Cyclic Gas/Steam Stimulation

of Heavy-Oil Wells
Robert F. Meldau, SPE, Husky Oil Operations Ltd.
Robert G. Shipley, SPE, Husky Oil Co.
Keith H. Coats, SPE, Intercomp Resource Development and Engineering Inc.

Summary
Injecting air with steam nearly doubled oil gas/steam mixture before injection. These
production from three cyclic steam stimulations in a stimulations used much less steam and much more
California reservoir producing 11 °API (0.99-g/cm 3) gas than those reported here. Downhole steam
oil. Computer model studies reveal several reasons generators under development also inject gas/steam
for better oil recovery, including greater pressure mixtures. 7
drawdown, gas drive of heated oil near the well, and This paper briefly describes the Paris Valley
trapped residual gas deep in the reservoir. thermal iilot and gives field results for cyclic
steaming. Computer model results help explain the
Introduction improved performance noted when gas is added to
Cyclic steaming is a key tool to achieving economic the steam.
production from reservoirs containing very viscous
oil. Thus, there is an incentive for the development of Field Results
better cyclic steaming techniques since heavy oil The wet combustion process used at Paris Valley
reservoirs in the U.S., Canada, and Venezuela consists of forming a heat pad by dry combustion
contain a large amount of potential reserves. and then switching to simultaneous air/water in-
Adding gas to cyclic steam first was suggested by jection. A water/air ratio of 200 bbl/MMscf
Pursley! and Weinstein 2 in 1974. They found oil (1.12x 10- 3 m 3 /m 3 ) was chosen to move the heat
increases of 50070 or more from gas/ steam forward at about the same calculated rate as the
stimulation compared with steam alone in physical combustion front but maintain high combustion
and computer models of Cold Lake wells in Alberta temperatures. Cyclic steaming was used in the
and matched field results for steam stimulation with producing wells before heat arrival from in-situ
and without gas. Their work was the basis for trying combustion. The pilot was terminated after 4 years in
air/steam stimulation of Paris Valley wells in March 1979 because of high compressed air cost, low
California. oil production rate, and controlled oil prices.
The idea of cyclic gas injection to produce heavy Fig. 1 shows the pilot location. Figs. 2 through 4
oil wells is not new. Shelton 3 reports the use of and Table 1 describe the reservoir. The oil pay is
propane-rich natural gas with mixed results in the underlain by a tilted oil/water contact. Note that the
field, and Clark 4 used exhaust gas with some success. estimated tank oil viscosity of the Upper Lobe at a
Cyclic carbon dioxide has stimulated heavy oil wells 5 reservoir temperature of 87°F (30.6°C) is about
successfully and is similar to gas/steam stimulation 227,000 cp (or mPa·s) compared with 23,000 cp for
since both provide viscosity reduction, oil expansion, the Lower Lobe. Typical primary production for an
and reservoir energy. upstructure well is 5 BOPD (0.80 m 3 /d) and 40
The Vapor Therm™ process has been used to BWPD (6.36 m 3 /d). Downstructure wells produce
recover heavy oil from several midcontinent reser- water with only a trace of oil.
voirs by cyclic gas/steam stimulation at oil/steam Each pilot pattern is about 5 acres (20 234 m 2 ),
ratios of 1.6: 1 to 2.2: 1. 6 In this process, water is and the distance between adjacent upstructure wells
mixed with hot combustion gas to produce a is about 150 ft (45.7 m) as shown in Fig. 2. The in-
0149·2136/81/0010·8911 $00.25 jectors were drilled off center to compensate for
Copyright 1981 Society of Petroleum Engineers of AIME gravity effects on injected air.
1990 JOURNAL OF PETROLEUM TECHNOLOGY
 TABLE 1 - ROCK AND FLUID PROPERTIES

Depth, ft 800
Temperature, of 87

-----l Initial pressure, psig

Net thickness, ft
Permeability, md
Dip, degrees
220
50
3,750
15

I
Porosity, % 32

I
N
~ Water saturation, %
Gas saturation, %
Oil content, bbllacre·ft
36
0
1,590

l, I
Oil gravity, ° API
Solution GOR, scf/bbl
Water salinity, ppm
10.5
<10
280
'",,- Tank Oil Viscosity (cp)

~
~ 0 "",,-
Temperature (OF)
87
Upper Lobe
227,000
Lower Lobe
23,000
PARIS ~" 100 94,000 11,000
200 340 120
VALLEY - , ..p ""',
FIELD
Santa
600

•
I~U~~::~
Fig. 1 - Pilot location.

"llTOOLE

g LOWER
GO
L..._1..JI!l!=
T.O.822'

Fig. 3 - Type log - Paris Valley field.

LEGEND
• PILOT PRODUCER
~ STEAM INJECTOR
t TEMP. OBSERVATION WELL

o 500 1000

FEET

Fig. 2 - Structure - top of Ansberry sand. Fig. 4 - Net pay - all lobes.

OCTOBER 1981 1991

 WEll 3 Cyclic steaming accounted for more than 90070 of
1 the oil produced during the pilot. Fig. 5 shows that
most of this response occurred in upstructure wells
with no bottom water and more than 60 ft (18.3 m) of
net pay. Cumulative steam/oil ratio in these up-
structure wells averaged 3:1 and varied from 2:1 to
5.5:1. Well 3's Cycle 3 had the best response with a
steam/oil ratio of 0.93:1.
Air/Steam Injection
Compressed air was added to the steam because it
was readily, available from the combustion

I
•
operations. The risk of damage from wellbore fires
• . . . /1 \ was recognized and minimized by injecting steam
before and after air/steam injection. The procedure
@---/ was to (1) inject steam down the tubing for about 1

o AREA PROPORTIONAL TO·\

CUMUlA TlVE OIL PRODUCTION
week, (2) inject steam down the annulus and air
down the tubing for about 10 days, and (3) inject
steam only for the last 4 to 7 days. Steam was in-
jected down both tubing and annulus just before
(30 MBBlS FOR WELL 3) starting air. A typical downhole completion for
injection is shown in Fig. 6.
Fig. 5 - Cumulative oil production. Table 2 reports steam and air volumes for Well
20's Cycles 3 and 5 and Well 3's Cycle 7. The
air/steam ratios were based on Pursley's physical
model data, 1 which indicate a more favorable
response at gas/oil ratios (GOR's) above 50 scflbbl
(8.9 m 3 /m 3 ). The amount of air injected was small
compared with steam injected in volume or heat. For
example, with an air/steam ratio of 100 scflbbl (17.8
m 3 /m 3 ), only 2.3% of the total volume is air and
STEAM 430 BPD, AIR 40 MCFPD OCT. 23, 1978 combustion provides only 3.1 % of the total heat if
COMPLETION IES LOG TRACER LOG RESULTS (1) steam is 65% quality at 300 psia (2070 kPa) and
,. (2) air generates 100 Btu/scf (3.73 x 106 J/m 3 ).
Casing
'-
5112·
Liner
There was no evidence of well bore fires or severe
27/S" corrosion due to air/steam injection. There were no
Tubing
pumping problems in Well 20. Well 3 had some
problems with sanding during backflow but no more
than during the previous cycle with steam alone. No
unusual emulsions were observed. Wellhead pressure
during air injection increased about 100 psi (690 kPa)
in Well 20's Cycle 3, decreased 50 psi (345 kPa) in
6""l Well 20's Cycle 5, and increased about 200 psi (1380
TRACER STEAM INJECTED "'1FT.
kPa) in Well 3's Cycle 7. This latter increase was
more than expected from the relative volume of air.
Fig. 6 - Radioactive tracer profile during air/steam in- However, it is possible that some plugging resulted
jection. from debris when steam was switched from the
tubing to the annulus. Well 20's steam injectivity
improved 50% in subsequent Cycle 4 and 100% in
Cycle 5.
Toward the end of air/steam injection in Well 20's
TABLE2-AIRISTEAM INJECTION Cycle 5, radioactive injection profiles were run using
both gas- and water-soluble iodine-131 tracers. First,
Well 20 Well 3 the gas tracer (methyl iodide) was injected at the
Cycle 3 Cycle 5 Cycle 7 wellhead into air going down the tubing; second, the
Date started 7/77 10/78 10178 gas tracer was injected into steam going down the
Days injection 26 24 20
Steam annulus; and third, the water tracer (sodium iodide)
Volume, Mbbl 16.2 10.5 9.2 was injected into steam going down the annulus. Fig.
Wellhead quality, % 74 62 66 6 shows the results of logging after each tracer in-
Heat content, MMBtu 5,821 3,570 3,140 jection. The air tracer entered toward the bottom and
Average pressure, pSig 400 260 440
Air steam tracers entered toward the top of the open pay
Volume, MMscf 1.5 3.7 3.6 interval as expected from the geometry of the
Air/steam ratio, scf/bbl 91 350 390 completion. The gas and water tracers in steam gave
roughly the same distribution. The injection tem-
1992 JOURNAL OF PETROLEUM TECHNOLOGY
 20

-
OIL CUT

. .: 40 e.o
"
(.)
.... 6O '0.
, I
, 0 FLUID LEVEL
FLUID LEVEL c::?'\ 8O J7 't,. ,.0
~~'"
XXXX

100
"" ", ,'"

.
0

.,"0
..".,
0

~
«
W
1-
II:
«
II:
.... ....
0 0

Fig. 7 - Response to cyclic steam - Well 20. Fig. 8 - Response to cyclic steam - Well 3.

TABLE 3 - RESPONSE TO CYCLIC AIR/STEAM

Well 20 Well 3
Cycle 2 Cycle 3 Cycle 4 Cycle 5 Cycle 6 Cycle 7
Steam volume, Mbbl 13.2 16.2 15.7 10.4 8.2 9.2
Air volume, MMscf 0 1.5 0 3.7 0 3.6
Air/steam ratio, scf/bbl 0 91 0 355 0 394
Comparable producing days 161 161 90· 90 97·· 97
Oil produced, bbl 2,449 4,701 270 503 2,375 4,203
Steam/oil ratio, bbl/bbl 5.4 3.4 58 21 3.5 2.2
Oil/steam ratio, bbl/bbl 0.19 0.29 0.02 0.05 0.29 0.45
Peak production test, BOPO 51 81 24 38 60 141
'Total cycle: 534 bbl in 142 days for 30:1 steam/oil ratio.
"Total cycle: 4,843 bbl in 270 days for 1.7:1 steam/oil ratio.

perature of 400°F (204°C) was well below methyl combustion gas production from the fireflood. The
iodide's decomposition temperature of 518°F steam/ oil ratio of Cycle 3 was 3.4 compared with 5.4
(270°C). for the previous cycle and with 30 for the subsequent
Cycle 4 at comparable times.
Oil Response Fig. 1 shows that the oil rate is abnormally low and
The response to air/steam stimulation is compared that the water cut is very high in both Cycles 4 and 5
with the previous stimulation with steam only in of Well 20. We have not found the reason for this
Table 3. Production times varied widely. However, low productivity, particularly in view of the
at comparable times, air/steam stimulation produced favorable injection profile seen in Fig. 6. However,
almost twice as much oil as the previous cycle with the results are not meaningful in evaluating
steam alone: gas/steam stimulation.
Well 3. The oil response in Well 3's Cycle 7 was
Increased Oil encouraging, when all facts are considered, but not
Well Cycle (bbl) (m 3 ) (0/0) conclusive. Well 3 was stimulated with air and steam
-- in late Oct. 1978 near the end of the project. The
20 3 2,252 358 92
gravel pack completion was due for replacement
20 5 233 37 86
after six steam jobs, but this work was not justified
3 7 1,828 291 77
with pilot shutdown imminent. Thus, we experienced
pump problems and failed to keep the well pumped
In addition, there was a substantial reduction in off, as seen in Fig. 8. Even so, Cycle 7 at comparable
steam/oil ratio (or increase in oil/steam ratio). producing times has a lower cumulative steam/oil
Well 20. Cycle 3 of Well 20 in 1977 gave the best ratio than Cycle 5 or Cycle 6 in spite of the higher
response (Fig. 7) and had the most reliable data. Oil fluid level. There was combustion gas production
and water production declined in a normal manner during Cycle 7 and the previous cycle, though
over a 6-month period. There was practically no probably from only part of the pay.
OCTOBER 1981 1993
 TABLE 4 - RADIAL MODEL PROPERTIES
LAYERS
Pore volume, MMbbl 2.3
Gross thickness, ft 96

1 Model area, acres

Number of layers
Blocks in radial direction
10
7
9
Block 1 center radius, ft 4.6
Dip, degrees none
"L Reservoir temperature, of 87
Rock compressibility, 1Q-6 ps i- 1 400
Net/gross thickness 0.9
Average initial pressure, psi 262
Initial solution GOR, scf/bbl o
Water/oil capillary pressure yes
Gas/oil capillary pressure no
Thermal conductivity overburden

WELLBORE OF CYCLIC
STEAM STIMULATION WEL~ ,2
3
I
BLOCKS
and underburden, Btu/ft·D·oF
Tank oil gravity, °API
24
10.5

Tank Oil Viscosity (cp)

Temperature rF) Layers 1 Through 3 Layers 4 Through 7
87 100,000 10,000
Fig. 9 - Cutaway view radial two·dimensional model. 200 330 110
400 8.0 7.6

TABLE 5 - DESCRIPTION OF LAYERS

Air Vertical Initial

Thickness Porosity Rock Permeability Permeability' Water
Layer (ft) (%) Type (md) (% horizontal) ~
1 12 32 1 2,500 0 32
2 12 32 1 2,500 100 32
3 24 31 2 .500 2 58
4 12 30 1 2,500 2 32
5 12 28 1 4,000 2 32
6 12 28 1 4,000 100 32
7 12 27 1 2,500 2 32
'Vertlcal (Z) transmissibility factor with cell above.

TABLE 6 - DESCRIPTION OF BLOCKS

Block Number

Radius, ft
Well
- -1 2 3 4 5 6 7 8 9
0.5 6.5 11.0 18.5 31 52 88 148 250 381
Volume, % 0 0.03 0.05 0.15 0.43 1.2 3.5 10.0 27.6 57.1

Paris Valley Computer Model Radial Model Well 4

The performance of gas/steam stimulation was A two-dimensional radial model of Paris Valley Well
compared with steam alone in computer model runs 4 (like that shown in Fig. 9) was used to study
of a typical Paris Valley well. The purpose was to gas/steam injection. Model properties are given in
gain a better understanding of reservoir mechanisms Tables 4 through 7. The 96-ft (29.3-m), to-acre (40
and support the observed field behavior discussed 468-m2 ) model contains seven layers and nine radial
earlier. blocks. Vertical flow is restricted at 2070 of horizontal
This study used the numerical simulation model between most layers. Reservoir oil viscosity of the
developed for in-situ combustion by Coats 9 modified upper three layers is about 100,000 cp (or mPa·s)
to handle two oil viscosities. The steam modellO was compared with to,ooo cp for the lower layers. This
tried first but was unstable when producing at high agrees with field data for tank oil viscosity if con-
GOR's. The combustion reaction was not used since sideration is given to the small amount of solution
this would increase computer costs and complicate gas probably present. Hysteresis was not included in
interpretation of the results. The amount of heat the relative permeability functions of Table 7.
added by combustion would be small as noted Model properties were derived from a combination
earlier. of field, laboratory, and literature data. Parameters
1994 JOURNAL OF PETROLEUM TECHNOLOGY
 TABLE 7 - SATURATION AND RELATIVE PERMEABILITY

Intercomp Paris Valley

Symbol Type 1 Type 2-
Saturation, %
Connate water (= SWI R) SWC 25 33
Residual oil to water SORW 20
Residual oil to gas SORG 33 23
Critical gas SGC 5
Residual gas SGR 15
Change in saturation with temperature, %/100°F
Connate water SCWTP 5.0 6.7
Residual oil to water SORWTP -3.3
Residual oil to gas SORGTP -6.7 -3.3
Relative permeability, %
Water at residual oil KRWRO 10
Oil at connate water KROCW 100
Gas at residual oil KRGRO 60
Change in KRWRO with temperature, %/100°F KRWTP -1.7
Exponent for relative permeability curve--
Water NW 3.0 5.0
Oil with water NOW 1.1 1.1
Oil with gas NOG 1.5 1.5
Gas NG 1.5 1.5
'If different from Type 1.
"Typical normalized equation for relative permeability vs. water saturation:
KRW/KRWRO = [(SW - SWC)/(l - SORW - SWC)1 NW •

TABLE 8 - PARIS VALLEY WELL 4 into Layers 5 through 7 in accordance with steam
CYCLIC STEAM INJECTION injection profile results for tubing hung at the
bottom of the sand. All intervals were open to
Cycle
production, and observed fluid levels were used to
1 2 3
- estimate bottomhole pressure during backflow.
Date 9/75 7/76 4/77 History matching Well 4 observed production was
Injection time, days 14 36 37
Steam volume, Mbbl 4.4 7.6 13.2 not successful as seen in Table 9 because of
Estimated sandface quality, % 0- 37 68 limitations in field data and outside reservoir in-
Heat injected, MMBtu 580 1,850 4,320 fluences during later cycles. Oil production could not
Average wellhead pressure, psig 590 510 350 ~e matched for two reasons. First, field data, par-
'Hot water temperature = 449'F.
tIcularly water cut, proved unreliable upon close
study. Second, cyclic steam injection in nearby wells
affected Well 4 production and bottomhole
that affect the results and are most uncertain include
pressures. High water production rates in Cycles 2
(1) initial water saturation, (2) vertical permeability
between layers, (3) oil viscosity of the middle layers, and 3 also could not be matched in this radial model.
(4) rock compressibility, and (5) water relative Detailed study of well and reservoir behavior in-
permeability vs. temperature. dicates the presence of a tilted water table 50 to 100ft
Well 4 was chosen for model studies of cyclic (30.5 m) from Well 4 that contributed to water
steam because primary production rates were known production. Bottom water and interference effects of
cyclic steam response seemed typical of upstructur; nearby wells could be handled with a three-
wells, and injection profile logs were available to dimensional model, but we decided that comparative
estimate steam injection intervals. Initial water runs with a simple two-dimensional radial model
saturation was estimated from core data, and relative would be adequate to study the mechanisms of cyclic
permeability was adjusted to match the observed gas/steam stimulation.
primary rates of 4 BOPD (0.64 m 3 /d oil) and 43
BWPD (6.84 m 3 /d water). Table 8 gives the cyclic Cyclic Gas/Steam Performance
steam injection conditions. Steam was injected only Table 9 also summarizes the response to cyclic steam

TABLE 9 - SUMMARY OF WELL 4 CYCLIC STEAM RESPONSE

Intercomp Gas/Steam Oil Production Water Production

Source Model (scf/bbl) for Cycle (bbl) for Cycle (Mbbl)
Field 0 2 3 1 2 3
-
Run4 steam 0 2,064 6,295 9,020 14.1 33.4 51.0
Run4A combustion 0 3,155 2,868 6,780 13.4 8.1 20.4
Run15A combustion 80 3,015 2,915 6,506 12.5 8.0 19.9
Run15B combustion 320 3,502 7,435 6.4 17.0
3,879 8,005 7.5 17.4

OCTOBER 1981 1995

 30r-----,----~----~---- in Well 4 for (1) field data, (2) the steam computer
20
EXTRA OIL FROM GAS WITH STEAM (320 SCF/BBl) model with no gas, and (3) the combustion computer
model with 0, 80, and 320 scf of gas per barrel of
steam (0, 14.2, and 57 m 3 gas/m3 steam). The gas
and steam were injected simultaneously. Note that
the steam and combustion models gave almost
identical results. Adding 80 scflbbl (14.2 m 3 ) of gas'
in Cycles 2 and 3 resulted in 16070 more oil
production, and adding 320 scflbbl (57 m 3 ) resulted
00 in 26070 more oil. Fig. 10 shows the monthly oil and
C
water rates predicted for the base case with no gas
....
<> and the extra oil predicted for adding 320 scf of gas
per barrel (57 m 3 1m 3 ).
...ui 40

'"ca: As expected, the bottomhole injection pressure at

the end of Cycle 2 injection increased from 518 psia
......5
20
(3572 kPa) for steam alone to 638 psi (4399 kPa) and
682 psi a (4702 kPa) for the two cases with gas added.
This resulted from increased pressure in the reservoir
rather than the small increase in injection rate. Cycle
3 pressure increase was similar.
1875 18711 1877 1878 Maps comparing temperature, pressure, gas
saturation, and oil saturation from the computer
Fig. 10 - Predicted cyclic steam production - Well 4, radial model output are summarized in Table 10.
model. Examination showed that the results could be

TABLE 10 - COMPUTER MAPS OF WELL 4 CYCLIC STEAM, CYCLE 3

Radial Blocks
Run· After Layers·· 1 Through 4 5 6 7 and8 9
Temperature, • F
15B injection A 14 3 0 0 0
B 100 14 1 0 0
C 400 253 18 0 0
D 403 96 7 0 0
Pressure, psia
15B injection A 298 296 293 252 258
B 471 353 330 300 251
C 644 644 640 390 338
D 665 664 649 409 342
Pressure Increase Due to Gas Injection, psi
15B-4A injection A 98 86 70 5 0
B 200 106 92 54 10
C 131 133 150 67 86
D 109 107 97 83 118
15B-4A production A 29 45 47 22 5
B 23 38 32
C 10 11 24 80 99
D 6 5 28 80 92
Gas Saturations, %
4A injection A 0 0 0 0 0
B 0 0 0 0 0
C 20 0 0 0 0
D 20 0 0 0 0
15B injection A 7 4 3 0 0
B 5 5 8 5 0
C 25 8 5 6 4
D 18 5 5 5 1
15B production A 5 5 2 0 0
B 5 5 5 3 0
C 12 5 5 5 4
D 8 5 5 5 1
Oil Saturation Decrease Due to Gas Injection, %
4A-15B production A 3 2 0 0 0
B 7 0 0 0 0
C 7 1 3 3 2
D 1 0 1 3 1
"Run 4A - no gas; Run 158 - 320 sci gaslbbl steam injected.
""A= 1 and 2, 8=3, C=4 and 5, 0=6 and 7 ofTable5.

1996 JOURNAL OF PETROLEUM TECHNOLOGY

combined into four layers of equal thickness and five TABLE 11 - EFFECT OF GAS ON CYCLIC STEAM OIL 2
radial blocks with no loss of definition. At the end of
Cycle 3 steam injection, steam has moved out about Oil Extra
Produced Oil
50 ft (15.2 m) to Block 5 in the lower layers receiving Case Fluids Injected (bbl) (bbl)
injection and has moved up to Layer 3 near the well.
steam·only 45 Mbbl steam 8,300
Gas/steam injection gave somewhat higher tem- gas first 4.99 MMscf gas 10,000 1,700
peratures than steam alone in the middle layers near 45 Mbbl steam
the well and lower temperatures in the bottom layer. gas last 45 Mbbl steam 14,000 5,700
Pressures were substantially higher after gas/steam 4.99 MMscf gas
injection. This increase persisted deep into the model
even at the end of the production cycle.
The level of gas saturation throughout the model the upper layers had contained oil of lower viscosity,
from gas/steam injection is striking both at the end but it would have been of little benefit in wells where
of injection and at the end of production. Some gas most of the steam already goes into the top of the
moved out 300 ft (91 m) to the last radial block in the sand.
lower layers. Also, gas moved into upper layers even Gas/steam performance will vary greatly
though vertical permeability was only 2% of depending on reservoir properties and individual well
horizontal. Gas saturations after production are, in conditions. In some cases, gravity effects coupled
most cases, at the critical gas saturation of 5"70. with a dipping structure or local heterogeneities will
trap gas and displace oil. The method of gas injection
Discussion also can affect the results.
The field results, particularly those of Well 20's Weinstein 2 found in cyclic steam model studies of
Cycle 3, were encouraging but do not explain why gas Cold Lake wells in Alberta that oil production in-
with steam may perform better than steam alone. creased by 40% if the gas was injected after steam
The computer model results, however, suggest rather than before steam in the first cycle. He suc-
several mechanisms that assist oil recovery under cessfully matched field data with and without gas
Paris Valley reservoir conditions. They include (1) injection using natural gas and reservoir oil with 42
trapping gas at saturations up to the critical, (2) scflbbl (7.5 m 3 /m 3 ) of solution gas. His results for
increased gas drive of heated oil near the wellbore, injecting (1) steam alone, (2) natural gas then steam,
(3) movement of heat into upper, more viscous oil and (3) steam followed by gas are reproduced in
sands, and (4) greater drawdown due to higher Table 11. It is not easy to explain why the response is
reservoir pressures. so much better when gas is injected after the steam.
A major effect when gas is injected with steam Both schemes provide gas beyond the heated zone.
occurs out in the reservoir past the heated zone. Part The difference may be due partly to the gas pushing
of the gas that flows into this region is trapped at heat farther out into the reservoir and partly to three-
saturations up to the critical during back flow . This phase relative permeability functions in the model.
trapped gas, in turn, displaces oil, which is the other Farouq Ali * also modeled cyclic steaming of Cold
nonwetting phase. The effect can be seen Lake wells with and without added gas under
qualitatively in the data set at the bottom of Table somewhat different reservoir conditions. Solution
10, which compares oil saturations at the end of the gas was about 35 scflbbl (6.23 m 3 /m 3 ), initial gas
6-month production period. Little or no trapped gas saturation 0.1%, and critical gas saturation 5%.
effect would be expected if an initial gas cap or initial Thirty thousand barrels (4770 m 3 ) of steam and 2
mobile gas saturation exists in the reservoir. Also, the MMscf (56.6 x 10 3 m 3 ) of gas were injected each
amount of oil displaced by trapped gas in the model cycle for six cycles. The oil/steam ratio ,improved
obviously will depend on the estimate chosen for about 20% from 0.25:1 to about 0.30:1 with gas
critical gas saturation. Thus, this mechanism added to the steam. However, there was only a small
probably will be important only in liquid-filled advantage to injecting gas last rather than first.
reservoirs. The Paris Valley well response and computer
During back flow , most of the gas beyond the model results suggest that adding gas to cyclic steam
heated zone flows back to the well and is produced. should be considered for viscous, liquid-filled
In the process, it drives the heated oil to the wellbore, reservoirs. However, more model studies and more
thereby increasing recovery. In some reservoirs, gas field tests clearly are needed to guide future ap-
drive may occur naturally due to an outside gas plications.
source. Low-GOR wells, 'however, are good can- The choice of gas will depend mostly on
didates for gas/steam stimulation. availability. Natural gas is safest and can be
Gas also pressures up the reservoir. During recovered on backflow. Compressed air must be used
backflow, the resulting increase in drawdown in- with caution to avoid well damage but in some cases
creases oil rates, as seen in Fig. 10. (like Paris Valley) is readily available. Flue gas has
Another effect of the gas was to move heat into been used with steam, but corrosion problems must
upper layers containing more viscous oil. This oc- be considered. Carbon dioxide gave poorer per-
curred in spite of the 2% vertical permeability formance than air or methane in Pursley's laboratory
restriction. Oil recovery in the Paris Valley model
improved because steam injection was into the lower 'Farouq Ali, S.M.: private communication, U. of Alberta, Edmonton, Canada
layers. The effect probably would be even greater if (1979).

OCTOBER 1981 1997

scaled model tests at 20 scf of gas per barrel of steam 2. Weinstein, H.G.: "Mathematical Models for Thermal
(3.6 m 3 gas/m3 steam) but could give better per- Recovery Processes," paper presented at Heavy Oil Sym-
formance than air or methane in larger amounts. posium, U. ofZulia, Maracaibo, Venezuela, July 1974.
3. Shelton, J.L. and Morris, E.E.: "Cyclic Injection of Rich Gas
Into Producing Wells to Increase Rates From Viscous-Oil
Conclusions Reservoirs," J. Pet. Tech. (March 1972) 890-896.
I. Air I steam stimulation produced almost twice as 4. Clark, N.J., Roberts, T.G., and Lindner, J.O.: "Engine
much oil in a comparable time as the previous cycle Exhaust Gas Boosts Heavy Oil Recovery," Pet. Eng. (Aug.
1964) 43-47.
in limited Paris Valley field tests. 5. Reid, T.B. and Robinson, H.J.: "Lick Creek Meakin Sand
2. Air can be injected with steam into an oil well Unit Immiscible C0 2 /Waterflood Project," J. Pet. Tech.
without damage if suitable precautions are followed (Sept. 1981).
carefully. 6. Sperry, J.S.: "Heavy-Oil Recovery System Completes Three
3. Model studies indicate adding gas to steam in Field Tests in Mid-Continent Region," Oil and Gas J. (July
27,1981) 225-237.
Paris Valley wells increases oil production as a result 7. Fox, R.L., Donaldson, A.B., and Mulac, A.J.: "Develop-
of (1) trapping residual gas, (2) gas drive of hot oil ment of Technology for Downhole Steam Production," paper
near the well, (3) better flow of heat to upper oil SPE 9776 presented at the SPE 1981 California Regional
sands, and (4) higher drawdown from increased Meeting, Bakersfield, March 25-26.
8. Shipley, R.G., Meldau, R.F., and White, P.D.: "Paris Valley
reservoir pressure. Combustion Thermal Drive Pilot Demonstration Test - Final
4. The effect of adding gas to steam will vary Report," DOE/SAN/l000-3, U.S. DOE, Oakland, CA (Sept.
greatly depending on reservoir and well conditions. 1980).
5. Additional model studies and field tests are 9. Coats, K.H.: "In-Situ Combustion Model," Soc. Pet. Eng. J.
(Dec. 1980) 533-554.
needed to guide future application. 10. Coats, K.H.: "A Highly Implicit Steamflood Model," Soc.
6. Adding gas to cyclic steam should be considered Pet. Eng. J. (Oct. 1978) 369-383.
for viscous oil reservoirs, particularly if they are
liquid-filled and underpressured. SI Metric Conversion Factors
7. Water-soluble iodine-131 tracer gave almost the acre x 4.046 873 E+03
same steam injection profile as gaseous iodine-131 acre-ft x 1.223 482 E + 03
when injecting 430 BID (68.4 m 3 /d) of 70% quality °API 141.51C API + 131.5)
steam down the annulus at a temperature of 400°F bbl x 1.589 873 E - 01
(204°C). Btu x 1.055 056 E + 03
cp x 1.0* E-03 Pa·s
Acknowledgments cu ft x 2.831 685 E-02 m3
We thank Husky Oil and the u.s. DOE for per- OF CF - 32)/1.8
mission to present this paper. This study was done as ft x 3.048* E-OI
part of the work under cost-shared Contract EY-76- in. X 2.54* E-OI mm
C-03-1000. psi x 6.894 757 E+OO kPa
psi -1 x 1.450 377 E-OI kPa -1
scf x 2.863 640 E - 02 stdm 3
References ·Conversion factor is exact. JPT
1. Pursley, S.A.: "Experimental Study of Thermal Recovery Original manuscript received in Society of Petroleum Engineers office March
10, 1980. Paper accepted for publication May 13, 1981. Revised manuscript
Processes," paper presented at Heavy Oil Symposium, U. of received Aug. 17, 1981. Paper (SPE 8911) first presented at the SPE 1980
Zulia, Maracaibo, Venezuela, July 1974. California Regional Meeting, held in Los Angeles, April 9·11.

1998 JOURNAL OF PETROLEUM TECHNOLOGY