorporates the
cumulative effects of all the other parameters
Storage factor
Storage factors for the free-phase CO2 in pore volume
Storage factors for CO2 dissolved in remaining oil
Storage factors for CO2 dissolved in remaining water

The storage factor SCO2 from Eq. (14) reects the effective CO2
storage capacity in oil reservoir. Table 3 summarizes the terms
shown in Eqs. (12)(17).
2.3. Determination of key parameters
Rf , Ef and SCO2 are the three key parameters for the calculation
of theoretical (Eqs. (3)(6)) and effective (Eqs. (13)(17)) storage
capacities, and these three methods are introduced to determine
them.
2.3.1. Numerical reservoir simulation
The numerical reservoir simulation is a mature and accepted
method. The steps to establish a numerical reservoir simulation
include setting up a geological model, establishing a numerical
reservoir model, and adjusting parameters of rock and uid through
history matching. The Eclipse software (Schlumberger Ltd.) was
adopted to establish a numerical model for accurately simulating CO2 storage conditions, and the numerical models of three oil
reservoirs at the Xinjiang oileld were used.
Numerical reservoir simulation is considered as a very reliable
estimation method for reservoir development. Certain data should
be chosen to establish an actual numerical reservoir model, such as
reservoir grid, earth coordinates of production and injection wells,
well trajectory, log interpretation, well test, well measures, historical dynamic and other relevant data. For some oil reservoirs, this
data can often be difcult to obtain, which represents a challenge
for the calculation of the three parameters. Consequently, in some
situations, two additional methods of parameter determination are
presented in Sections 2.3.2 and 2.3.3.
2.3.2. Empirical formula
The three key parameters Rf , Ef and SCO2 can be obtained by
empirical formulae. The Rf can be obtained from an empirical relationship (Stevens and Kuuskraa, 1999) derived from seven eld EOR
projects (Fig. 3). At the Xinjiang oileld, Ef was assumed to be 0.58,
according to CO2 ooding that adopted by 2D visible experiment.
The SCO2 can be calculated by Eqs. (14)(17). Among these equations, the water recovery factor Rw can be estimated by Eq. (18)
(Rice, 2007).
Rw =

Rf (1 Swp )
Swp 1.8149912 10144 e(665492.66)/t

(18)

where t is production time, year.


2.3.3. Stepwise regression
Mathematics and statistics were used in the stepwise regression method (Zhu, 2012). It has been widely adopted to establish
the functional relationship between the dependent and independent parameters so as to analyze experimental data. Dependent

Fig. 3. Oil recovery empirical relationship (Stevens et al., 1999).

parameters Rf , Ef and SCO2 were obtained from conceptual models of numerical reservoirs which are established with different
independent parameters including permeability, heterogeneity,
oil saturation, uid mobility, injection and production pressures,
reservoir pressure, and injection time. The stepwise regression was
adopted to establish the relationship between the dependent and
independent parameters. This method will be helpful for evaluating
CO2 storage capacity in oil reservoirs.
The basic conceptual model for the numerical reservoir was
an inverted pattern, including four production wells and a central
injection well. Each grid block measured 25 m 25 m 2 m, over
the entire grid of 37 25 5 cells. Top depth and reservoir temperature were 1980 m and 335.3 K, respectively. Relative permeability
graphs are shown in Fig. 4. The three-parameter PengRobinson
equation of state was used for ash calculation. Different conceptual models with different independent parameters of a numerical
reservoir were expressed using the basic conceptual model of a
numerical reservoir.
The independent parameters, i.e., permeability, heterogeneity,
oil saturation, uid mobility, injection and production pressures,
reservoir pressure, and injection time were transformed to be
dimensionless. The dimensionless equations of the independent
parameters are shown in Table 4. Table 5 summarizes the terms
in the dimensionless equations. The permeability ratio between
vertical and horizontal permeability (kv/h ) and permeability variation coefcient (Vk ) were used to characterize the heterogeneity
of the permeability. The oilCO2 mobility ratio (Mgo ) was adopted
to characterize the mobility of CO2 in the reservoir. Because the

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

29

Fig. 4. Curves of wateroil and gasoil relative permeability. Fig. 4-1 is the curves of wateroil relative permeability, and Fig. 4-2 is the curves of gasoil relative permeability.

Table 4
The dimensionless equations and denitions of the independent dimensionless parameters.
Equations

Denitions

Kv/h = kv /kh
Sop

Ratio of vertical permeability to horizontal permeability


Present oil saturation, (1 Swp )

Mgo =

o 
krg
o
o 
kro
g

Mobility ratio of CO2 displacing oil




n
2  n
n
n
 
ki
hi ki /
hi 1
i=1
i=1
i=1
n
/
hi ki /
hi 1
n

Vk =

i=1

hi 1

i=1

Permeability variation coefcient (Liu et al., 1993)

i=1

PinjD = Pinj /PMM


PpD = Pp /PMM
PrD = Pr /PMM
tD = Vinj /POIP

Dimensionless injection pressure


Dimensionless production pressure
Dimensionless reservoir pressure
Dimensionless injection time

Table 5
Parameters for the dimensionless equations.
Parameter

Units

Description

kv
kh
o
krg
o
kro
o

m2
m2
m2 /m2
m2 /m2
Pa s

g

Pa s

hi

ki
Pinj
PMM
Pp
Pr
Vinj

m2
Pa
Pa
Pa
Pa
m3

Vertical permeability
Horizontal permeability
End-point relative permeability of CO2
End-point relative permeability of oil
Oil viscosity at reservoir pressure and
temperature
CO2 viscosity at reservoir pressure and
temperature
Thickness of layer i, subscript i is the
layer number
Permeability of layer i
Bottom hole injection pressure
Minimum miscible pressure
Bottom hole production pressure
Average reservoir pressure
Cumulative volume of injected CO2 at
reservoir pressure and temperature

design is an independent quadratic design requiring three levels


(low, intermediate and high) for each independent dimensionless
parameter so as to capture quadratic effects. Treatment combinations are at midpoints of the process space, which essentially means
that a BBD never uses all the extreme values at the same time. The
experimental design included 120 groups of simulations, which
contained eight independent dimensionless parameters with each
parameter having three levels. All the simulation runs were performed by Eclipse.
The stepwise regression method (Zhu, 2012) was applied in
parametric regression to establish the estimation model shown by
y = C0 +

n


ai xi +


ij

i=1

bij xi xj +

n


ci xi2

(19)

i=1

The model included multinomial and random error of the


parameters. This non-linear model can be converted into a linear
one through parameters substitution. The converted equation is
shown as follows:

miscibility between CO2 and crude oil is determined by minimum


miscible pressure, the dimensionless parameters of reservoir pressure, injection pressure and production pressure were introduced.
The dimensionless injection time tD was used to characterize the
cumulative volume of injected CO2 . The values of the independent
dimensionless parameters are shown in Table 6.
BoxBehnken design (BBD) (Box and Behnken, 1960) was
introduced to design the experiment. BoxBehnken experimental

y1

e1

y
e

2
2
2
, X = (xij ) , e = , =
mp

.
...
...
..

Y =

ym

em

(20)

Table 6
The values of dimensionless independent parameters corresponding to the three levels.
Levels

Kv/h

Sop

Mgo

Vk

PinjD

PpD

PrD

tD

Low
Intermediate
High

0.1
0.55
1

0.40
0.55
0.70

27.5
51.75
76.0

0.05
0.455
0.86

1.21
1.305
1.40

0.61
0.705
0.80

0.89
0.985
1.08

0.2
0.7
1.2

30

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

Fig. 5. Relation graphs of predicted values and actual values. Fig. 5 1Fig. 5-3 are the relation graphs of predicted values and actual values for Rf , Ef and SCO2 respectively.
The straight lines in the gures represent places where predictive values are equal to actual values. These lines were used as the standard to measure matching between the
predicted and the actual values.

Eq. (20) can be expressed by Eqs. (21) and (22).


Y = X + e

Q () =

m


(21)

[yi (1 xi1 + 2 xi2 + + p xip )]

(22)

When Q() was at its minimum, the corresponding would be


the regression coefcient. When X was a non-singular matrix, the
regression coefcient was as follows. Table 7 summarizes the terms
in Eqs. (19)(23).
= (X  Y )

XY

(23)

Rf , Ef and SCO2 obtained from the performed simulations were


taken as the dependent parameters. It is assumed that the independent dimensionless parameter would not affect the dependent
parameter when the signicant level is very low. The least significant level of p = 0.05 was used. Rf , Ef and SCO2 were calculated by
the stepwise regression method. The equations are as the follows:

j=1

Rf = 0.7066 + 0.2716PinjD 0.1468kv/h Sop + 0.001151kv/h Mgo + 0.07318kv/h Vk 0.2682kv/h PinjD


+ 0.04611 kv/h tD + 0.8285Sop PpD + 0.1237Sop tD 0.2110Vk PinjD 0.2606Vk tD 0.1629(kv/h 1.622)

(24)

2 0.4109(t 1.087)2
1.866(Sop 0.3794)2 0.7967(Vk 0.7191)2 0.00001665Mgo2 0.2228PpD
D

Ef = 1.258 0.2471kv/h Sop + 0.002169kv/h Mgo + 0.1172kv/h Vk 0.3054kv/h PinjD + 0.1702kv/h tD


+ 0.3359Sop tD + 0.001323Mgo PinjD 0.2372Vk PinjD 0.2563Vk tD + 0.3503PinjD tD + 0.07813PrD PpD

(25)

0.3416(kv/h 1.158) 0.7520(Sop 0.3967)2 0.00003974Mgo2 0.6096(Vk 0.8701)2 0.5066(tD 0.4603)2


SCO2 = 0.6575 0.3143kv/h Sop + 0.1564kv/h tD 0.003718Sop Mgo + 0.4652Sop Vk 0.3406Sop PinjD
+ 0.001081 Mgo Vk + 0.001024Mgo tD 0.1993Vk PinjD 0.5189Vk tD + 0.4828PinjD tD
2

(26)

+ 0.1410 PpD tD 0.1665(kv/h 1.057) 1.729(Sop 0.6120) 0.6182(Vk 0.6091) 0.2247tD2


Table 7
Parameters for stepwise regression.
Parameter

Description

y
C0
ai , bij , ci
xi , xj

Dependent parameters
A constant
Regression coefcients
Independent parameters, i and j are the order number of
/ j
the parameters, and i =
Matrix of dependent parameters
Dependent parameters
The number of simulations
Matrix of independent parameters
Independent parameters or their combinations
the number of independent parameters (x), n = 8
The number of independent parameters and their
combinations
Matrix of random errors
Random errors
Matrix of multinomial coefcients
Multinomial coefcients
Sum of squared errors of each parameter

Y
y1 , y2 ,. . ., ym
m
X
xij
n
p
e
e1 , e2 , . . ., em

1 , 2 , . . ., p
Q()

According to the above equations, predicted values were compared with actual values to acquire their relational graphs, as shown
in Fig. 5. The straight lines in the gures represent places where the
predicted values are equal to the actual values. As shown in those
gures, the predicted and actual values of Rf , Ef and SCO2 are well
matched, and their correlation coefcients are 0.9538, 0.9411 and
0.9519, respectively. This demonstrates that the established equations obtained by the stepwise regression method can be used to
estimate Rf , Ef and SCO2 . It should be pointed out that the dimensionless parameters are limited in ranges. The correlations can be
used within the ranges. Further studies need to be conducted if the
values of some dimensionless parameters are beyond ranges. The
variation tendency of Rf , Ef and SCO2 with different parameters are
obtained using the three equations. As shown in Fig. 6, four graphs
have been added to show the variation tendency of Rf , Ef and SCO2
with different Kv/h , Sop , Mgo and Vk respectively. Because the inuence of PinjD , PpD and PrD to Rf , Ef and SCO2 are insignicant, their
corresponding graphs will not be given in this paper.
Rf , Ef and SCO2 increase with the increase of tD . The growth is
obvious at the beginning but the growth tapers off after tD = 0.8

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

31

Fig. 6. Rf , Ef and SCO2 change with tD with different values of the parameters. Fig. 6 1Fig. 6-4 are the variation tendency of Rf , Ef and SCO2 with different Kv/h , Sop , Mgo , Vk
respectively. And Rf , Ef and SCO2 change with tD .

Fig. 7. Rf , Ef and SCO2 change with tD with different values of the parameters. Rf and SCO2 are different with that of Fig. 6. The denominators of the new Rf , Ef and SCO2 are all
total pore volume. Fig. 7-1Fig. 7-4 are the variation tendency of Rf , Ef and SCO2 with different Kv/h , Sop , Mgo , Vk respectively. And Rf , Ef and SCO2 change with tD .

32

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

Fig. 8. The absolute differences of Rf , Ef and SCO2 change with tD . Fig. 8 1Fig. 8-3 show the variation trends of the absolute differences respectively. The absolute differences
are the different values of Rf , Ef and SCO2 respectively, which correspond to the maximum and minimum values of Kv/h , Sop , Mgo and Vk respectively.

because of CO2 breakthrough. Rf , Ef and SCO2 increase with the


increase of Sop , and decrease with the increase of Kv/h , Mgo and Vk .
In this paper, oil recovery is dened as the ratio of cumulative produced oil to POIP. POIP is often the oil-in-place after water ooding,
which makes it less than OOIP. Therefore, the oil recovery in this
paper is seemingly higher than the oil recovery that dened as the
ratio of cumulative produced oil to OOIP. The injection pressure
and reservoir pressure in many models are higher than minimum
miscible pressure. Therefore, most oilCO2 systems in these models
can achieve miscibility which can leads to higher oil recovery. There
are still some other situations that can achieve higher oil recovery,
such as homogeneous reservoir, low density and viscosity of the oil,
and well shot wells. All these situations can easily lead to increase
of swept volume which may further result in oil recovery increasing. However, the oil recovery in this paper has rarely been used in
other references. For comparison, the oil recovery can be converted
to the usual oil recovery. An example of this conversion has been
presented. POIP is 0.7 times of OOIP after water ooding. The oil
recovery is 0.25 after CO2 ooding, with the denominator of POIP.
If it is converted to the oil recovery with the denominator of OOIP,
the value is 70% of the oil recovery, i.e. 0.175.
In order to compare with Ef , Rf and SCO2 were converted to the
values which the denominators are total pore volume. The trends
are shown in Fig. 7. The overall trends are the same as Fig. 6, but the
new Rf and SCO2 are less than the former because their denominators are all total pore volume. Ef is higher than Rf and SCO2 . And Rf
is higher than SCO2 . In order to analyze the further trends of these
correlations, the curves of the absolute differences of Rf , Ef and SCO2
are shown in Fig. 8. The absolute differences values of Rf , Ef and SCO2 ,
which correspond to Vk , Kv/h and SCO2 respectively, are the maximum values. It indicates that the most inuential parameters on
Rf , Ef and SCO2 are Vk , Kv/h and Sop respectively. The absolute differences of the three key parameters, corresponding to Kv/h , increase
with the increase of tD , which indicates that the inuence of Kv/h to
Rf , Ef and SCO2 increase with the increase of tD .The absolute differences of the three key parameters, corresponding to Vk , decrease
with the increase of tD , showing the inuence of Vk to Rf , Ef and SCO2
decrease with the increase of tD .

oil and water in most reservoirs is relatively simple and uid properties are relatively favorable. The oil is of low density and low
viscosity, and is classied as light oil. The high pressure, low porosity, low permeability and strong heterogeneity make some parts of
the oileld difcult to develop via water ooding.
Three reservoirs in Xinjiang oileld (CN, XS and BQ) were chosen to validate the three methods for estimating the three key
parameters. Their locations are shown in Fig. 9. The porosity is
0.160.20, and permeability is 4.7 1015 59.74 1015 m2 . Viscosity of the oil is 1.1 103 4.4 103 Pa s. The specic gravity
of in-place oil is 0.7240.814. The original reservoir temperature
is 335.25351.45 K and its pressure is 20.6 106 29.5 106 Pa.
The irreducible water saturation ranges from 30.4% to 47%. Average reservoir depth is 22252597.5 m, and the development area
is 7.3 106 39.7 106 m2 . Reservoir thickness is 6.910.8 m. Oil
reserve is 7.54 109 28.83 109 m3 . The average values of the
independent parameters including permeability, heterogeneity, oil
saturation, uid mobility and reservoir pressure were obtained
from numerical reservoir simulations and were used to depict the
reservoir features. Then, the independent dimensionless parameters were attained through the dimensionless formulae. The values
of these parameters for the three oil reservoirs are shown in Table 8.
The numerical reservoir simulation, empirical formula and stepwise regression methods were used to estimate the three key
parameters. The numerical simulation has been adopted widely
to simulate CO2 storage in formations (Kumar et al., 2004; Ozah
et al., 2005; Jahangiri and Zhang, 2012), and has been proven to
be acceptable. Therefore, it is assumed that the CO2 storage capacities obtained through the numerical reservoir simulation are the

3. Example storage capacity calculation of Xinjiang


oilelds reservoirs
Xinjiang oileld is in the northern part of Xinjiang Uygur
Autonomous Region in China with an area of 13.487 104 km2
(Fig. 9). It is a representative oileld characterized with features
like low permeability, low porosity, and high pressure. Both permeability and porosity show strong heterogeneity. The distribution of

Fig. 9. Location of the three reservoirs (BQ, XS and CN) in Xinjiang oileld. The lightcolored region is the uplift belt, and the dark-colored region is the depression belt.
The dashed line depicts the boundaries of the regions, and the solid line means the
border line of the oileld.

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

33

Table 8
The values of independent dimensionless parameters of example reservoirs in Xinjiang oileld.
Reservoir

kv/h

Sop

Mgo

Vk

PinjD

PpD

PrD

tD

BQ
CN
XS

0.1
0.13
0.11

0.47
0.45
0.56

41.9
37.4
44.8

0.28
0.35
0.24

1.42
1.32
1.39

0.79
0.61
0.72

1.01
0.89
1.12

0.47
0.61
0.52

Table 9
Calculated values of the key parameters and their corresponding storage capacities
using three different methods for three reservoirs in Xinjiang oileld.

Rf , m3 /m3

Ef , m3 /m3

SCO2 , kg/kg

POIP, 109 m3

MCO2 t , 109 kg

MCO2 e , 109 kg

limited. The calculated CO2 storage capacity of the oileld shows


that it has a very large storage potential.

Reservoir

Numerical
simulation

Empirical
formula

Stepwise
regression

4. Discussion

BQ
CN
XS
BQ
CN
XS
BQ
CN
XS
BQ
CN
XS
BQ
CN
XS
BQ
CN
XS

0.46
0.52
0.43
0.56
0.65
0.59
0.32
0.41
0.37
28.83
22.46
7.54
13.96
12.99
3.77
5.71
5.55
1.70

0.54
0.59
0.49
0.58
0.58
0.58
0.38
0.46
0.42
28.83
22.46
7.54
15.22
13.53
4.03
6.78
6.22
1.93

0.49
0.47
0.40
0.54
0.70
0.53
0.35
0.43
0.35
28.83
22.46
7.54
14.59
12.54
3.53
6.24
5.91
1.61

Most oil reservoirs in Xinjiang oileld have been developed via


water ooding, and most of them are experiencing a stage of high
water production. To store CO2 in such reservoirs, the dissolution of
CO2 in remaining oil and water should be considered. Considering
the heterogeneity of permeability as well as the gravity separation,
which is caused by the density variation of oil and CO2 , sweep efciency must be contemplated. In the present work, the calculation
methods of CO2 storage capacity were based on Rf , Ef and SCO2 of the
reservoirs in the Xinjiang oileld. The three key parameters above
can be obtained in different ways. For oil reservoirs with abundant
and accurate data, the numerical simulation method can determine
those parameters with relatively high precision. When less data
is available, the stepwise regression method based on numerical
conceptual reservoir simulation can be utilized to estimate CO2
storage capacity. The empirical formula method is mainly based
on experimental and empirical data, which are sometimes difcult
to obtain. Therefore, applicable conditions of this method must be
strictly considered before application. Because the stepwise regression method requires less data and has a wide range of application,
it can be utilized to predict storage capacity for oil reservoirs when
lacking the required data.

standard values in this paper. The theoretical and effective storage


capacities of the reservoirs were estimated according to the calculation model of CO2 storage capacity. The result is shown in Table 9.
It shows that Rf , Ef and SCO2 obtained by the stepwise regression are
closer to the numerical simulation as compared to the empirical formula. SCO2 in CN reservoir is higher than in other reservoirs which
makes CN reservoir more suitable for storing CO2 . The theoretical and effective storage capacities of the reservoirs increase with
POIP in general. CO2 storage capacities obtained by the stepwise
regression method are very close to the results obtained through
the numerical simulation. As shown in Fig. 10, most of the relative errors of theoretical storage capacity are less than that of
effective storage capacity. The relative error of stepwise regression
method is 3.089.12%, and that of the empirical formula method
is 4.1218.75%. The average relative errors of the two methods are
6.02% and 10.73%. Therefore, the accuracy of stepwise regression
method is higher than the empirical formula method. This suggests
that the stepwise regression method is more suitable to evaluate
the CO2 storage capacity of Xinjiang oileld when needed data is

5. Conclusions
Based on the volumetric balance theory, this work presented
calculation models for CO2 storage capacity with the CO2 dissolved
in remaining oil and water considered. In these models, Rf , Ef and
SCO2 were the three key parameters. Numerical reservoir simulation, empirical formula, and stepwise regression methods were
used to determine these parameters in the models. After comparison of the results obtained from the three different methods, the
numerical reservoir simulation method is the most reliable. However, it requires substantial geological, reservoir and experimental
data. A reduced amount of data would impact the accuracy and precision of the results. The empirical formula method is mainly based
on experimental and empirical data. Less data is required, but even
so, some are still difcult to acquire. Its applicable condition must
be strictly considered. And the estimation accuracy is relatively
poor. The stepwise regression method is based on the numerical
reservoir simulation method, and only requires small representative blocks of the oil reservoir to build the numerical simulation
after which many parameter values can then be obtained easily. It
does not require a substantial amount of data, but the calculation
results are accurate.
Acknowledgements

Fig. 10. Relative errors of the storage capacities of the three reservoirs in the empirical formula and stepwise regression methods.

This work was supported by the National Basic Research Program of China (973 program, grant no. 2011CB707302), and Chinese
National Major Science and Technology Projects (2011ZX05016006 and 2011ZX05009-004-001).We are grateful to all staff
involved in these projects, and also thank the journal associate
editor and the reviewers.

34

C. Liao et al. / International Journal of Greenhouse Gas Control 28 (2014) 2534

Then, Eq. (3-1-1) can be simplied as follows

Appendix A

MCO2 t = CO2 r A h  (1 Swp )

Detailed derived steps for Eqs. (13)(17)


Eqs. (3), (5), (10), (11), (12) are shown as follows:
MCO2 e = Ce MCO2 t
MCO2 t = Mfree-phase

(12)

CO2

+ MCO2 in sweptoil + MCO2 in sweptwater

(3)

SCO2
Ce

After substituting the last equation into Eq. (12), Eq. (13) is
obtained.
MCO2 e = Ce MCO2 t = CO2 r A h  (1 Swp ) SCO2

(13)

References
MCO2 in sweptoil = CO2 r Ef POIP (1 Rf ) mCO2 in oil
Mfree-phase

CO2

(5)

= CO2 r POIP Rf + PWIP Rw

(10)

MCO2 in sweptwater = CO2 r Ef (PWIP Vpw ) mCO2 in water

(11)

When Eqs. (5), (10) and (11) are substituted into Eq. (3), a new
equation is obtained, as shown in Eq. (3-1).

MCO2 t = CO2 r [POIP Rf + PWIP Rw ]


+ Ef POIP (1 Rf ) mCO2 in oil
+Ef (PWIP Vpw ) mCO2 in water

(3-1)

where, POIP = A h  (1 Swp )


PWIP = A h  Swp
PWIP Vpw = A h  Swp (1 Rw )
(The second equation is based on the hypothesis that continuous
CO2 ooding is adopted only.)
The two equations above are substituted into Eq. (3-1), and then,
Eq. (3-1-1) is obtained.
MCO2 t = CO2 r

[A h  (1 Swp ) Rf + A h  Swp Rw ]

+ Ef A h  (1 Swp ) (1 Rf ) mCO2 in oil


+ Ef (A h  Swp (1 Rw ) mCO2 in water


= CO2 r A h  (1 Swp )
+ Ef (1 Rf ) mCO2 in oil + Ef

Rf +

Swp
Rw
1 Swp

Swp
(1 Rw ) mCO2 in water
1 Swp

= CO2 r A h  (1 Swp ) (Sfree-phase CO2 + SCO2 in sweptoil + SCO2 in sweptwater )

(3-1-1)

where,
Sfree-phase

CO2

= Rf +

Swp
Rw
1 Swp

SCO2 in sweptoil = Ef (1 Rf ) mCO2 in oil


SCO2 in sweptwater

Swp
= Ef
(1 Rw ) mCO2 in water
1 Swp

(15)
(16)
(17)

The SCO2 is dened as below


SCO2 = Ce (Sfree-phase

CO2

+ SCO2 in sweptoil + SCO2 in sweptwater )


(14)

Bachu, S., Shaw, J., 2003. Evaluation of the CO2 sequestration capacity in Albertas
oil and gas reservoirs at depletion and the effect of underlying aquifers. J. Can.
Petrol. Technol. 42 (9), 5161.
Bachu, S., Shaw, J., 2005. CO2 sequestration in oil and gas reservoirs in western Canada: effect of aquifers, potential for CO2 -ood enhanced oil recovery
and practical capacity. In: Proceedings of 7th International Conference on
Greenhouse Gas Control Technologies, Volume 1: Peer-Reviewed Papers and
Overviews. Elsevier, pp. 361369.
Bachu, S., Bonijoly, D., Bradshaw, J., Burruss, R., Holloway, S., Christensen, N.P., Mathiassen, O.M., 2007. CO2 storage capacity estimation: methodology and gaps. Int.
J. Greenhouse Gas Cont. I, 430443.
Box, G.E.P., Behnken, D.W., 1960. Some new three level designs for the study of
quantitative variables. Technometrics 2, 455475.
Brennan, S.T., Burruss, R.C., Merrill, M.D., Freeman, P.A., Ruppert, L.F., 2010. A probabilistic assessment methodology for the evaluation of geologic carbon dioxide
sequestration. In: U.S. Geological Survey, pp. 131.
CSLF (Carbon Sequestration Leadership Forum), 2005. A taskforce for review and
development of standards with regards to storage capacity measurement,
http://www.cslforum.org
CSLF (Carbon Sequestration Leadership Forum), 2007. Task force for review
and identication of standards for CO2 sequestration capacity estimation,
http://www.cslforum.org
Duan, Z.H., Sun, R., 2003. An improved model calculating CO2 solubility in pure water
and aqueous NaCl solutions from 273 to 533 K and from 0 to 2000 bar. Chem.
Geol. 193, 257271.
Enick, R.M., Klara, S.M., 1990. CO2 solubility in water and brine under reservoir
conditions. Chem. Eng. Commun. 90 (1), 2333.
Goodman, A., Hakala, A., Bromhal, G., et al., 2011. U.S. DOE methodology for the
development of geologic storage potential for carbon dioxide at the national
and regional scale. Int. J. Greenhouse Gas Cont. 5, 952965.
IPCC, 2005. IPCC Special Report on Carbon Dioxide Capture and Storage. Cambridge
University Press, Cambridge, United Kingdom/New York, USA, p 442.
Jahangiri, H.R., Zhang, D., 2012. Ensemble based co-optimization of carbon dioxide
sequestration and enhanced oil recovery. Int. J. Greenhouse Gas Cont. 8, 2233.
Kovscek, A.R., 2002. Screening criteria for CO2 storage in oil reservoirs. Petrol. Sci.
Technol. 20 (7-8), 841866.
Kumar, A., Noh, M., Pope, G.A., et al., 2004. Reservoir simulation of CO2 storage in
deep saline aquifers. In: SPE/DOE Symposium on Improved Oil Recovery.
Liu, Z., Xin, Q., Wang, W., et al., 1993. The Description Principle and Method Technology of Reservoir. Beijing Petroleum Industry Press, Beijing, China, pp. 5657.
Ozah, R., Lakshminarasimhan, S., Pope, G., et al., 2005. Numerical simulation of the
storage of pure CO2 and CO2 H2 S gas mixtures in deep saline aquifers. In: SPE
Annual Technical Conference and Exhibition.
Rice, J.A., 2007. Mathematical Statistics and Data Analysis. Duxbury Advanced Series,
third ed.
Stevens, S.H., Kuuskraa, V.A., 1999. Sequestration of CO2 in depleted oil and gas
elds: barriers to overcome in implementation of CO2 capture and sequestration. In: IEA Greenhouse Gas R&D Programme (PH3/22).
Stevens, S.H., Kuuskraa, V.A., Gale, J., 2001. Sequestration of CO2 in depleted oil
& gas elds: global capacity, costs and barriers. In: Proceedings of the Fifth
International Conference on Greenhouse Gas Control Technologies, CSIRO.
Shaw, J., Bachu, S., 2006. Screening, evaluation, and ranking of oil reservoirs suitable
for CO2 -ood EOR and carbon dioxide sequestration. In: Proceedings of SPE/DOE
Symposium on Improved Oil Recovery held in Tulsa, Oklahoma, USA, pp. 2226.
Shen, P.P., Liao, X.W., 2009. The Technology of Carbon Dioxide Stored in Geological Media and Enhanced Oil Recovery. Petroleum Industry Press Beijing, China,
4752.
Xue, H.T., Lu, S.F., Fu, X.T., 2005. Forecasting model of solubility of CH4 , CO2 and N2
in crude oil. Oil Gas Geol. 26 (4), 444449 (in Chinese with English abstract).
Zhu, B., 2012. Developing support software for rapidly predicating the formationsensitivity with analytic method of single correlation and multiple regression.
Sci. Technol. Eng. 12 (32), 86178621 (in Chinese with English abstract).

You might also like