Professional Documents
Culture Documents
This paper is to be presented at the Petroleum Societys 6th Canadian International Petroleum Conference (56th Annual Technical
Meeting), Calgary, Alberta, Canada, June 7 9, 2005. Discussion of this paper is invited and may be presented at the meeting if
filed in writing with the technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will
be considered for publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to
correction.
Abstract analyzing the effect of viscosity and the produced gas-oil ratio
(GOR) on the performance of the process. Consequently,
The Vapor Extraction (VAPEX) process has been proposed optimal values for the number of cycles of solvent injection, the
as a viable alternative to steam-based heavy oil recovery soak time and the amount of back production based on the
methods. In this process, a vaporized solvent is injected into a optimal soaking time, is determined.
horizontal well placed higher in the formation, and the diluted The influence of well geometry on oil recovery is also
heavy oil is produced by gravity drainage from a horizontal considered here. Properly positioning the horizontal injectors
production well situated below. One shortcoming of this and producers can enhance significantly the overall oil
process is the slow diffusion of the solvent into the bulk of the production rate.
oil.
The mass transfer mechanism in the VAPEX process
involves molecular diffusion and convective dispersion within a Introduction
porous medium at a microscopic scale phenomena that are
not well understood. In this study, we are proposing an Currently, the economic extraction of viscous heavy oil is a
alternative method to VAPEX. Instead of directly injecting the major challenge in the petroleum industry. To deal with this
vaporized solvent and producing oil, cyclic solvent soaking is problem, Butler and Mokrys (2) have proposed the concept of
applied to a heavy oil reservoir in order to maximize the mixing injecting light hydrocarbon solvent into the reservoir (VAPEX)
time between the solvent and the heavy oil. as an alternative to Steam-Assisted Gravity Drainage (SAGD).
In this paper, the most effective solvent mixture, which must This approach is especially beneficial for thin pay zone
be in its gaseous phase and also close to its dew point pressure reservoirs where the heat losses to the over- and under-burden
under the prevailing reservoir conditions, is determined using a have a negative impact on the economics of such processes. The
1)
swelling test, built in the Winprop ( model. The solvent mixture VAPEX process does not require water recycling and treatment,
is then verified by comparing different solvent compositions yields much lower carbon dioxide emissions and can be
using thermodynamic simulations. The optimal soaking time for operated at reservoir temperature. The capital and operational
a certain amount of injected solvent is also examined by
1
costs are estimated to be much less than those of a SAGD distribution function. Seven pseudo-components, including the
project (3). light ends, were used to characterize the fluid composition in
The concept of VAPEX is very similar to that of SAGD. In the simulator (see Table 1). The initial reservoir fluid
VAPEX, a vapor chamber, rather than a steam chamber, is composition is then compared with the produced oil in a later
formed around the horizontal injector, and the diluted oil flows section to demonstrate the alteration of composition during
by gravity towards the producer. The principal mechanisms in solvent soaking.
this process are quite complex. Toluene, as a solvent, has been
observed, in an experiment (4), to extract Athabasca and Suncor Equation of State (EOS)
coker feed bitumen from a Hele-Shaw cell. In those
It is very important to tune the equation of state to be used
experiments, the solvent-bitumen contact region is composed of
in the simulator with the experimental data so that the phase
a frontal layer bounded by two different, sharp solvent
behavior of the reservoir fluid and the solvent mixture under
concentration layers. In the contact region, the vapor finger
consideration could be more accurately predicted.
displaces the frontal layer, and the solvent within the frontal
The Peng-Robinson Equation of State (1) was tuned by
layer drains partially by dissolution and partially by diffusion,
matching some of the following experimental PVT (13) data:
into the oil.
differential liberation, saturation pressure, density and viscosity
Many simulations (5,6,7), based on the laboratory scale, have
of the reservoir fluid at different pressures. A modified
been conducted to investigate asphaltenes precipitation and
Pederson state model (1) was chosen as the correlation to
solvent dispersion phenomena by using STARS (8). Most such
compute the viscosity of the heavy oil. After tuning the EOS,
simulations used porous media having permeabilities of 6 Darcy
the calculated oil viscosity at saturation pressure was found to
or greater. However, simulations using field scale values for
be 2883 cp, which is very close to the experimentally
permeability have yet to be undertaken.
determined value of 2941 cp.
As is well known (9), the solvent has to be injected in its
gaseous phase, and it has to contain some liquid phase so that
the contact interface between the heavy oil and the solvent can
be enlarged, and so that the viscosity of the heavy oil can be
2-D Simulations
reduced effectively. This places very strict limits on permissible
reservoir operating conditions. Thus finding an effective light Reservoir Description
hydrocarbon solvent mixture is crucial to the proper design of a
2D simulations were run using the compositional simulation
successful project.
model GEM (14). A Cartesian grid block system was employed
Because conventional solvent injection is a slow process,
to construct the reservoir model. (see Table 2). The reservoir
which involves molecular diffusion (10) and convective
depth was 450 m with a reference pressure of 4076 kPa. A
dispersion (11) within the porous medium, we propose that
homogeneous reservoir system with a porosity of 33% and a
solvent soaking be applied, as an alternative, to promote more
horizontal permeability of 2 Darcy was employed. The vertical
effective dilution. Through compositional simulations the
permeability was 1 Darcy. The initial water and residual oil
optimal soaking time could be obtained based on the selected
saturations were 0.25, and the irreducible oil saturation at the
injection rate and the predetermined solvent mixture.
gas-liquid interface was 0.2. The Stone II model (14) was used to
A properly designed well geometry (12) is also very helpful
generate the relative permeability curves.
in maximizing the effectiveness of solvent soaking for the
recovery of heavy oil. Different well spacing configurations
have been tested with the aim of finding the best well locations.
Well Configuration
This paper will first introduce the reservoir model used in Two horizontal well pairs were used to produce the oil from
this simulation study and then describe the methodology applied the reservoir. The horizontal distance between the two well
for determining solvent composition, optimal soaking time, and pairs was 80 m, and the vertical separation distance between the
well spacing. Results of the simulations will then be presented horizontal injector and producer was 5 m for the base case
and discussed. (Figure 1). This vertical distance arose from a drilling design
perspective, where 5 m is the minimum achievable distance for
drilling horizontal well pairs. Both of the injectors were located
Properties of Reservoir Fluids at layer 12 and the producers were located at layer 17. The
overall horizontal well length was set equal to the dimension of
the j direction, which was 600 meters.
Characteristics of the Oil
The reservoir fluid used for this study was obtained by Solvent Injection Rate Determination
recombining stock tank oil (Lloydminster type) with methane. Generally, the solvent injection rate should be designed
The viscosity and density of the dead oil at saturation pressure based on the final results of an economic analysis, and in this
and room temperature were 12,895 cp and 984.2 kg/m3, analysis the net cumulative solvent oil ratio (NCSOR) (15) is one
respectively. The API gravity of the oil was measured to be of the main economic evaluation criteria. Later on, the NCSOR
12.27 API. It has been shown (7) that, injecting a solvent into a value is reported as a reference for evaluating the three
moderately viscous oil is more effective in reducing viscosity simulation cases. On the other hand, since solvent injection is
than with a more viscous heavy oil. Therefore, 15 sm3 of an extremely high-pressure process, in order to prevent the
methane per standard cubic meter of oil was added to the dead reservoir from being fractured, the pressure distribution profile
oil prior to the PVT experiments. The resulting saturation throughout the reservoir during the whole process needs to be
pressure for the mixture was measured to be 4068 kPa at 21oC, monitored, which in turn limits the solvent injection rate.
and after flashing to standard conditions, a final GOR of 14.5 Different solvent injection rates, ranging from 2000 to 4500
sm3/sm3 was obtained. m3/day, were tested and compared in the simulation work. The
An extended analysis has been undertaken in order to simulation results indicated that, within this range, the higher
1)
characterize the C7+ fractions. Using the Winprop ( model, the the solvent injection rate, the more enhanced the cumulative oil
heavy end components of the reservoir fluid were split into production rate. However, because of pressure constraints, an
three pseudo-components using a two-stage exponential injection rate of 4500 m3/day was found to be appropriate. At
2
this injection rate, the prevailing pressure during the whole reaching a peak value. The no soaking case gave the highest
process was around 6000-7000 kPa, which is acceptable based peak GOR value, which means that more of the injected gas
on a formation fracture pressure of 10,000 kPa. was produced rather than dissolved in the oil, as compared with
the other three cases. The longer soaking times give the lowest
Solvent Selection peak GOR values. The results for the 3-month and 4-months
soaking times were very close to each other. We therefore
The determination of the most effective solvent mixture is selected the 3-month soaking time as the optimal choice.
extremely important for designing a successful solvent-based Moreover, it is worth mentioning that the viscosity reduction
heavy oil recovery process. The solvent mixture has to satisfy of the heavy oil occurs instantaneously if the concentration of
the following critical principles under the reservoir conditions: the solvent is large enough at the production well. This
it must contain sufficient vapor phase to fill the cavity chest; it phenomenon has been observed in previous work (7). The
must contain enough liquid phase to dissolve in the viscous oil amount of solvent required for this depends on the time it takes
and dilute it, resulting in a saturation pressure of the solvent for it to be transported over the 5 m distance that separates the
mixture being close to the prevailing reservoir pressure; the injection from the production well.
solvent dissolving into the heavy oil should have a low
viscosity, preferably below 100 cp, which is the maximum 6-Month Injection Case
viscosity constraint for producing the heavy oil to the surface.
The tool used in this paper for searching for the most Similarly, for the 6-month solvent injection case, Figure 5
effective solvent mixture is the swelling test, built into the shows the viscosity profile over time. It looks similar to the 3-
Winprop (1) model. Different solvent mixtures, containing C1, month injection case, with a disturbance (spike) in the middle
C2 and C3 at different mole fractions, were tried in order to find part of the figure. This may be caused by asphaltenes
out the best one. Finally, a solvent composition of 40% C1 and precipitation, but needs to be investigated further. As more
60% C3 was selected by considering the operating pressure solvent has been injected into the reservoir in this case, the time
range of 6000-7000 kPa when injecting at 4500 m3/day. (see at which the 100 cp level is achieved is about two months
Figure 2). The figure shows the saturation pressure of the earlier than in the 3-month injection case.
solvent, and the reservoir fluid mixture is computed to be An 8-month soaking time resulted in a lower long-term
around 6500 kPa at a 50% solvent concentration (mole viscosity as compared with the 6-months soaking time. and a
fraction). This satisfies the solvent criterion described above, similar long-term viscosity to the 10-month soaking time.. Also,
whereby the solvent mixture, while being in its gaseous phase, the GOR profile shown in Figure 6 indicates that an 8-month
was nearly a liquid. soaking time is better than a 6-month soaking time and that it is
very close to a 10-month soaking time. Therefore, an 8-month
Optimal Soaking Time soaking time was determined as the optimal soaking time for
the 6-month solvent injection case.
Based on the most effective solvent mixture and the
The cumulative production rate can be used as another way
predetermined injection rate, 3-month, 6-month and 9-month
to determine the optimal soaking time. In Figure 7, the
injection cycles were considered, followed by a soaking period
cumulative oil production at the end of 10 years of operation
before production. For each injection cycle, the corresponding
(injection-soak-production) is compared to illustrate how much
optimal soaking time was determined as follows. The soaking
time can be neither too long nor too short. If this time is too oil can be produced for different soaking times. As can be seen
long, production is delayed; and if it is too short, the solvent in Figure 7, the longer the soaking period is, the higher the
does not fully dissolved into the heavy oil. cumulative oil production is.
The optimal soaking time for each injection cycle, at a fixed The no soaking and 6-month soaking cases, which resulted
injection rate of 4500 m3/day, was found by examining the in the lowest cumulative amounts of production, had similar
viscosity and GOR profiles for each injection period. Different amounts of cumulative productiont. The difference between the
soaking times were selected for each of the three injection 6- and 8-month soaking times was around 450 m3, which was
periods. The optimal soaking time will result in the most greater than the difference between the 8- and 10- month
effective viscosity reduction and the lowest GOR. This soaking cases. This is another confirmation that the 8-month
procedure will also be applied later to the cyclic solvent soaking time is the optimal soaking time for the 6-month
injection simulations. solvent injection case.
5
30293, presented at the International Heavy Oil SPE/Petroleum Society of CIM 65525, presented at the
Symposium, Calgary, Alberta, June 19-21, 1995. 2000 SPE/Petroleum Society of CIM International
16. ESCOBAR, E., VALKO, P., LEE, W.J. and Conference on Horizontal Well Technology, Calgary,
RODRIGUEZ, M.G., Optimization Methodology for Alberta, November 6-8, 2002.
Cyclic Steam Injection with Horizontal Wells
6
Table 1. Recombined reservoir fluid composition of Table 4. NCSOR Comparison for the 3 Cases
Lloydminster oil sample
produced oil density, kg/m3 977
No. Component Composition MW SG
(mole fraction) density mole mixture
1 C1 23.30% 16.04 0.300 gas kg/m3 fraction density
methane 0.65 40% 1.4
2 C2 0.000% 30.07 0.356
propane 1.91 60%
3 C3 0.008% 44.10 0.507
4 IC4-NC6 0.330% 83.23 0.679 Case 1 Case 2 Case 3
5 C7-C23 28.45% 208.4 0.845
gas consumption, m3 2.48E+06 2.70E+06 2.57E+06
6 C24-C32 15.05% 331.5 0.904
cum. oil production,
7 C33+ 32.86% 800.0 1.035
m3 28197 29720.2 31337.5
gas mass, ton 3468 3773 3603
Table 2. 2-D Simulation settings
oil mass, ton 27548 29037 30617
Grid type Cartesian NCSOR 0.126 0.130 0.118
Grid system 120*1*20
Grid dimension 1m*600m*20m
Porosity 33%
Table 5. Reservoir fluid composition before and after
Permeability i,j 2000 mD
production
Permeability k 1000 mD
Reservoir top depth 450 m No Component Origional oil Produced oil Change
Temperature 21 oC (mole fraction) (mole fraction)
Initial pressure at 450 m 4076 kPa 1 C1 23.30% 0.26% -23.0%
Initial oil saturation 0.75
2 C2 0.000% 0.00% 0.0%
Number of pseudo-Components 7
EOS PR -1979 3 C3 0.008% 9.56% 9.6%
Solvent composition 40%C1 + 60%C3 4 IC4-NC6 0.330% 0.24% -0.1%
Number of horizontal injectors 2 5 C7-C23 28.45% 33.5% 5.1%
Number of horizontal producers 2 6 C24-C32 15.05% 17.7% 2.7%
7 C33+ 32.86% 38.7% 5.8%
Table 3. Cases design
Case 1
Injection Soaking Production Table 6. Comparison of different vertical well spacings
Cycle No. time time time
1 3 0 3 4m 5m 6m 7m
2 6 0 12
Cumulative
3 9 0 18
4 6 0 63 production, m3 30142 31338 32375 29276
3
time in months Difference, m 866.4 2062 3099.9 0
Case 2
Injection Soaking Production
Cycle No. time time time
1 4 3 4
2 8 9 12
3 12 12 56
Case 3
Injection Soaking Production
Cycle No. time time time
1 3 3 3
2 6 8 12
3 9 10 18
4 6 8 34
7
horizontal injectors
horizontal producers
Lloydminster Oil
Swelling Calc.
8000 1.5
1.4
7000
Swelling Factor
Pressure (kPa)
Saturation
1.3
6000
1.2
5000
1.1
Psat S.F.
4000 1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Composition (mol fraction)
3 months
300 4 months
200
100
0
0 500 1000 1500 2000
time, days
8
GOR vs.time
100
no soak
2 months
80
3
GOR, m /m
3 months
4 months
3
60
40
20
0 1000 2000 3000 4000
time, days
200 8 months
10 months
150
100
50
0
0 200 400 600 800 1000 1200 1400
time, days
GOR vs time
170 no soak
6 months
150
GOR, m3/m3
8 months
10 months
130
110
90
70
0 500 1000 1500 2000
time, days
9
Cumulative production vs.time
10000
production ,m 3
Cumulative oil
9000
8000
no soak
7000 6 months
8 months
10 months
6000
0 500 1000 1500 2000 2500 3000 3500 4000
time, days
10 months
60 12 months
40
20
0
0 250 500 750 1000 1250 1500
time, days
GOR vs.time
190 no soak
6 months
GOR, m3/m3
170 10 months
12 months
150
130
110
90
70
0 500 1000 1500 2000
time, days
10
Cumulative production vs.time
14000
production,m 3
Cumulative oil
12000
10000
no soak
6 months
8000
10 months
12 months
6000
0 500 1000 1500 2000 2500 3000 3500 4000
time, days
production profile
70
60 no soak
production rate,
3 cycle
50
4 cycle
m3/day
40
30
20
10
0
0 1000 2000 3000 4000
time, days
25000
20000
15000
10000 no soak
3 cycle
5000 4 cycle
0
0 1000 2000 3000 4000
time, days
11
GOR vs. time
200
no soak
180 3 cycle
GOR, m3/m3
4 cycle
160
140
120
100
80
0 1000 2000 3000 4000
time, days
production, m 3
production rate,
cumulative oil
25000
40
m3/day
20000
30
15000
20
10000
10 5000
0 0
0 1000 2000 3000 4000
time, days
Figure 14. Comparison of production profiles with and without diffusion coefficient
12