Академический Документы
Профессиональный Документы
Культура Документы
K. Alikhlalov(+), SPE, University of Houston, B. Dindoruk(*), SPE, Shell International E&P Inc.
Abstract
Estimated total world in-place oil volumes are between 9 trillion and 13 trillion barrels, of which 30% is conventional
crude. The rest is 30% oil sands and bitumen, 25% is extra heavy oil and 15% is heavy oil. Normally definition of heavy
oil is based on gas free properties of the oil. An oil is classified as heavy oil if the dead oil viscosities are between 100
centipoises to 10000 cp at reservoir temperatures with APIs between 10 to 20. Extra-heavy oil is commonly defined as oil
having a gravity of less than 10 and a reservoir viscosity more than 10 000 centipoises. When reservoir viscosity
measurements are not available, extra-heavy oil is considered to have a lower limit of 4 API (WPC Report 2007). Though
industry has known about heavy oils for a century, it still provides a minute fraction of the worlds production. Because in
many cases oil will not flow at economical rates (mainly due to high viscosities). One of the most important technical
challenges for Enhanced Oil Recovery (EOR) engineers is to develop an economical and feasible recovery process for heavy
oil reservoirs.
Depending on the reservoir-fluid system, commercial rates can be achieved by injecting (heat) steam. However,
injectivity can be one of the key factors that can lead to difficulties to deliver enough heat to contact significant portion of the
reservoir with steam (or for any other potential injectant). The main reason for such low injectivities in targeted heavy oil
reservoirs originates from the excessively high in-situ oil viscosities. In order to facilitate the injection of reasonable quantities
of steam in a continuous manner, cyclic steam injection techniques are often utilized at early stages of such projects, especially
for extra heavy oil reservoirs. From the full project time cycle point of view, a combination of cyclic and continuous steam
injection could be used to overcome this problem. In this paper, we have considered a five-spot pattern and studied optimum
time for the conversion of cyclic steam injection to continuous steam injection for various type of reservoir and fluid properties
using a commercial numerical simulator (STARS, 2009). To our knowledge, this is the first study that addresses the cyclic to
contuinuous steam injection conversion time and correlates it to simple measurable quantities for the benchmark system with
the reservoir/fluid properties selected.
Introduction
Steam injection process includes the injection of steam that can be generated at the surface or downhole. The main
advantage of downhole generation is to reduce the heat losses (i.e., can be a major issue for deeper reservoirs) in wellbores.
The whole process can be continuous or cyclic.
Cyclic steam injection is also known as huff-and-puff or steam soak process. In a broad sense, cyclic steam injection
is a well stimulation process which involves the transfer of heat to the reservoir near the wellbore by periodical injection of
steam. In huff-and-puff one well acts as both injector and producer. The intention of the heat injection is to improve oil
mobility (or perhaps near wellbore clean-up) during the injection and the soaking period by reducing the oil viscosity. Cyclic
steam injection was first applied in late 1950's in Venezuela (Trebolle et. al 1993) to recover bitumen from tar sands which
could not be produced by primary recovery due to very high oil viscosity.
Ultimate recovery by cyclic steam injection is in the range of 5-25 % of total oil in place. Typically 5000 to 20000
bbl-coldwater equivalent (CWE) of steam is injected into a well over some days and weeks during the cyclic steam injection
process. After the injection process, the well is closed for several days to several weeks for the soaking period. Soaking time
usually lasts 3 to 14 days. During this period, it is preferred to maximize heat transfer to the reservoir and to minimize heat
losses to the cap rock. After soaking the well is put back on production (approximately one to several months depending on the
project). The entire cyclic steam injection process is shown in Figure 1 below. Basically each cycle consists of three stages:
injection period, soak period and production period.
(+)
(*)
Now with BP
Adjunct Faculty, University of Houston, Department of Chemical Eng, Petroleum Engineering Program.
SPE 146612
Cycle 1
Cycle 2
Steam
Cycle 3
Steam
Steam
Cycle 4
Steam
80
60
40
20
Soaking +production
Soaking +production
Soaking +production
Soaking +production
0
0
50
100
150
200
250
Time, days
Figure 1: Cyclic Steam Injection Process.
Continuous steam injection is also called steamflooding or steam drive. From design and engineering point of view,
steam drive process is a more complicated process than huff and puff. In steamflooding, at least two separate wells are used,
one for steam injection and the other for oil production. Figure 2 below shows the steam drive process. As can be seen in
Figure 2, just the mere existence of two wells, immediately introduces a larger reservoir scale between the wells, and what
happens to the reservoir and fluid properties between these two wells effects the propagation of the steam and thus the
recoveries.
Injected steam heats the formation around the wellbore and forms a steam zone that grows with continuous steam
injection. Injected steam decreases the oil saturation in the steam zone to a very low value, pushing the mobile oil out of the
steam zone. As the steam zone grows, more oil is moved from the steam to the unheated zones. There the oil accumulates to
form an oil bank.
Unlike the cyclic steam injection, in the steam drive process steam is injected continuously and injected fluid drives
the oil from injector to producer. While steam zone grows, more oil is pushed from the steam zone to unheated zones ahead of
the steam front. There the oil accumulates to form an oil bank. The injected steam condenses to hot water, which also moves
across the steam front, heating and displacing the accumulated oil. Steam creates different regions of temperatures and fluid
saturations while it moves through the reservoir between wells. Figure 2 shows this process.
SPE 146612
Model Description
Many of the commercial simulators are very similar in their mathematical formulations, however there are still
differences in terms of the numerical solution techniques employed and as well as the overall simulator capabilities. In
principle, they all are capable of simulating non-isothermal methods, such as cold waterflooding, hot waterflooding, and
steamflooding and steam injection. In this study, we have used a commercial thermal simulator (STARS 2009, CMGComputer Modeling Group) to investigate the cyclic to continuous steam injection time.
Five-spot well pattern is one of the most commonly utilized configurations. The injection fluid, which is normally
water, steam or gas, is injected simultaneously through the four injection wells to displace the oil toward the central production
well (Figure 3). Therefore, we have built a quarter of a 5-spot well spacing (using Cartesian symmetry element for the
pattern), in order to demonstrate the study project objectives. The following assumptions were made in order to construct this
model:
.
Since the purpose of this study is to investigate injectivity and as well as the optimum conversion time of cyclic
steam injection to steamflooding we have used a set of reservoir and fluid properties that make sense for the proposed
problem. For example, we have not used relatively light oil for our investigation, since such fluid systems will not
exhibit severe injectivity issues for the problem posed here. In our study, we have used the properties of Kern River
field (California) as presented by Chu et al. (1975) with slight modifications (Table 1 and Table 2).
The quarter of 5-spot well spacing where single production and single injection well is assumed to be located on the
corners of the square. Total injection rate decreased four times in order to scale the quarter pattern rates to the full
pattern area.
Only water and oil will be present on the model. No gas effect is considered in this study.
There is no aquifer support on the production. The only water influence is from the injection well (condensing steam).
Injector
Modeled Area
Producer
Reservoir Properties
The reservoir and fluid data taken from Kern River field in California are shown in Table 1. Table 2, shows the
numerical grid system used for the study
Input Parameters for Simulation
Number of Grid Blocks in I (x-Direction)
26
26
10
2.64
ac
Model Area
0.66
ac
X length
170.5
ft
Y length
170.5
ft
Z length
60
ft
Depth
900
ft
Permeability I
3000
md
Permeability J
3000
md
Permeability K
1500
md
SPE 146612
0.000735
1/psi
Oil Compressibility
0.000005
1/psi
Water compressibility
0.000003
1/psi
4000
cp
Oil API
12
Steam Viscosity
0.11
cp
Oil Density
61.5
lb/ft^3
Water Density
62.5
lb/ft^3
Porosity
0.3
Reservoir Temperature
95
Reservoir Pressure
50
Psi
Thermal Conductivity
38.4
Btu/(ft*day*F)
Horizontal Permeability
Vertical Permeability
1.5
Relative permeability data (rock-fluid properties) were also taken from Kern River field (Chu et al. 1975).
Liquid- Gas Relative Permeability
0.9
0.9
0.8
0.8
0.7
0.6
w ater
0.5
oil
0.4
0.3
0.7
0.6
Liquid
0.4
0.3
0.2
0.2
0.1
0.1
Gas
0.5
0.2
0.4
0.6
0.8
Sw , Water Saturation
0.2
0.4
0.6
0.8
SI, Saturation
Numerical Scheme
Grid orientation effects are known to impact the quality of the numerical results in finite difference schemes. Such
grid orientation effects can yield incorrect breakthrough times and as well as distorted recoveries (i.e., Palagi et al. 1994).
Therefore we have used a nine point formulation option instead of a default five point formulation. Nine point formulation
allows flow between a gridblock and all eight surrounding blocks. However, the nine-point option increases the numerical
work required to solve the flow equations. Before embarking on full scale parametric study, we have performed an extensive
grid-sensitivity runs to optimize the number of grids to be used for the base case (26 X 26 X 10 grids). Detailed analysis of
grid sensitivity is shown in Alikhlalov (2010).
Results and Discussion
Cyclic Steam Injection and Injectivity
Cyclic steam injection was used to heat heavy and extra heavy oil reservoirs to maintain production rates at the
economic Oil Steam Ratio (OSR) where injectivity may be an issue for continuous injection, at least at the startup phases, the
injectivity issue is mainly caused by heavy oil viscosity. Therefore, in this study, the performance of the huff and puff process
is investigated by focusing on the injectivity.
SPE 146612
Injectivity Index (Ii) is a measure of the injection rate achievable in a injection well under a given pressure
differential. Injectivity index is a function of fluid and rock properties and in the case of heavy oils, predominantly high oil
viscosity is the main reason for lower injectivities.
We have investigated the early performance of the injection wells in terms of injectivity. Various initial oil viscosities
are introduced for the same base case and thus the injectivity response is analyzed for different sets of heavy oils. A set of
modified Kern River viscosity data is used as part of the base case (Chu et al. 1975). The viscosity of the base case is equal to
4000 cp at a reservoir temperature and pressure (95 oF and 50 psi). The other viscosities are generated by multiplying the base
case viscosity set (Figure 6). As can be seen in the chart below, the highest viscosity case run was for 1000000 cp oil.
Viscosity-Temperature Relationship
1000000
mo1
mo5
mo-viscosity, cp
100000
mo10
mo15
10000
mo20
1000
mo25
mo50
100
mo75
mo100
10
mo125
mo150
mo200
mo250
0.1
10
100
1000
Temperature, F
The main objective in cyclic steam injection is to achieve a defined injection rate with the given constraints (i.e., net
heat injection per unit time, maximum pressure that reservoir can take, etc.). The huff and puff process in our study contains
20 days of injection, 8 days of soaking and 30 days of production. The details of the cyclic steam injector constraints are given
in Table 3.
During the cyclic steam injection, injectivity index of cyclic steam injector increases from cycle to cycle (as the
reservoir heats up). However, as expected, it decreases while increasing the oil viscosities. In our models, the injectivity index
was defined numerically. We have used the wellblock pressures as the pressure that the injector has to overcome, rather than
the reservoir pressure. Therefore numerical injectivity index is the ratio of steam rate divided by the difference between the
pressure at the injector and wellblock pressure at that point in time. Results of injectivity index for various viscosity sets are
given in the following section (Figures 7 and 8).As seen in Figures 7 to 8, initial injectivity is inversely proportional to oil
viscosity, and it takes longer to improve the viscosity as the oil becomes more and more viscous.
20
days
Injection Pressure
519.58
psi
Injection rate
1000
bbl/day
Injection Temperature
470
0.8
Mass f raction
Soaking time
days
Production time
30
days
SPE 146612
22
20
18
16
14
12
Mo1
10
8
6
4
2
0
0
200
400
600
800
1000
1200
1400
Cycle Days
20
18
16
14
12
Mo10
10
8
6
4
2
0
0
200
400
600
800
1000
1200
1400
Cycle Days
Figure 8: Injectivity Index vs. Day, Mo10 (Base Case Viscosity X 10).
Figures 9 (in Days) and 10 (in pore volumes injected, PVI) show how injectivity index varies during cyclic steam
injection process for all viscosity sets considered.
SPE 146612
Injectivity Index
Mo1
22
Mo5
20
Mo10
18
Mo15
16
Mo20
14
Mo25
12
Mo50
10
Mo75
Mo100
Mo125
Mo150
Mo200
0
0
200
400
600
800
1000
1200
Mo250
1400
Days
Figure 9: Injectivity Index vs. Day, for all viscosity sets considered.
Injectivity Index
Mo1
22
Mo5
20
Mo10
18
Mo15
16
Mo20
14
Mo25
12
Mo50
10
Mo75
Mo100
8
`
Mo125
Mo150
Mo200
Mo250
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
PV Injected
Figure 10: PV Water Injected vs. Injectivity Index, for all viscosity sets considered.
As shown in Figures 9 and 10, at early stages injectivity index increases for high viscosities. It happens for viscosity
range of mo50-mo250 (200000-1000000 cp initial viscosity oil) and it lasts only 1-2 cyclic steam injection cycles depending
on the magnitude of the oil viscosity. Analyzing plots of cumulative oil production and cumulative liquid production versus
pore volume injection that are shown in Figures 11 and 12 respectively, it was observed that for high viscosity range (at that
particular period of time), cumulative oil production decreases while cumulative liquid production slightly increases.
Cumulative produced liquid is sum of produced oil and produced water that is injected during injection cycle. This means that,
for high viscosity range, injected water equivalent steam at early stages did not decrease heavy oil viscosity deep into reservoir
SPE 146612
as in low viscosity cases. Due to this, in the first cycles for high viscosity cases, water saturation around the injector well was
higher than for low viscosity range (leading to higher total mobility in the vicinity of the well). This high water saturation
makes injection process improvement more significant for higher viscosity oils than for lower viscosity oils right after the
initial cycle (and also part of the reason is that we are using the block pressures as a standard yardstick). After some time,
injected steam affects the high viscosity oil deeper into the reservoir.
As mentioned earlier, injectivity is one of the key factors that cause difficulties to contact a significant part of the
reservoir with steam and in fact cyclic steam injection is used to overcome this injectivity problem and maintain production at
the economic Oil Steam Ratio (OSR). While a certain injectivity level is achieved, the next is to convert cyclic steam injection
into steam drive in order to get better sweep by driving the oil to producers continuously.
1.0E+01
Mo5
Mo10
PV Produced Oil
1.0E-01
Mo15
Mo20
Mo25
1.0E-03
Mo50
Mo75
Mo100
1.0E-05
Mo125
Mo150
Mo200
1.0E-07
1.0E-02
1.0E-01
1.0E+00
1.0E+01
Mo250
PV Injected Water
Figure 11: PV Oil Produced vs. PV Water Injected (During Cyclic Steam Injection).
10.00
Mo5
PV Produced Liquid
Mo10
Mo15
Mo20
1.00
Mo25
Mo50
Mo75
Mo100
0.10
Mo125
Mo150
Mo200
Mo250
0.01
0.01
0.10
1.00
10.00
PV Injected Water
Figure 12: PV Liquid Produced vs. PV Water Injected (During Cyclic Steam Injection).
SPE 146612
Total
Time,
days
324
412
522
Cum
Oil,
bbl
25360
26623
28188
Table 4: Total Thermal Project Results at various Conversion Injectivity Index, for different
Instantaneous Oil Steam Ratio (OSR Ins) ,Mo1 (Base Case Viscosity).
Mo10
Results at Ins OSR=4 %
Injectivity Conversion
OSR
Cum
Index,
Time,
Oil,
Total Time, RF, WC, Cum,
bbl/d/psi
days
bbl
days
%
% bbl/bbl
7
363
669
52.38 96
0.07
24096
11
482
744
53.71 96
0.07
24690
16
601
836
55.29 97
0.07
25421
Cum
Oil,
bbl
9317
9906
10524
Table 5: Total Thermal Project Results at various Conversion Injectivity Index, for different
Instantaneous Oil Steam Ratio (OSR Ins) , Mo10 (Base Case Viscosity X 10).
The runs are stopped at the same water cut (96 %) values for both cases. It is observed that when the injectivity index
of 7 reached, it is the optimal time to convert from the cyclic steam process to the steamflooding process for the cases
analyzed. Because that conversion time yields the maximum recovery when water cut reaches 96 % and economical limit of
OSR of 4 % for all cases (results at intermediate OSR value of 10 % is also shown in Tables 4 and 5). The elapsed time to
reach injectivity index value of 7 increases with the oil viscosity (Fig. 13). In other words, with increasing oil viscosity value,
it requires more cyclic steam cycles to reach the injectivity index of 7. This also means that with increasing oil viscosity, it
10
SPE 146612
requires more water equivalent of steam injection (and thus more heat). This may also mean that some projects, depending on
the local economics/oil price premise, will not go forward for some of those high viscosity cases (or perhaps higher injector
density/smaller pattern area is needed).
A simple linear relationship is defined between the optimum conversion time and pore volumes of water injection.
PV Injected
2.50E+00
2.00E+00
1.50E+00
1.00E+00
5.00E-01
Simulation Results
Correlation Results
0.00E+00
0
2000000
4000000
6000000
8000000
10000000
Figure 13:PV Injected vs. Viscosity Ratio (Oil Viscosity / Steam Viscosity of 0.11 cp
(Simulation and Correlation Results).
Viscosity ratio (oil viscosity / steam viscosity) versus pore volume water injected for each viscosity value is shown
in Figure 12. The relationship in Figure 12 shows the amount of water required to be injected for different viscosity ratios, in
order to get a predefined injectivity of 7. In other words, when converted to real time, it indicates the optimum conversion time
from cyclic steam injection to steamflooding.
A simple linear equation can be used to represent this relationship (Eq. 1). This correlation is valid for viscosity range
of 4000 cp to 1000000 cp. Again, the reservoir and base case fluid properties are from Kern River field California (Chu et al.
1975):
(1)
Where A = 4.9682E-01
B = 2.3269E-07
After the conversion of the cyclic steam injection to the continuous steam flood, runs continued using instantaneous
oil-steam-ratio limit of 2-4 % depending on the viscosity ratio of the system. We have used the lower limit for heavier (more
viscous) oils since very high viscosity models did not achieve instantaneous OSR of 4 %. Therefore, we have presented our
results at more than one Ins OSR cut-off levels. We also show the Recovery Factor and Cumulative Oil results for such lower
Instantaneous oil steam ratios. Tables 6 to 8 indicate the summary of the simulation runs for OSR = 4-2% (again, the purpose
of extending the runs to this level was to see what levels of recovery can be achieved).
SPE 146612
11
Mo
1
5
10
15
20
25
50
75
100
Ins OSR 4 %
Project Life, day
RF, %
466
75.04
615
61.06
669
52.38
730
47.76
758
43.65
807
41.16
941
28.82
950
18.96
879
11.94
Mo
1
5
10
15
20
25
50
75
100
125
150
200
250
Ins OSR 3 %
Project Life, day
RF, %
506
78.01
695
67.15
765
59.61
823
54.71
850
50.63
899
48.05
1079
39.15
1147
32.08
1169
23.67
1065
14.43
1048
11.01
1233
10.35
1457
10.01
Mo
1
5
10
15
20
25
50
75
100
125
150
200
250
Ins OSR 2 %
Project Life, day
RF, %
566
81.18
800
72.18
894
66.59
969
62.67
1018
59.66
1075
57.47
1271
49.37
1367
43.99
1454
38.64
1478
32.53
1500
25.68
1686
21.47
1641
13.21
12
SPE 146612
Sensitivity Runs
Limited scope sensitivity runs were performed in order to capture the impact of some of the key parameters on the
conversion time (more details can be seen in Alikhlalov, 2010). Specifically, the impact of lower kv/kh and injection rates are
studied, and the results are given below.
Estimation of the Conversion time for kv/kh=0.1
Our earlier results were based on where kv/kh=0.5. Overall, the results obtained based on kv/kh=0.1 are very similar to
the case discussed earlier where kv/kh was equal to 0.5. The main difference between the two cases, as expected, is the time to
reach the optimum conversion time. For the case with lower vertical permeability, conversion time is reached later because of
relatively poorer vertical communication between layers. Steam override effect decreases appreciably when vertical
permeability is decreased from 1500 mD to 150 mD. Injectivity versus PV injected plot for kv/kh=0.1 was very similar to the
base case with kv/kh=0.5 and will not be shown here for brevity.
Similar to the base case, a simple linear relationship between pore volume injected versus viscosity ratio was developed:
PV Injected=A+B*Viscosity Ratio,
(2)
Where A = 1.1356E+00
B = 3.9287E-07.
Figure 14 shows the relationship of PV Injected vs. Viscosity Ratio for both kv/kh cases
PV Injected
3.50E+00
3.00E+00
2.50E+00
2.00E+00
1.50E+00
1.00E+00
KvKh=0.1
5.00E-01
KvKh=0.5
0.00E+00
0
2000000
4000000
6000000
8000000
10000000
SPE 146612
13
The Neuman type rate reduction is given by Equation 3. Injection rate is most drastically reduced with the Neuman
reduction schedule, especially during the first year of injection reduction. Steamflood oil production results from gravity
drainage as well as horizontal movement of the heated oil, where Neuman schedule eliminates the horizontal component of the
fluid flow which stops (or reduces) steam production. This rate reduction scenario is designed to stop steam production at the
producing well. Neuman rate reduction equation is given by (Neuman 1985):
t*
2
Q (t )
= sin 1
Qo
t
0.5
(3)
Methods
Constant Initial Rate
Neuman's method
Linear Reduction
Mo25
Project Life, OSR Ins. 4 %
807
916
856
RF, %
41.16
43.11
41.78
Table 9: Results of Rate Reduction Scenarios, Mo25 (100000 cp initial viscosity oil).
Figure 15: RF, All Methods, and Mo25 (100000 cp initial viscosity oil).
14
SPE 146612
Figure 16: OSR Ins., all methods, Mo25 (100000 cp initial viscosity oil)
Figure 17: Cumulative Water Injection bbl, all methods, Mo25 (100000 cp initial viscosity oil).
SPE 146612
15
References
Alikhlalov, K.S:"Conversion of Cyclic Steam Injection to Continuous Steam Injection," MSc thesis, University of Houston, Houston, May
2010
Chu, C and Trimble A. E..:Numerical Simulation of Steam Displacement-Field Performance Applications, SPE 5016, AIME 49th
Annual Fall Meeting, Houston, TX, October 6-9, 1974
Hong, K.C.:Steamflood Reservoir Management-Thermal Enhanced Oil Recovery, Penn Well Publishing company, Tulsa, Oklahoma, 1994
Neuman, C.H.:A Gravity Override Model of Steamdrive, Journal of Petroleum Technology, January 1985: 163-9
Palagi, C.L. and Aziz, K.: Modeling Vertical and Horizontal Well With Voronoi Grid, SPE 24072, SPE Western Regional Meeting held
in Bakersfield, CA, March30-April 1, 1994
Stars Manual CMG (2009, Computer Modeling Group)
Trebolle, R. L., Chalot, J.P., and R. Colmenaras.: The Orinoco Heavy-Oil Belt Pilot Projects and Development Strategy. Paper SPE 25798
presented at the 2nd International Thermal Operations Symposium, Bakersfield, CA, February 1993
World Energy Council, 2007 Survey of Energy Resources, http://www.worldenergy.org/documents/ser2007_final_online_version_1.pdf
Nomenclature
BT
breakthrough time
FsIns
instantaneous oil-steam-ratio
porosity
Ii
injectivity index
kro
krog
krw
Kv/Kh
anisotropy
Mo(1-250)
oil viscosity
Pave
PV
pore volume
Pwf
Qinj
16
SPE 146612
RF
recovery factor
So
oil saturation
Sw
water saturation
Tr
reservoir temperature
Ts