Journal of Petroleum Science and Engineering 187 (2020) 106823

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering

journal homepage: http://www.elsevier.com/locate/petrol

Measurements of CO2 molecular diffusion coefficients in crude oils from

swelling-time curve and estimation using viscosity from the
Stokes-Einstein formula
Ronald Ssebadduka *, Hiroyuki Kono 1, Kyuro Sasaki, Yuichi Sugai, Ronald Nguele
Resources Production and Safety Engineering Laboratory, Department of Earth Resources Engineering, Kyushu University, Fukuoka, Japan

A R T I C L E I N F O A B S T R A C T

Keywords: The CO2 diffusion coefficients in crude oils were measured from curves of oil-swelling vs. elapsed time with CO2
Diffusion coefficient dissolution in the oils. The oil swelling was measured with CO2 up to a pressure of 10 MPa at 50 � C. The swelling
Viscosity coefficient and diffusion coefficient of CO2 and CH4 gases in the oil column were compared with CH4 gas. The
Swelling factor
diffusion coefficient of the heavy oil was evaluated as approximately 1.1–1.6% of that of bitumen. The swelling
Heavy oil
Bitumen
factors increased with pressure, and the diffusion coefficients in CO2 supercritical range were more than twice as
Supercritical high as those in the CO2 gas phase. An empirical equation to estimate gas solubility in crude oils vs. API gravity
has been presented as well. Gas diffusion coefficients were shown to relate to oil viscosity based on the Stokes-
Einstein formula and a new correlation between the two with the absolute average deviation (AAD%) of about
15.5%, derived.

1. Introduction
Global warning caused by green-house gases is the number one cause
of climate change and it is mostly due to increasing atmospheric CO2
concentration. There are two geological strategies to reduce CO2 emis­
sion, the CCS (CO2 Capture and Storage) and CCUS (CO2 Capture, Uti­
lization and Storage). CCS is expensive and generates no revenue, for
this reason, in the last few years the concept of CO2 capture, utilization
and storage (CCUS) has been advanced, based on the fact that utilizing
CO2 captured from anthropogenic sources, particularly in CO2 enhanced
oil recovery (EOR) operations, would help generate revenue that will,
accordingly, offset capture costs and facilitate deployment. The oil
production by injection of captured CO2 from a coal power plant for
example, is expected to help cover the cost associated with CO2 capture
since oil production is associated with profitability. With “next gener­
ation” CO2 storage technology and a monetary reward for storing CO2,
the oil produced by CO2-EOR could be 100þ% “carbon-free (Bachu,
2016; Ferguson et al., 2009)”.
Heavy oil production has been extensively studied especially in
Canada which is rich in heavy and extra heavy oils. Recently, CO2 in­
jection is being used as one of the cold EOR processes for heavy oil re­
covery. To achieve a higher oil production rate and oil recovery factor by
injecting soluble gases such as CO2 or CH4 into oil reservoirs, the CO2
molecular diffusion coefficient and solubility in oil are important factors
to consider in CO2-EOR (Grogan et al., 1989; Grogan and Pinczewski,
1987; Renner, 1988; Sigmund, 1976). If the gas diffusion coefficient can
be estimated without high pressure measurements, numerical simula­
tions of oil flow in the reservoir are easy to carry out (Upreti and
Mehrotra, 2002; Yang and Gu, 2006).
Dissolution of soluble gas contributes to cold production, as it re­
duces oil viscosity and capillary pressure (Alshmakhy and Maini, 2012;
Beal, 1946; Beecher and Parkhurst, 1926; H. H. Li et al., 2013; X. X. Li
et al., 2013; Miller and Jones, 1981; Sasaki et al., 2014; Seyyedi and
Sohrabi, 2017). It also causes swelling of the oil, which increases oil
saturation in micro pores and subsequently the relative permeability of
the oil (Chung et al., 1988; Or et al., 2016; Sasaki et al., 2014; Tunio
et al., 2011). Therefore, accurate measurements of the gas diffusion and
dissolution behavior in crude oil are required, since gas dissolution is
expected to increase oil mobility in the reservoir. These mobility
changes could be used to understand oil drainage mechanisms required
for the cold production process. However, there is little published data
on gas solubility, diffusion coefficients, and swelling factors of heavy
oils. This is especially true for heavy oil because the measurement of gas
dissolution in oil needs a long time up to a few weeks to reach steady

* Corresponding author.
E-mail addresses: ronuka19@yahoo.com (R. Ssebadduka), hiroyuki.kono@japex.co.jp (H. Kono).
1
Present, Japan Petroleum Exploration Co., Ltd.

https://doi.org/10.1016/j.petrol.2019.106823
Received 12 August 2019; Received in revised form 11 December 2019; Accepted 17 December 2019
Available online 20 December 2019
0920-4105/© 2019 Elsevier B.V. All rights reserved.
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

state conditions under high pressure (Li and Dong, 2009; Sheikha et al.,
2006).
In this study, gas diffusion coefficients were measured based on
swelling factors of CO2 and CH4 in different crude oils. The basic
methodology to measure the diffusion coefficients from volume expan­
sion was first reported by (Riazi, 1996) applying the fluids system be­
tween CH4 gas and C5H12 in a PVT cell under constant volume and
pressure. In this report, oil swelling, which is proportional to gas solu­
bility, was measured against the elapsed time to estimate gas diffusion
coefficient in oil by matching with the unsteady-state diffusion equation.
It is relatively easy to measure the coefficient from direct measurements
of gas solubility. The oil swelling was measured from visual inspection of
photographs showing the surface movement of the oil meniscus in a
column which was placed in a high-pressure cell with visualization
windows. Furthermore, the equation of the gas diffusion coefficient was
also formulated from the Stokes–Einstein formula, in which the diffusion
coefficient is inversely proportional to a fractional power of the vis­
cosity. It was compared with the empirical equation. The result shows a
possibility to estimate the gas diffusion coefficient from the known oil
viscosity.
2. Measurement apparatuses and procedures
2.1. Oils, gases and sandstone core
2.1.1. Crude oils and gases
As shown in Table 1, hexadecane and two dead oil samples, one
medium and one heavy, were used for present measurements. The heavy
oil sample was produced in Japan and had the following parameters:
specific gravity, 0.9713 (50 � C); API gravity, 11.5; average molecular
weight, 317; and viscosity, 821 mPa s (30 � C), 412 mPa s (40 � C) and
227 mPa s (50 � C). The medium oil sample was produced in Oman and
had the following parameters: API gravity, 29.3; and viscosity, 9.7 mPa s
(30 � C).
CO2 (>99.9% pure) and CH4 (>99.9% pure) were used as the soluble
gases for injection into the oil samples.
3. Monitoring oil swelling and gas diffusion
3.1. High pressure cell
A high-pressure cell with two visualization windows was used for the
measurement (see Fig. 1). Maximum pressure of the cell was 10 MPa and
the maximum temperature setting of the thermostat was 90 � C. Water
from a water bath kept at constant temperature by a thermostat was
circulated through two jackets around the cell, and used to maintain a
temperature of 50 � C in the pressure cell covered with thermal
insulation.
The gas was injected in the cell up to pressures of 10 MPa using a gas
booster (MAXIMATOR, DLE 15-75-NU-CST). The cell pressure was
monitored with an electric pressure gauge (GE DRUCK Ltd, model DPI
282) with 0.01 MPa resolution. Oil swelling was evaluated by photog­
raphy of the surface movement of the oil column, and then visual
inspection of the photographs. The surface movement was generated
with gas dissolution in the oil columns. Swelling-time curves were
constructed to estimate the diffusion coefficients in the oil from the
unsteady state diffusion equation.
4. Measurement procedure and analysis for oil swelling
Approximately 2 cm3 of oil was placed in the column with a
measuring scale, and the mass of the oil was recorded before the mea­
surements. The oil was then placed inside the pressure cell, and the
pressure cell was sealed. Two different measurements were conducted
for oil compressibility and swelling factors.
a) Oil compressibility test with He gas
A compressibility test was conducted with He gas, which is not sol­
uble in the oil. Air remaining in the cell was replaced with He gas. The
cell temperature was then controlled by circulating hot water. To
determine if the entire oil sample in the cell was at a homogenous
temperature, the oil surface was monitored until the oil expansion
caused by the heating, which took about 2 h, stopped. After confirming
the oil had reached a stable temperature, high pressure He gas was
injected into the sample. Any downward movement in the oil surface
was measured and recorded against the gas pressure. The measurement
was repeated at several pressures up to 9 MPa. Taking P0 as the standard
atmospheric pressure (0.103 MPa); P as the pressure after He gas in­
jection; and V0 and V as the oil column volumes at P0 and P, respectively,
oil compressibility, κ was determined using the following equation:
V V0 1
κ¼ ⋅ (1)
V0 P P0
b) Swelling test with solvent gases
The oil swelling factors were measured using a swelling test with CO2
and CH4 gases injected into the cell. For each oil and gas combination
(heavy oil-CO2, heavy oil-CH4, medium oil-CO2, medium oil-CH4),
measurements were conducted at several different gas pressures, and a
swelling factor was calculated for each measurement. The swelling co­
efficient for each oil-gas combination was then evaluated from the
swelling factors. The gas injection procedure for the oil compressibility
test was used, but this time the measured response was of the upward
movement of the oil surface as shown in Fig. 2. The swelling factor, Sr,
was calculated using the following equation:
ΔV * ΔV *
Sr ¼ 1 þ ¼1þ (2)
V0* V0 ½1 κ⋅ðP P0 Þ�
where, V0* and ΔV* are the compressed original volume of the oil col­
umn and the increased volume after oil swelling respectively. The
swelling factor was measured for gas pressure, P. As shown in Fig. 3, the
oil column length after pressure increases in the cell was monitored with
the time interval of few hours to measure the swelling volume of the oil
at pressures up to 10 MPa.

Table 1
Oil samples used for present measurements. 4.1. Mathematical model of gas diffusion in oil
Sample (Dead oil) API (kg/m3) Viscosity (cP) Compressibility (
� 10 3MPa 1) The gas diffusion model involves the solvent gas dissolving in the oil
from the gas-oil interface and then diffusing through the oil as shown in
@50 C �
@30 C �
@50 C �
@30 C �

0.1 MPa 0.1 MPa 4 MPa 4 MPa Fig. 4. The basic derivative expression of one dimension unsteady state
gas diffusion in columns is as shown in equation (3) (Cussler, 2009;
Hexadecane 55.59 2.8 2.3 0.5 0.5
C16H34 (770.1) Sinha and De, 2012; Westwater and Drickamer, 1957):
Omani Medium Oil 40.27 5.0 9.7 0.6 0.5
(833.1) ∂C ∂2 C
¼D 2 (3)
Japanese Heavy Oil 13.06 220 821 0.5 0.5 ∂t ∂x
(972.5)
Initial and boundary conditions are:

2
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

Fig. 1. Schematic diagram of the measurement system.

Fig. 4. Unsteady dissolution and diffusion 1-D model in the oil column during
contact with high pressure gas.
Fig. 2. Swelling factor VS. Time.

Fig. 3. Typical photos showing heavy oil swelling by CO2 dissolution.

3
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

CðL; tÞ ¼ CA0 ; Cðx; ∞Þ ¼ CA0

∂C (4)
ð0; tÞ ¼ 0
∂x

where C ( ) is the gas concentration at time t and location x, CA0 ( ) is

gas concentration at the surface, and L (m) is the depth of the oil column.
The following assumptions are considered in the present study.

(a) Diffusion coefficient is constant

(b) Oil column length at the time, tc, that showed 66.7% of gas
saturation is substituted to L, since L increased with time.
(c) Density-induced natural convection is negligible
(d) Effect of oil evaporation into gas phase is negligible

The gas concentration, C, is given by the following equation:

Cðx; tÞ 4X∞
ð 1Þnþ1 �
¼1 exp Dλ2n t cosðλn xÞ
CA0 π 2n 1
(5)
n¼1

2n 1
; λn ¼ π; n ¼ 1; 2; 3; ⋯
2L
Z L
Cðx; tÞdx
8 X
∞
1 �
RðtÞ ¼ 0
¼1 exp Dλ2n t (6)
LCA0 π2 n¼1 ð2n 1Þ2

5. Measurement results and discussion

5.1. Swelling factor-time curve

The compressibility test of the heavy oil measured with He gas was κ

Fig. 6. Generalized plots of the unit ratio of swelling factors.

Fig. 7. Measured swelling factor against gas pressure.

¼ 0.40 � 10 3 and that of the medium oil was κ ¼ 0.60 � 10 3 MPa 1

(see Figs. 5–7). These values were used to evaluate net oil swelling by
correcting oil surface movement. Fig. 8 shows the oil swelling factors
obtained by correcting the compressibility with Eq. (2). Oil swelling
factors of the medium oil against CO2 were larger than those of the
heavy oil, while the diffusion time to reach equilibrium gas dissolution
in the heavy oil was 30 times longer than that of the medium oil. In this
study, a unit ratio of the oil swelling factor (Sasaki et al., 2014) before
equilibrium dissolution is defined as:
Fig. 5. Measurement results of swelling-time curve.

4
R. Ssebadduka et al.
Fig. 8. Gas solubility against swelling factor.
Rs ðtÞ ¼
Sf ðtÞ 1
(7)
Sf ∞ 1
where Sf∞ is the swelling factor at equilibrium condition. The unit ratio
of swelling–time curves of Rs(t) can be assumed to be the same function
of time, R(t) in Eq. (6), which shows the time curve of cumulative gas
dissolution into the oil. Both the oil samples produced similar Rs(t) time-
curves (Fig. 9). The progress of oil swelling with time is closely related to
the gas diffusion coefficient, D, in the oil.
5.2. Oil swelling coefficient
The swelling factor at equilibrium below the bubble point pressure is
assumed to be proportional to the pressure and is defined as:
Sf ∞ ¼ 1 þ fsw P (8)
where, fsw is a proportional constant to pressure which is defined as the
swelling coefficient in this study. The swelling coefficient depends on
characteristics of the oil and gas. The swelling factors of heavy and
medium oil samples were measured with increasing gas pressure as
shown in Fig. 10. The relationships between the pressure and swelling
factor for each oil are indicated by a trend line following Eq. (8). The
swelling coefficients of the heavy and medium oils against CO2 were
determined as:
Fig. 9. Typical example of matching Eq. (6) to the measurement results of
swelling ratio.
5
Journal of Petroleum Science and Engineering 187 (2020) 106823
Fig. 10. Correlation between API gravity and diffusion coefficient of CO2.
�
Heavy Oil CO2 : fsw ¼ 0:0137 MPa 1�
(9)
Light Oil CO2 : fsw ¼ 0:0244 MPa 1
These results show that the swelling coefficient of CO2 in the medium
oil is approximately 1.8 times higher than that of the heavy oil for a
pressure range of up to 8 MPa. There was no apparent change in the
relationship between pressure and the swelling factor before and after
the supercritical pressure around 8 MPa, thus CO2 dissolution and oil
swelling were not so sensitive pressure less than 10 MP.
The swelling coefficients of CO2 in the heavy and medium oils were
approximately five-to-six times higher than those of CH4:
�
Heavy Oil CO2 : fsw ¼ 0:0025 MPa 1�
(10)
Light Oil CO2 : fsw ¼ 0:0056 MPa 1
5.3. Equation to estimate gas solubility in oil
The gas dissolution is taken to be proportional to the oil swelling,
based on the assumption that when gas dissolution occurs in the oil, the
gas molecules in the oil phase behave in a similar manner to when they
are in the liquid phase, with limited movement.
The gas solubility Cs (mmol/g) can be estimated from the swelling
factor, Sf∞, using the following empirical equation based on the mea­
surement results as shown in Fig. 8 irrespective of the difference be­
tween CH4 and CO2
�
Cs ¼ 38:1 Sf ∞ 1 (11)
Suppose ρgl (g/cm3) is the apparent density of the dissolved gas in oil,
and is defined as the ratio of the mass of the dissolved gas (g) over the oil
swelling volume expressed as (1- Sf∞).
ρgl ¼ 1:6g=cm3 for CO2
(12)
ρgl ¼ 1:69g=cm3 for CH2
These values of ρgl were determined from Eq. (11) by using molecular
weights of the gasses. The liquid density of CO2 ranges from 0.70 to 1.03
g/cm3 at 2–7 MPa, and CH4 liquid density is around 0.45 g/cm3 for
pressures from 2 to 7 MPa. The apparent densities of CO2 and CH4 in the
heavy oil are around 50% larger than those of liquid CO2 and CH4,
respectively.
5.4. Gas diffusion coefficient in heavy and medium oils
The time curve of Rs(t) is a function of the diffusion coefficient, D
(m2/s), oil column length L(m), and time, t(s). Curve fitting typically
shown in Fig. 9 was performed on the swelling factor data with Eq. (6),
and the gas diffusion coefficients for the oils were calculated from the
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

Table 2 used to confirm the nature of the correlation with focus on the constant
Summary of gas diffusion coefficients. and dependent variable.
Dead oils Gas P *supercritical D
9 5.5.2. Relationship between viscosity and API
MPa � 10 m2/s
Several data for viscosity and oil density of different crude oils
Present Meas. including heavy, light, and medium oils from different companies and
Japanese Heavy Oil CO2 2.81 0.61
κ ¼ 0.40 � 10 3 (MPa 1) CO2 4.88 0.70-0.73
for different countries (engineering toolbox, n.d.) representing a wide
API gravity ¼ 11.5 at 50 � C CO2 7.82* 0.75-0.78 range of API values were used for coming up with the probable equation
CO2 10.06* 1.5-1.6 for the correlation of API gravity at temperature of 15 � C and viscosity at
CH4 6.89 0.49-0.54 20 � C. Different measurements of both the API at 15 � C and viscosity of
Omani Medium Oil CO2 2.82 39–40
Athabasca bitumen (Bachu, 2016.) at 20 � C were used to check whether
κ ¼ 0.60 � 10 3 (MPa 1) CO2 4.90 35–55
API gravity ¼ 29.3 at 50 � C CO2 7.84* 67–69 the correlation is universal to them as well. This was found not to be the
CO2 10.10* 96–97 case but the resultant correlation is found to be accurate in determina­
CH4 3.04 2.3-2.5 tion of the API and viscosity of the present data samples we used. This
CH4 8.97 2.2–2.3 correlation, shown as equation (16) is found to be true for API values
from references
Lloydminster heavy oil CO2 2.0 0.20
from about 13 to 60 � API. In addition, all plots give a general expression
(Yang and Gu, 2006) CO2 3.5 0.46 of the form,
ρ ¼ 0.988g/cc at 23.9 � C CO2 4.2 0.53
CO2 6.0 0.55
μ ffi C2 ⋅expða ⋅ APIÞ (15)
CH4 4.9 0.21
CH4 5.0 0.22 where, C2, and a, are constants for the plots shown in Fig. 11.
CH4 6.0 0.12 When constants C2 and a, have values, 4 � 103 and - 0.15 respec­
CH4 14.0 0.19 tively, the relationship between μ and API gravity in crude oil is
Athabasca bitumen CO2 4.0 0.23
(Upreti and Mehrotra, 2002) CO2 8.0* 0.4
expressed by the following equation.
API gravity ¼ 8.3 at 50 � C CH4 8.0 0.15
μ ffi 4 � 103 expð 0:15 ⋅ APIÞ (16)
Therefore, the relationship between the diffusion coefficient, D and
API is expected to have the following equation.
swelling-time curves (see Table 2). D ffi C1 ⋅C2 1 ⋅expð a ⋅ APIÞ (17)
The diffusion coefficients of the medium oil were 50–80 times higher
than those of the heavy oil. In both oils, the diffusion coefficients D ffi C1 ⋅4 � 10 3
expð0:15 ⋅ APIÞ (18)
increased slightly as the gas pressure increased and the diffusion co­
efficients of supercritical CO2 were more than twice as high as those sub- In this study, using Hexadecane, Oman’s medium oil and heavy oil
critical CO2. Diffusion coefficients for other oils in literatures are also produced in Japan, the CO2 gas solubility, swelling ratio, gas diffusion
listed in Table 2. The diffusion coefficients including those ones were coefficient in oil, oil level basic physical property values such as evap­
plotted against API gravity in Fig. 10. Several measured data are oration rate were measured. In particular, as shown in Fig. 1, a method
distinguished as being in the range of the supercritical CO2. As a result, a of correcting the evaporation rate of oil into CO2 gas was proposed, and
probable exponential correlation between the diffusion coefficient in gas the swelling coefficient of oil by gas melting was measured. Hexadecane
phase and API gravity is expressed as; is an important component of heavy oil, and since it has been used
before in other experiments (Campos et al., 2009; Hardy, 1958; Mat­
D ¼ 2:9 � 10 11
⋅expð0:247 ⋅ APIÞ (13) thews et al., 1987; Simnick et al., 1977; Tanaka et al., 1993; Zwanzig
and Harrison, 1985), it was used as a material for judging the validity of
this measurement.
5.5. Stokes-Einstein formula
5.5.3. Relationship between CO2 gas diffusion coefficient and viscosity
5.5.1. Relationship between viscosity and gas diffusion coefficient in crude
Empirically motivated modifications of the Stokes–Einstein formula,
oil
in which the diffusion coefficient is inversely proportional to a fractional
An important relation from which hydrodynamic radii of the
power of the viscosity, are discussed. We suggest that the results should
aggregate is routinely measured is the classic Stokes-Einstein equation.
be analyzed in terms of an effective hydrodynamic radius
Here, we report a modification in the classical Stokes-Einstein equation
using a mixture theory approach, in order to accommodate the changes 2:0 � 10 5 ðm2 PaÞ
D ffi (19)
in viscosity of the solvent due to the changes in solute size and shape, to μðPa⋅sÞ
implement a more realistic model (Achuthan et al., 2011). Specifically,
The constant of the prediction equation (17) was 2.9 ☓ 10 5, but the
we have focused on validating this model in proto fibril lateral associ­
one from the empirical formula for the measured value is 2.0☓10 – 5
ation reactions along the aggregation pathway, which has been experi­
m2Pa. According to the above formula (19), the CO2 gas diffusion co­
mentally well characterized.
efficient in oil can be predicted from the viscosity μ (Pa⋅s) of oil gener­
Regarding the relationship between viscosity and diffusion coeffi­
ally. The percentage absolute average deviation (AAD%) of this
cient, it is generally known that the diffusion coefficient in solvent is
expression is about 15.5% which is comparable to 15.4% reported for
inversely proportional to the viscosity of solvent, from Stokes-Einstein’s
earlier expressions of diffusion coefficients (Riazi and Whitson, 1993). In
relation (Zwanzig and Harrison, 1985).
another report, the slope measured by linear least squares of the line that
Assuming that viscosity is μ (Pa⋅s) and C1 is a constant, the rela­
gave the correlation was given as 0.55 (Renner, 1988), in another
tionship between D and μ is expressed by the following equation,
report it was 0.44 (Hayduk and Cheng, 1971), and 0.47 in yet a
C1 different paper (McManamey and Woollen, 1973), these values are
D ffi (14)
μ much higher than the line slope of 0.0314 presented in this study.
However the form and correlation coefficients and constants for the lines
Combining equations (13) and (14) produces an exponential rela­
in these reports are different from ours. For example the correlation
tionship between viscosity and API gravity. Graphical data analysis is

6
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

Fig. 11. Relationship between viscosity and API of different crude oils.

diffusion coefficients over time reveals very similar trends.

6. Conclusion

In this study, oil swelling by gas dissolution and gas diffusion coef­
ficient in oil were investigated experimentally. Oil swelling was
measured by CO2 or CH4 dissolution in heavy and medium dead oils at
gas pressures up to 10 MPa. The oil compressibility and oil swelling
factors were measured by monitoring surface movement of oil in a
column placed in a high pressure cell with visualization windows. The
gas dissolution is regarded as the volume of gas liquefaction in oil, and
an equation to estimate the solubility based on the swelling factor was
presented. The apparent gas densities of CO2 and CH4 were calculated to
estimate the solubility from the swelling factor. Unsteady gas dissolution
and diffusion in the oils were mathematically modeled, and the diffusion
coefficients were evaluated by curve fitting of time-swelling factors with
the theoretical model of gas dissolution. Furthermore, the equation of
CO2 gas diffusion coefficient in relation to oil viscosity was presented
based on Stokes–Einstein formula.
The present study is summarized as follows;

1) The oil swelling coefficients factors for CO2 gas dissolution in Jap­
anese heavy oil and Omani medium oil were measured in the high
pressure cell as fsw ¼ 0.0137 MPa 1 and 0.0244 MPa 1, respectively.
2) In this study, the CO2 gas diffusion coefficients in the oils were
evaluated as D ¼ 0.61 to 1.6 � 10 9 m2/s (Japanese heavy oil) and
35 to 97 � 10 9 m2/s (Omani medium oil). The diffusion coefficients
in the CO2 supercritical range were found to be more than twice as
high as those in CO2 subcritical gas range.
Fig. 12. Relationship between CO2 gas diffusion coefficient vs Viscosity in oil.
3) From this work, a new correlation,
Dðm2 =sÞ ffi 2:0 � 10 5 ðm2 PaÞ=μðPa ⋅sÞ was found and can be used to
coefficient for slope ¼ -0.55 was reported as 0.993 (Renner, 1988). estimate the CO2 gas diffusion coefficient (D) using the oil viscosity
The modified Stokes-Einstein equation incorporates an effective (μ).
viscosity for the mixture consisting of the macromolecules and solvent
where the lateral association reaction occurs. This effective viscosity is
Author contribution
modeled as a function of the volume fractions of the different species of
molecules. The novelty of our model is that in addition to the volume
Ronald Ssebadduka and Kyuro Sasaki conceived of the presented
fractions, it incorporates previously published reports on the dimensions
idea. Ronald Ssebadduka developed the theory and designed and per­
of the protofibrils and their aggregates to formulate a more appropriate
formed the experiments with the help of Ronald Nguele and Hiroyuki
shape rather than mere spheres. The net result is that the diffusion co­
Kono., Yuichi Sugai was a consultant in the presentation of the results
efficient which is inversely proportional to the viscosity of the system is
and the methodology used., Kyuro Sasaki, Yuichi Sugai, and Hiroyuki
now dependent on the concentration of the different molecules as well as
kono were both very helpful in the provision of the technical details used
their proper shapes. Comparison with experiments for variations in
in this work. Ronald Ssebadduka was the one who wrote the manuscript

7
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

with help from Kyuro Sasaki. Ronald Nguele helped a lot in the pre­ Exploration Co., LTD. (JAPEX). The authors would like to thank to Mr.
sentation of the graphical data in the manuscript. Masanori Nakano (JAPEX) and Mr. Sunao Imai (JAPEX) for their kind
supports.
Acknowledgement

The Japanese oil samples were provided by Japan Petroleum

Nomenclature

C gas concentration ( )
CA0 maximum gas concentration at surface ( )
Cs gas solubility in oil (mol/g)
D diffusion coefficient (m2/s)
fsw swelling coefficient ( )
L depth or length of oil columns (m)
P gas pressure (MPa)
P0 standard atmospheric pressure (MPa)
R(t) unit ratio of cumulative dissolved gas at t ( )
Rs(t) unit ratio of oil swelling factor ( )
Sr swelling factor ( )
Sf∞ swelling factor at equilibrium ( )
Sf(t) swelling factor at t ( )
t time (s)
V oil column volume (cm3) at P
Vo original oil column volume (cm3) at P0
V0* compressed original volume of the oil column
x length from bottom in oil column (m)
ΔV* increased volume after oil swelling (cm3)
κ oil compressibility (MPa 1)
λn variable (m 1)
ρgl apparent density of the gas in oil (g/cm3) or (kg/m3)
μ viscosity (Pa⋅s)

Appendix

Mathematical model

The one dimensional unsteady state molecular diffusion equation is hardly solved in any general way. The solution depends critically on the
boundary and initial conditions.
∂C ∂2 C
¼D 2 (20)
∂t ∂x
Using the error function (erf) and the shrinking Gaussian then Fourier series;
�
C ¼ a exp λ2 Dt CosðλxÞ (21)
�
∂c=∂t ¼ aλ2 D exp λ2 Dt cosðλxÞ (22)
�
∂C=∂x ¼ aλD exp λ2 Dt sinðλxÞ (23)

� �
∂2 C ∂x2 ¼ aλ2 D exp λ2 Dt cosðλxÞ (24)

The derivative ∂C= , is of the first order and thus calls for a single initial condition (at all x values); typically, the initial concentration distribution
∂t
is given as, C ¼CAO (x) at t ¼ 0.
�
For the spatial derivative ∂2 C ∂x2 , a second order, two boundary conditions usually apply
C ¼CAO (x,t) at x1 ¼ 0
C¼C2 (x,t) at x2 ¼ L, but at this point there is an impermeable boundary so C2¼CAO, this is the same as the concentration after infinite diffusion time.
The impermeable boundary also implies no flux and thus no concentration gradient at that boundary. ∂C=∂x ¼ 0
From equation (21); Fourier transform and express the initial condition as a sum of cosine waves.
X
∞
�
C¼ a exp λ2 Dt cos λx
(25)
n¼0
� �
2n 1
λ¼
2L

8
R. Ssebadduka et al. Journal of Petroleum Science and Engineering 187 (2020) 106823

For n ¼ 1, 2, 3 …
Fourier transform: a ¼ 4π
The different cosine functions add up to,

4 X∞
ð 1Þnþ1 �
C ðx;tÞ ¼ CAO CAO exp λ2 Dt cos λx (26)
π n¼1
2n 1

References McManamey, W.J., Woollen, J.M., 1973. The diffusivity of carbon dioxide in some
organic liquids at 25� and 50� C. AIChE J. 19, 667–669. https://doi.org/10.1002/
aic.690190347.
Achuthan, S., Chung, B.J., Ghosh, P., Rangachari, V., Vaidya, A., 2011. A modified
Miller, J.S., Jones, R.A., 1981. In: A Laboratory Study to Determine Physical
Stokes-Einstein equation for A β aggregation. In: BMC Bioinformatics. BioMed
Characteristics of Heavy Oil after Co 2 Saturation. Society of Petroleum Engineers -
Central, p. S13.
SPE/DOE Enhanced Oil Recovery Symposium, vol. 1981. EOR, pp. 259–268. https://
Alshmakhy, A., Maini, B.B., 2012. Foamy-oil-viscosity measurement. J. Can. Pet.
doi.org/10.2523/9789-ms.
Technol. 51, 60–65.
Or, C., Sasaki, K., Sugai, Y., Nakano, M., Imai, M., 2016. Swelling and viscosity reduction
Bachu, S., 2016. Identification of oil reservoirs suitable for CO2-EOR and CO2 storage
of heavy oil by CO2-gas foaming in immiscible condition. SPE Reserv. Eval. Eng. 19,
(CCUS) using reserves databases, with application to Alberta, Canada. Int. J. Greenh.
294–304. https://doi.org/10.2118/179738-pa.
Gas Control 44, 152–165.
Renner, T.A., 1988. Measurement and correlation of diffusion coefficients for Co2and
Beal, C., 1946. The viscosity of air, water, natural gas, crude oil and its associated gases
rich-gas applications. SPE Reserv. Eng. 3, 517–523. https://doi.org/10.2118/15391-
at oil field temperatures and pressures. Trans. AIME 165, 94–115.
PA.
Beecher, C.E., Parkhurst, I.P., 1926. Effect of dissolved gas upon the viscosity and surface
Riazi, M.R., 1996. A new method for experimental measurement of diffusion coefficients
tension of crude oil. Trans. AIME G-26 51–69. https://doi.org/10.2118/926051-g.
in reservoir fluids. J. Pet. Sci. Eng. 14, 235–250. https://doi.org/10.1016/0920-
Campos, C.E.P.S., Villardi, H.G.D.A., Pessoa, F.L.P., Uller, A.M.C., 2009. Solubility of
4105(95)00035-6.
carbon dioxide in water and hexadecane: experimental measurement and
Riazi, M.R., Whitson, C.H., 1993. Estimating diffusion coefficients of dense fluids. Ind.
thermodynamic modeling. J. Chem. Eng. Data 54, 2881–2886. https://doi.org/
Eng. Chem. Res. 32, 3081–3088. https://doi.org/10.1021/ie00024a018.
10.1021/je800966f.
Sasaki, K., Sugai, Y., Chanmoly, O.R., Kono, H., 2014. CO2 solubility characteristics of
Chung, F.T.H., Jones, R.A., Nguyen, H.T., 1988. Measurements and correlations of the
crude oils related to carbon capture and utilization ( CCU ). Res. Feature “ Micro Rev.
physical properties of CO2/heavy-crude-oil mixtures. SPE Reserv. Eng. 3, 822–828.
” 5–8.
https://doi.org/10.2118/15080-PA.
Seyyedi, M., Sohrabi, M., 2017. Pore-scale investigation of crude oil/CO 2 compositional
Cussler, E.L., 2009. Diffusion mass transfer in fluid systems, ‫ﺯﻥ ﻭ ﻓﺮﻫﻨﮓ‬.engineering
effects on oil recovery by carbonated water injection. Ind. Eng. Chem. Res. 56,
toolbox. n.d. the engineering toolbox [WWW Document]. https://www.engineeri
1671–1681. https://doi.org/10.1021/acs.iecr.6b04743.
ngtoolbox.com/crude-oil-petroleum-viscosity-gravity-density-d_1959.html.
Sheikha, H., Mehrotra, A.K., Pooladi-Darvish, M., 2006. An inverse solution methodology
Ferguson, R.C., Nichols, C., Leeuwen, T. Van, Kuuskraa, V.A., 2009. Storing CO2 with
for estimating the diffusion coefficient of gases in Athabasca bitumen from pressure-
enhanced oil recovery. Energy Procedia 1, 1989–1996. https://doi.org/10.1016/j.
decay data. J. Pet. Sci. Eng. 53, 189–202. https://doi.org/10.1016/j.
egypro.2009.01.259.
petrol.2006.06.003.
Grogan, A.T., Pinczewski, W.V., 1987. Role of molecular diffusion processes in tertiary
Sigmund, P.M., 1976. Prediction of molecular diffusion at reservoir conditions - 1.
Co2Flooding. JPT, J. Pet. Technol. 39, 591–602. https://doi.org/10.2118/12706-
Measurement and prediction of binary dense gas diffusion coefficients. J. Can. Pet.
PA.
Technol. 15, 48–57. https://doi.org/10.2118/76-02-05.
Grogan, A.T., Pinczewski, V.W., Ruskauff, G.J., Orr, F.M., 1989. Diffusion of CO2 at
Simnick, J.J., Lawson, C.C., Lin, H.M., Chao, K.C., 1977. Vapor-liquid equilibrium of
reservoir conditions: models and measurements. SPE Reserv. Eng. 3, 93–102.
hydrogen/tetralin system at elevated temperatures and pressures. AIChE J. 23,
https://doi.org/10.2118/14897-pa.
469–476. https://doi.org/10.1002/aic.690230410.
Hardy, R.C., 1958. Viscosity of normal-hexadecane. J. Res. Natl. Bur. Stand. 61, 433.
Sinha, A.P., De, P., 2012. Mass Transfer: Principles and Operations.
https://doi.org/10.6028/jres.061.036.
Tanaka, H., Yamaki, Y., Kato, M., 1993. Solubility of carbon dioxide in pentadecane,
Hayduk, W., Cheng, S.C., 1971. Review of relation between diffusivity and solvent
hexadecane, and pentadecane þ hexadecane. J. Chem. Eng. Data 38, 386–388.
viscosity in dilute liquid solutions. Chem. Eng. Sci. 26, 635–646. https://doi.org/
https://doi.org/10.1021/je00011a013.
10.1016/0009-2509(71)86007-4.
Tunio, S.Q., Tunio, A.H., Ghirano, N.A., El Adawy, Z.M., 2011. Comparison of different
Li, H., Zheng, S., Yang, D., 2013. Enhanced swelling effect and viscosity reduction of
enhanced oil recovery techniques for better oil productivity. Int. J. Appl. Sci.
solvent(s)/CO 2/heavy-oil systems. SPE J. 18, 695–707. https://doi.org/10.2118/
Technol. 1, 143–153.
150168-PA.
Upreti, S.R., Mehrotra, A.K., 2002. Diffusivity of CO2, CH4, C2H6 and N2 in Athabasca
Li, X., Li, H., Yang, D., 2013. Determination of multiphase boundaries and swelling
bitumen. Can. J. Chem. Eng. 80, 116–125. https://doi.org/10.1002/
factors of solvent(s)-CO2-heavy oil systems at high pressures and elevated
cjce.5450800112.
temperatures. Energy Fuels 27, 1293–1306. https://doi.org/10.1021/ef301866e.
Westwater, J.W., Drickamer, H.G., 1957. The mathematics of diffusion. J. Am. Chem.
Li, Z., Dong, M., 2009. Experimental study of carbon dioxide diffusion in oil-saturated
Soc. 79, 1267–1268. https://doi.org/10.1021/ja01562a072.
porous media under reservoir conditions. Ind. Eng. Chem. Res. 48, 9307–9317.
Yang, C., Gu, Y., 2006. Diffusion coefficients and oil swelling factors of carbon dioxide,
https://doi.org/10.1021/ie900145c.
methane, ethane, propane, and their mixtures in heavy oil. Fluid Phase Equilib. 243,
Matthews, M.A., Rodden, J.B., Akgerman, A., 1987. High-temperature diffusion of
64–73. https://doi.org/10.1016/j.fluid.2006.02.020.
hydrogen, carbon monoxide, and carbon dioxide in liquid n-Heptane, n-Dodecane,
Zwanzig, R., Harrison, A.K., 1985. Modifications of the Stokes-Einstein formula. J. Chem.
and n-hexadecane. J. Chem. Eng. Data 32, 319–322. https://doi.org/10.1021/
Phys. 83, 5861–5862. https://doi.org/10.1063/1.449616.
je00049a012.