Heavy oil recovery by inert gas injection with horizontal wells

Heavy oil recovery by inert gas injection with horizontal wells

ELSEVIER PETROLEUM SCIENCE & ENGINEERING Journal of Petroleum Science and Engineenng 11 (1994) 213-226 Heavy oil recovery by inert gas injection wit...

753KB Sizes 3 Downloads 84 Views


PETROLEUM SCIENCE & ENGINEERING Journal of Petroleum Science and Engineenng 11 (1994) 213-226

Heavy oil recovery by inert gas injection with horizontal wells M . R . Islam a,*, A. C h a k m a a,**, K.N. Jha b aDepartment of Chermcal and Petroleum Engmeermg The Um verslly of Calgar)?, CalgarE Alberta, Canada bCANMET Energy Research Laboratories Energy, Mines and Resources Canada, Ottawa, Ontarzo, Canada (Received January 1, 1994, revised version accepted March 2, 1994)

Abstract Gas mjection is one of the oldest enhanced oil recovery techmques used for the recovery of light oils. Recently, there has been a growing interest m this techmque for both hght and heavy od reservoirs The apphcatton of gas injection to horizontal wells for the recovery o f heavy oils from shallow reservoirs has been mvestlgated using a 2D physical model Oil recovery usmg gravity-stablhzatlon may be as high as 70% of the oil-m-place for moderately vtscous heavy oil reservotrs. The ~mportance o f m a m t a m m g a stable displacement front was demonstrated through several displacement tests under unstable as well as pseudostable displacement regimes. Unstable or pseudostable flow regimes are hkely to prevad m the situation where gravity stabdlzatlon does not take place.

1. Introduction

Injection of natural gas into an oil reservoir was probably the first process suggested for improved oil recovery. Gas injection for improved oil recovery has been practiced prior to 1900 (Craft and Hawkins, 1959). In the earlier days, gas injection was primarily intended for pressure maintenance, thus initial productivity was increased due to increased reservoir pressure. This type of gas injection was not intended to increase the ultimate recovery. However, recently gas injection has been used as a tool to increase the ultimate recovery of oil. In this type of application, the injected gas acts not only as a pressure maintenance agent but also displaces the oil from the formation. The displacement mechanism and, *Present Address Department of Geological Engineering South Dakota School of Mines Rapid City, S D_, USA **Corresponding author.

more importantly, the sweep efficiency of the injected gas, determine the success of gas rejection-based enhanced oil recovery (EOR) processes. Displacement of oil by gas injection is now considered to be an economically viable method for the improved recovery of light oils even with oil prices as low as $18 per barrel (Fayers, 1991 ). The effectiveness of any gas displacement process also depends on its displacement efficiency. With gas recycling, displacement efficiency of 5 MSCF/bbl is also considered to be economical when the gas price is in the range of $1-2/MSCF. For high permeability reservoirs or where gravity segregation IS possible, gas injection may become more attractive. In the absence of high vertical permeability or the possibihty of gravity segregation, gas lnjectxon can still be attractive if the reservoir is relatively thin. This type of gas injection is based on a frontal drive mechanism which is similar to water injection and is commonly referred to as "dispersed gas rejection".

0920-4105/94/$07 00 © 1994 Elsevier Science B V All rights reserved SSDI 0920-4105 ( 94 ) 00015-V


M R Islam et aL /Journal of Petroleum Sctence and Engmeermg 11 (I 994) 213-226

There are a number of heavy oil reservoirs in regions bordering Alberta and Saskatchewan, where the reservoir thickness is relatively low and in most cases less than l0 m. The main geometric characteristics of these reservoirs are that they are relatively thin but cover a wider reservoir area. In such cases, conventional heavy oil recovery methods such as steam injection are usually economically less attractive due to excessive heat losses through the overburden. Inasmuch as gas injection, In general, is supposed to work better when the reservoir is thin, it should also be attractive for the shallow reservoirs in question. Horizontal wells provide excellent means of covenng a wider reservoir area than vertical wells. Therefore, the use of gas injection with horizon-

Nomenclature A,nj a b g H h


ko,B kgor khj L M P, S, t I4," W,

area open to flow through wells scaling factor thickness o f the porous m e d i u m acceleration due to gravity height or thickness o f field height or thickness o f the model initial oil-in-place effective permeablhty to the phase i effective permeability to oll at residual gas saturation effective permeability to gas at residual oll saturation thermal conductivity o f j length or well spacing end point m o b i h t y raUo pressure in the phase t saturation in the phase t time width o f the reservoir or model mjectton rate for

Greek letters 0 p, g,

porosity density o f the phase viscosity o f the phase t

Subscrtpts c g o R r w

cap rock gas phase oll phase reference quantity reservoir rock aqueous phase

tal wells should provide a means for economical recovery of heavy oils from shallow reservoirs.

2. Literature review

Gravity assisted inert gas injection is one of the oldest recovery techniques used in the enhanced oil recovery. Leverett (1941) was the first researcher to point out the importance of gravitational and capillary forces in an immiscible displacement process. Katz (1942) reported experimental evidence of the importance of gravitational forces on immiscible displacement of oil. Ever since these earher pioneering works, there have been numerous studies undertaken in the study of oil recovery due to gravity assisted gas injection (Stahl et al., 1943; Lewis, 1944; Terwilliger et al., 1951; Higgins, 1953). Traditionally gas injection has been used for light oil reservoirs. Only recently, gas injection has been considered for heavy oil recovery (Garcla, 1983; Mayer et al., 1986; Chung et al., 1988 ). Mayer et al. (1986) conducted a series of post waterflood immiscible CO2 injection corefloods with Wilmington heavy oil of 480 mPa s viscosity. They found recovery to increase with CO2 injection. Nectoux (1987) studied the effect of gravitational and viscous forces as well as that of composition on oil recovery by gravity drainage. He found gas displacement fronts to be stable at low flow rates due to gravitational forces. He conducted experiments by matching velocities of the field with that obtained in the laboratory. For an unfavorable mobility ratio, this may lead to optimistic results (Bentsen, 1985). Therefore, Nectoux's experimental results may not be suitable for extrapolation to field predictions. Nectoux also attributed the lowering of oil viscosity to the interactions between the injected gas and the oil-in-place. This implies that gas composition affects recovery of oil. He also reported that a high rate of gravity drainage was not found to be as effective as a low rate. The interchange of components between the injected gas and the displaced oil is well known. Near the injection point, the gas extracts components from the oil,

M R Islam et al / Journal of Petroleum Science and Engmeerlng 11 (I 994) 213-226

whereas at locations far from the injection point gas may give off lighter components. In general, these compositional effects are considered to be beneficial for oil recovery (Ypma, 1985 ). Chung et al. (1988)investigated the application of gravity assisted immiscible CO2 injection for the recovery of heavy oil using Berea sandstone cores. Recovery of the IOIP was found to be at 60% compared with 44% obtainable from conventional waterflood. Kantzas et al. (1988a,b) reported unusually high recoveries of over 95% IOIP using gravity stable inert gas injection. Very high recovery ofoil by inert gas injection (over 95% of the oil-in-place) was reported by Kantzas et al. (1988a). The authors did not provide adequate explanation to justify such high recoveries. In addition, they did not attempt to interpret the results for field application. Islam et al. (1992) reported a series of coreflood results using reconstituted in situ combustion gases. Their stability analysis indicated that for a horizontal coreflood with very high mobility ratio, the flow regime was likely to be unstable. Danesh et al. (1989) presented an analysis based on capillary and Bond number. The latter is equivalent to the instability number of Bentsen (1985). They observed that for a Bond number greater than 10-5 the residual non-wetting phase is affected by gravitational forces. They argued that field displacement is more of a capillary controlled process. Several successful field cases of immiscible gas injection has been reported in recent years. Garcia ( 1983 ) reported one such case in which 14% additional oil was recovered by gas injection. Spivak et al. ( 1988 ) reported a WAG process in Wilmington reservoir of California. Their laboratory test indicated 22% recovery with CO2 alone and 25.1% with a mixture containing 85% CO2 and 15% N2. Babson (1989) provides a review of the gas injection projects in California. An example of successful gas-assisted gravity drainage in the Hawkins field of Texas has been reported by Carlson ( 1988 ). He pointed out that 80% of the oil is expected to be produced.


3. Scaled model

Scaled models are important in order to be able to interpret laboratory data for field applications. Unfortunately, due to the complexity of the reservoir-fluid system, not all the appropriate groups can be scaled properly. Lozada and Farouq All (1989) provide six different sets of scaling criteria for immiscible gas injection studies. A 2D scaled model based on their work has been used in this work. 3 1 . 2 D or low pressure model

Figure 1 shows a schematic of the low pressure 2D model. It essentially consisted of a rectangular box with dimensions of 30 × 30 × 0.65 cm. The model has a thick glass front and plexiglass back. The model was equipped with two horizontal wells; one placed at the top and the other one at the bottom. The top horizontal well was used for gas injection, whereas the bottom horizontal well was used as the production well. The 2D model has been used for low pressure runs and hence may also be called the "low pressure model". The design of the 2D has been based on Approach #5 of Lozada and Farouq Ali ( 1989 ). Table 1 provides a list of scaling groups satisfied in this design. The key feature of this model is that it scales the gravitational forces properly. This is because the ratio of the viscous forces as well as the ratio of the viscous to gravi'IX]



. . . . . . . . . . . . :_-=_-z_............












Fig 1 Schematac of the low pressure 2D model


M R Islam et al / Journal o f Petroleum Science and Engtneermg 11 (1994) 213-226

Table 1 Scaling groups for the low pressure 2D model Same flmds, different porous medium and different pressure drops

W H 0RSoafloRL2 PoRg!azR Poa //oRk~a L ' L" ta koa Poa ' Poa 'Pwa'/~ako~'

S,~ Soa Po,



¢ha, S~oR'Swa' Poa' P,,,akwaPwaL' L 2

tational forces are taken into account. In order to maintain geometric similarity, which is important for systems with horizontal wells, the criterion of having the same pressure drop in the model as in the field is relaxed in this model. At low pressure conditions, gravitational forces are important and the 2D model represents gravity stabilized gas flow situations well. Consequently, the low pressure 2D model was used for the study of gravity-stable gas injection runs. It is also suited for studying constant injection pressure cases. This model is essentially based on the following scaling group ( 1981 ):

gob Note that this group is a combination of several dimensionless groups listed in Table 1. This group has to be the same in the model and in the prototype. Therefore, if the sand-fluid system to be modeled in the laboratory and in the field were to be the same, in order to keep the scaling group unchanged one has to increase the permeability of the medium by the same factor which is used to scale down the reservoir height. For example for the case where Hnetd is l 0 m and the model thickness hmodel is 30.5 cm, if the reservoir permeability is 2 D, the laboratory model permeability will have to be about 66 D. In order to achieve this desired higher permeability in the laboratory model coarse glass beads were used. An infinite conductivity horizontal well is assumed for the low pressure model. Time scaling follows the following scaling group:

t(kg) h( ~)ASolloh ) ° 5

4. Experimental set-up The experimental set-up essentially consisted of the 2D scaled model and other peripheral devices and instrumentation. The 2D model was scaled to represent the Aberfeldy field in the Lloydminster area. The top horizontal well was used for gas injection, whereas the bottom well was used as the production well (see Fig. l ). The model was packed with glass beads or unconsolidated sand and was saturated with water by previously pulling vacuum in the system. During packing and oll saturation periods, effluent was collected through the bottom well placed underneath the glass bead pack. Oll flood was carried out through the top horizontal well and effluent was collected until an irreducible water saturation was established. At this point the cell was ready for a complete gas injection run. The bottom well was closed and the central outlet was opened. This injection production configuration is equivalent to a systematic vertical cross section with a gas displacement front moving areally. The wells were fabricated from stainless steel with slots. In order to prevent local plugging, both injection and production wells were wrapped with screens. Each run was carried out by repacklng and resaturating the core. Gas injection was carried out at a constant injection pressure. The gas volume was measured with a wet gas meter and the liquid volume was measured by collecting samples in a fraction collector.

5. Results and discussion A series of model studies was conducted using different crude oils. The permeability requirement of the scaled model led to the use of coarse glass beads of approximately 66 D permeability. Two different crude oils having viscosities of

M R Islam et al / Journal of Petroleum Sctence and Engmeermg 11 (1994) 213-226

1000 mPa s and 4000 mPa s, respectively, were used. Table 2 provides a summary of these scaled model runs. Visual observation of the gas displacement front was made for each case. The stability for each case, as observed visually, is marked in Table 2 as 'stable' or 'unstable'. In order to verify the stability, a stability analysis was made following Sharma and Bentsen (1987). They define instability number, L as: i_ll~v(M - 1-N~) kgortre



(M+ 1 ) ( M 1 / 3 +

4hEb 2 1

)2 h 2 + b


(1) where, M is the end point mobility ratio,/~ is the gas viscosity, kgor is the effective permeability to gas at residual oil saturation, h and b are the width and the thickness of the porous medium, respectively, v is the velocity of the displacement front, and tre is the effective pseudo-interfacial tension between oil and gas. Also, Ng in Eq. 1 is given by:

N~ - ApgkgorCOS ol #gv



where Ap is the density difference, g is the acceleration due to gravity, and tx is the angle of inclination from the vertical plane ( a = 90 ° means horizontal system). These parameters were determined independently and are given for both CO2 and N 2 c a s e s in Table 3. The instabihty numbers, as calculated from Eq. 1, are also reported in Table 2 for various cases. They were calculated under the assumption that the pressure gradient is small enough to justify a negligible gas compressibility. Instability numbers were calculated for the scaled model and the field. The instability number calculations require the knowledge of injection rates. For this, the highest rate attained in a process was used because this is the flow rate for which viscous fingering is most likely to occur. Table 2 also lists the breakthrough recovery for each of the runs. There appear to be three different ranges of breakthrough recoveries. The high-

Table 2 Summary of the charactenstlcs of the experimental results Run no

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Gas type

N2 N2 N2 N2 N2 N2 CO2 CO2 CO 2 CO2 CO2 CO2 N2 N2 N2 N2 CO2 CO2 CO2 COz CO2

Injection pressure

Oil viscosity (mPa s)

Instablhty number



Recovery at breakthrough (%IOIP)

0 1 5 10 15 15 0 1 5 10 15 15 1 5 10 10 1 5 10 10 5

67 65 51 49 7 41 71 70 45 42 11 38 42 36 6 22 46 39 7 21 38

1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000 4000 400O 400O 4000 4000 4000 4000 4000 4000

-14 12 9 21 5 32_7 32_7 -114 10_1 21 2 32_4 32.4 1.8 12_6 26 26 16 12 3 25 7 25.7 12.3

Stable Stable Stable Stable Unstable Stabilized Stable Stable Stable Stable Unstable Stabilized Stable Stable Unstable Stablhzed Stable Stable Unstable Stabilized Stable


M R Islam et al /Journal of Petroleum Sczence and Englneermg 11 (1994) 213-226

Table 3 Parameters used for stability analysis Oxl viscosity (mPa s)

Gas type

Packing type

k~o~ (D)

o-e (dyne/cm)

Moblhty ratio

1000 4000 1000 4000 7500 7500 7500

N2 N2

Glass bead Glass bead Glass bead Glass bead Res. sand Res_ sand Res sand

33 33 40 40 5 5 5

1433 1433 1100 l 100 1433 1433 1433

70,000 280,000 65,000 260,000 53,000 53,000 530,000

CO2 CO2 N2 N2 N2

est one ranges from 65-67% of the initial oil-inplace (IOIP), the medium one ranges from 4 9 51% of the IOIP, and the lowest one is less than 10% of the IOIP.

5.1. Effects of gas injection pressure Figure 2 compares oil recovery performance for the 1000 mPa s oil. In this figure, oil recovery values are plotted as a function of the field scale

time of production. Recoveries for all the cases are reported by extrapolating results for a 500 m long horizontal well. Note that all the runs show initial increase in oil production rates. There is a time delay between gas injection and the initiation of oil production. As oil production starts at the production well, the rate increases to reach a steady-state value. However, the steady-state production rates are not sustained because the gravitational overhead decreases as the gas dis-


I 175

% I~-

Run 4 10 psi



13:: ~,-

Run 3 : 5 psi


O o




0 t i i



I w


Run 2 1 psi I


Run 1 gravity drainage I 200





Days of Production Fig 2. Oll production for N2 injection ( 1000 mPa sod )

I ~O0


M R Islam et al / Journal of Petroleum Scwnce and Engmeermg I I (I 994) 213-226

placement front travels downward. The oil production rate by free gravity drainage reaches a value as high as 10 m3/day. Following this, the oil production rate declines steadily, although production continues for a long time. This gravity drainage run was continued until the oil rate became nil. The total laboratory displacement test time corresponded to a filed time of 11 years. At the end of this period, 67% of the oil-in-place was recovered. This appears to be the maximal attainable recovery through free or gas-assisted gravity drainage for this particular oil. The role of gas injection is to enhance this production so that the same amount of oil may be recovered at a shorter time period. Gas injection at an injection pressure of 6.9 kPa ( 1 psi) increases oil production substantially. However, as the gas displacement front travels downward, the oil production rate declines. At the end of 400 days, both free gravity drainage and gas-assisted gravity drainage give essentially the same oil production rate. The oil production rate is further enhanced by injecting gas at injection pressures of 34.5 kPa (5 psi) and 68.9 kPa ( 10 psi). These runs were continued until gas breakthrough took place. After gas breakthrough, oil cut dropped to a negligible value. Sharma and Bentsen (1987) found viscous fingering to take place at an instability number greater than rt 2 for an oil/water system. Note that Runs 1 and 2, for which the instability numbers were low (less than n2 ), gave significantly higher oil recovery than that of Runs 3 and 4. Instability numbers for Runs 3 and 4 were higher than n 2, even though visual observation indicated a stable displacement front. As the injection pressure was increased further (103.4 kPa, 15 psi), profuse viscous fingering took place. As the &splacement test was continued further, a single finger became dominant and propagated with much higher velocity than the rest of the fingers and caused an early gas breakthrough. This led to a sudden drop in oll production. However, unlike previous cases, oil flow rate was not nil and a final recovery of 14% of the IOIP took place when the gas-oil ratio ( G O R ) reached a high value of 7000 m3/m 3. Thin G O R value was arbitrarily set in order to compare the final recov-


enes of different runs. The last experiment in this series of N2 injection runs was conducted with a high injection rate (corresponding to an injection pressure of 103.4 kPa, or 15 psi) but with periodic shut-in's of the production/injection well. Each time a viscous finger was apparent, the wells were closed and the displacement front was allowed to be stabilized. Once the front was stabilized, the injection and production were resumed again. In a field situation, this process would correspond to periodic shut-m's to stabilize the gas displacement front. Cumulative recoveries for the visibly stable runs are plotted in Fig. 3. Note that the free gravity drainage case recovers oil in an extremely slow but steady pace. With an additional gas pressure of only 6.89 kPa (1 psi), oil recovery was increased to some 70% in two years time. Higher injection pressure cases, on the other hand, cease to produce oil only after 100 days of production. During this time, however, some 40-50% of the IOIP is recovered. If a short term benefit is sought, it is worth producing these wells at the highest allowable rates (controlled by stability of the displacement front). However, if maximum ultimate oil recovery is more important, it is more prudent to operate with low injection-pressure gas-assisted gravity drainage This case appears to give much higher recovery than the higher injection pressure cases.

5 2. Comparison of C02 and N2 mjectton runs Oil vlscostty I000 mPa s Runs 7-12 were conducted using C O 2 a s the displacing phase. As was indicated in Table 2, Runs 7-10 were visibly stable. Production rates of these stable CO2 front cases are compared with stable N2 injection runs in Fig. 4. As can be seen from Fig. 4, gas composition does not appear to affect the recovery of 1000 mPa s oil for low pressure cases. The difference between the two gases is obvious for higher injection pressure cases. Note that CO2 injection for both 34.5 kPa (5 psi) and 68.9 kPa (10 psi) cases lead to a quicker response in oil production than that of corresponding N2 injection cases. However, a drop in production rate occurs more quickly for


M R. Islam et al / Journal of Petroleum Sctence and Engmeenng I 1 (1994) 213-226 8000


Run 3:5 psi


Run 4 10 psi



r J


0 o

w i m I

¢r o

Run 2. 1 psi












~ ~ R'un 1: gravity drainage


E 0


200O i

/ .




.1 (

/' i










Days of Production Fig. 3 Cumulative od production for N2 rejection ( 1000 mPa s o11) 2OO

Run 4:10 psi










/ ~;-"'-






Run 10:10 psi Run3 5psi

i i


5 psi


' 2s41[ :" Run 1: l p s l ; R u n 8 l p s l drainage (Run 1, Run 7) -0

1 0


I 200





YI 600

Days of Production Fig. 4 Comparison ofoll production rates for COz and N2 ( 1000 mPa s o11)


M R Islam et al / Journal of Petroleum Sctence and Engtneermg I 1 (1994) 213-226

the CO2 injection cases. This happens despite the fact that instability numbers in both cases are very close to each other. As can be seen in Fig. 4, recovery was greatly improved by switching from gravity drainage to 6.89 kPa (1 psi) gas-assisted gravity drainage. For both these cases, breakthrough recoveries were quite high (71% and 70%, respectively). However, oil recovery was much quicker for the 6.89 kPa ( 1 psi) case. Breakthrough recovery for the 34.5 kPa (5 psi) run was 55% IOIP, which is significantly lower than the 6.89 kPa ( 1 psi) case. For an injection pressure of 68.9 kPa (10 psi), the breakthrough recovery was 49% of the IOIP. These values were substantially lower than the 6.89 kPa ( 1 psi) or the free gravity drainage cases even though no viscous fingering was observed visually. The stability analysis indicates that Run 9 (34.5 kPa, 5 psi) is almost at the boundary of the stable flow regime (see Table 2 for Instability numbers). Run 10, on the other hand, falls under the unstable flow regime ( I > rr2). Simdar to what was observed with nitrogen, CO2 injection at 103.4 kPa (15 psi) gave rise to another flow regime characterized by very low breakthrough recovery. For Run 10 the breakthrough recovery was only 11% of the IOIP. The breakthrough time, for this case was delayed substantially (up to 38%) by periodically shutting in the production/injection wells. Figure 5 also compares cumulative oil recovery for CO2 with those of nitrogen injection cases. As can be seen in this figure, very little difference is evident for gravity drainage or 6.89 kPa ( 1 psi) injection pressure case. For the injection pressure of 103.4 kPa (15 psi), carbon dioxide recovers more oil during the early stages of the test, but eventually flow tapers off leading to a lower final recovery than that of the nitrogen case. For an injection pressure of 68.9 kPa (10 psi), on the other hand, carbon dioxide shows higher recovery throughout the displacement test. Cumulative oll recovery during the early stages of producnon was the highest for an injection pressure of 34.5 kPa (5 psi). Higher injection pressure increases production in the short term but poses the risk of invoking viscous fingering which may lead to an early gas breakthrough. The run


with 68.9 kPa ( 10 psi) injection pressure showed high oil recovery during the early stages but led to a quick breakthrough resulting in poor ultimate oil recovery. Oil vtscostty of 4000 mPa s Runs 13-16 were conducted with 4000 mPa s oil using different rejection pressures with nitrogen. As can be seen from Table 2, Runs 13 and 14 were visibly stable. Runs 17-20 were conducted using CO2 as the displacement phase. An oil viscosity of 4000 mPa s was used for these runs as well. Runs 17 and 18 were visibly stable among the CO2 rejection runs. Oil production rates of these runs are shown, along with stable COz injection runs in Fig. 6. The trend in this case is similar to what was observed with the 1000 mPa s viscosity oil. However, oil flow rate was quite limited. For this case, the onset of viscous fingering takes place at an rejection pressure of 68.9 kPa (10 psi). This particular case corresponds to an unstable displacement front as per the analysis of Sharma and Bentsen ( 1987 ). Visual observation confirmed this observation. Similar to previous runs with 1000 mPa s oil, shut in's were tried in this series of runs as well. Note that the periodic shut in and production does stabilize the gas displacement front and the gas breakthrough is delayed substantially. Higher breakthrough recovery takes place for the 6.89 kPa ( 1 psi) injection pressure case (Run 13). Run 14 shows a substantial drop in breakthrough oll recovery. Visual observation indicated that the front was stable for this case even though theoretical prediction indicated otherwise ( I > n2). A drastic drop m breakthrough recovery takes place for an injection pressure of 68.9 kPa ( 10 psi) for which clear viscous fingerlng took place. Profuse viscous fingering led to very early breakthrough and a recovery of only 6% of the IOIP. Similar to previous cases, substantial improvement in breakthrough recovery was obtained by periodically shutting in injection/production wells. It was noted, however, that the time needed for stabihzlng the displacement front was h~gher than that of the prewous low-viscosity oil. When CO2 runs are compared with N2 runs, it can be seen that the oll flow rate

M R Islam et al /Journal of Petroleum Sczence and Engtneenng 11 (1994) 213-226


Run 1 0 10 psi



, RL~ 3 5 psi

. . . . . . . .

N 2






Run4 lOpsi

Run 8 1 psi


t /Run9


o 13



1 psi


Q. 4000 S






d r a i n a g e (Run 1; R u n 7)

E 0















Days of Production Fig. 5. Comparison of cumulative oil recovery for CO2 and N2 ( 1000 mPa s oil),

CO 2 ......... f


N 2






~, ,,



, Run14 5psi


Run 18 5 psi


L 4

c O o

13 O n

O # i 1









Days of Production Fig 6 Comparison o f o d producUon rates for CO2 and N2 (4000 mPa s o11)


81 pue IE sunH aoj soma uononpoad po3o uosuedmo o g ~[:I

uo!lonpoJcl ,to s~eO 0O9 I

O09 I

O01, I


00~ I


/o 0



0 -13

S (3_ cO


m .

0 --



23 £

(D 8~ u n H




El. L

8 ~t

(ltO s edtU 00017) ~N pue ZOO ~oj LtOA03~IFO ~^qvlnmnojo uosuedmo D

L ~I=I

uo!lonpoJcl ,to sAe(] OOg

0O9 I




,sO I, ~;I. unEI


00~ I











< 0

0 -O(X~

)sd£ t ~ u n H






.-n o o 0 < t~

3 OOO~

!sd ¢3 8 I. unH



9Z g-g l g (P66I) I I guuaautXu51puv aouatoN vunaloslad f o ivuanof / i v la utVFl 8 14[

M R Islam el al /Journal of Petroleum Sctence and Engtneermg 11 (1994) 213-226


7 Run 21: no Swi

.. o ..........


3, I












Days of Production Fig. 9. Comparisonofcurnulatlveoil recoveryfor Runs 21 and 18.

is distinctly higher for CO2 injection for both injection pressures of 6.89 kPa (1 psi) and 34.5 kPa (5 psi). This effect is even more obvious when cumulative oil recoveries of these cases are compared. This is shown in Fig. 7 which compares cumulative oil recovery curves for Runs 13 and 14 (N2 injection) along with those for Runs 17 and 18 (stable CO2 injection). It appears that for a higher oil viscosity, CO2 is likely to recover more oil than N2 provided that both cases are in the same flow regime (1.e., similar instability numbers). Substantial improvement in oil recovery took place for an injection pressure of 34.5 kPa (5 psi) even though the breakthrough time for this case was shorter than the 6.89 kPa ( 1 psi) case. This led to a lower ultimate oil recovery for the 34.5 kPa (5 psi) case. For this particular oil viscosity, CO2 appears to increase the breakthrough time slightly. Even though the operating pressure is very low, it is likely that oil swelling played some role. This is particularly the case because the oil viscosity is so high that any swelling

at the displacement front would make a s~gnificant difference. Clear viscous fingering took place for Run 19 for which breakthrough recovery was only 7%. The instability number for this run indicates that indeed this run should show unstable displacement front. Run 20 was a repeat of Run 19 except that periodic shut m's were tried in order to stabilize the displacement front. Breakthrough time was delayed for this run to 21% of the IOIP.

5 3_ Effect ofconnate water All the previous runs have been carried out m the presence of connate water. In order to determine the effect of connate water on oll recovery, Run 21 was conducted in the absence of connate water. The visual model was saturated with 100% crude oil of 4000 mPa s viscosity. Figure 8 compares oil production rates of this run with those of Run 18 which was an identical run except that 10% connate water present. Figure 9 shows cu-

M R Islam et al / Journal of Petroleum Sctence and Engtneenng 11 (1994) 213-226

mulattve oil recovery for these runs. Oil recovery was found to be slightly higher in the absence of connate water. However, when these results are compared as fraction of initial oil-in-place, both cases show almost identical results. It appears that in both cases, the injected gas acts as a nonwetting phase and 'ignores' the aqueous film on the rock surface.

6. Conclusions The feasibility of inert gas injection in heavy oil reservoirs was investigated. A series of lowpressure model tests indicated that the recovery using gravity-stabilization may be as high as 70% of the oil-m-place for moderately viscous heavy oil reservoirs. The importance of maintaining a stable displacement front was demonstrated through several displacement tests under unstable, as well as pseudostable displacement regimes. Unstable or pseudostable flow regimes are likely to prevail in the situation where gravity stabilization does not take place.

Acknowledgement Financial assistance provided by the Federal Panel on Energy R & D ( P E R D ) through CANMET, Energy, Mines and Resources Canada, in support of this work is gratefully acknowledged. Mr. C. Talukder and Mr. Y. Hu carried out the laboratory work.

References Babson, E C , 1989 A review of gas lnjectxon projects in Callforma SPE 18769, presented at SPE Cahf Reg. Meet, Bakersfield, Cahf Bentsen, R G_, 1985 A new approach to mstabdlty theory in porousmedla Soc Pet. Eng J , 1 5 765-779_ Carlson, L O_, 1988. Performance of Hawkins Field Umt under gas drive pressure maintenance operations and development of an enhanced oil recovery project SPE/DOE 17324, presented at SPE/DOE 6th Syrup EOR, Tulsa, Okla


Chung, F T H , Jones, R A and Burchfield, T_E, 1988 Recovery of viscous oil under high pressure by CO2 displacement - - A laboratory study SPE 17588, SPE lnt Meet. Petroleum Engineering, Tlanjln, China Craft, B C and Hawkins, M F, 1959_ Applied Petroleum Reservoir Engineering. Prentice-Hall, Inc, Englewood Chffs, N J , pp. 355-408 Danesh, A, KrlnlS, D , Harderson, G.D and Peden, J M , 1989 Pore level visual mvestigatxon of miscible and immiscible displacements J Pet Sci. Eng, 2(3) 167_ Fayers, F J., 1991 A summary o f R & D needs for gas inJection processes. Proc 6th Eur. Symp Improved O11 Recovery, pp 75-98 Garcla, F M , 1983 A successful gas rejection project in heavy oil reservoir. SPE 11988, presented at SPE Callf Reg Meet, San Francisco, Cahf Higglns, R.V, 1953 Factors that influence gravity drainage Pet Eng., (May) 83-92 Islam, M R., Erno, B_P and Davis, D , 1992 Hot gas and waterflood equivalence ofm-sltu combustion_ J Cen Pet Technol_, 31 (8) 44-52 Kantzas, A , Chatzxs, I and Dulhen, F.A_L., 1988a Enhanced oll recovery by inert gas rejection SPE/DOE 17379, 6th Symp Enhanced O11 Recovery, Tulsa, Okla Kantzas, A , Chatzis, I and Dulhen, F A L, 1988b. Mechanisms of capillary displacement of residual oi1 by gravity assisted Inert gas mject~on SPE 17506, SPE Rocky Mountain Reg Meet., Casper, Wyo_ Katz, D.L_, 1942 Posslbdltles of secondary recovery for the Oklahoma City Wdcox sand Trans AIME, 146.28 Leverett, M_C, 1941 Capdlary behavlour m porous sohds. Trans AIME, 142 152_ Lewis, J O_, 1944 Gravity drainage in od fields Trans AIME, 155 133 Lozada, D. and Farouq All, S M , 1989 Experimental design for non-equdxbrlum lmmxsclble carbon dioxide flood. Proc 4th U N I T A R / U N D P Int Conf Heavy Crude and Tar Sands, 4- 275-296 Mayer, E H_, Erlougher, R C_, Sr, Splvak, A. and Costa, A., 1986. An analysis of heavy oil xmmlsclble CO2 tertiary core flood data. SPE/DOE 14901, presented at SPE/DOE 5th Syrup. EOR, Tulsa, Okla_ Nectoux, A, 1987 Eqmhbrlum gas oil drainage velooty, gravitational and composmonal effects Proc_ 4th Eur Symp Improved Oll Recovery, Hamburg, Germany Sarma, H K. and Bentsen, R G , 1987 An expertmental verification of a modified mstabdlty theory for tmmlsclble displacement m porous medm_ J Can_ Pet Technol., (JulyA u g ) , 2 6 ( 4 ) 88-99 Splvak, A., Garrison, W H and Nguyen, J P, 1988 A review of the LBOD Immiscible COz ProJect, Tar Zone Fault Block V, Wdlmlngton Field, California SPE 17407, presented at SPE Cahf. Reg. Meet., Long Beach, Calif. Stahl, R.F, Martin, W A and Huntington, R L,2 1943. Gravitational drainage of hqmds from unconsolidated Wdcox Sand. Trans AIME, 151 138-145


M R. Islam et al /Journal of Petroleum Sctence and Engmeermg I1 (1994) 213-226

Terwdhger, P L., Wdsey, L.E., Hall, H.N. and Bridges, P M., 1951. An experimental and theoretical investigation of gravity drainage performance Trans_ AIME, 192 285

Ypma, J, 1985 Composmonal effects m gravity dominated nitrogen displacement. SPE 14416, presented at 60th Annu. Tech Conf Exhlb., Las Vegas, Nev