Laboratory Investigation of

Enhanced Light-Oil Recovery By

CO2/Flue Gas Huff-n-Puff Process
Y.P. ZHANG, S.G. SAYEGH, S. HUANG
Saskatchewan Research Council
M. DONG
University of Regina

have been mentioned in the literature(4-6): a) oil viscosity reduc-

Abstract tion; b) oil swelling; c) solution gas drive; d) relative permeability
This paper focuses on phase behaviour measurements with hysteresis due to reduced water saturation, drainage/imbibition,
reservoir oil-CO2 mixtures and on coreflooding tests in the huff- and wettability alternation; e) repressurization; f) gas diffusion and
n-puff mode to characterize the system, determine the influential mass transfer; and, g) interfacial tension reduction in the zone near
mechanisms, and supply data for simulation of the field imple- the wellbore.
mentation. The results indicate that significant amounts of CO2 The purpose of this work is to investigate the potential for ap-
could dissolve in the oil, which caused oil swelling and viscosity plying the CO2 huff-n-puff process in a medium-gravity oil reser-
reduction. During the puff cycle, the oil retained CO2 preferen- voir in Saskatchewan. There are many variables in the application
tially to methane; thus, the beneficial swelling and viscosity ef- of the huff-n-puff process, so laboratory measurements were car-
fects were maintained over an extended portion of this cycle. ried out to characterize the system, determine the influential mech-
Corefloods were performed to investigate the effect of water- anisms, and supply data for simulation of the field implementation.
flood residual oil saturation and injection gas composition (CO2 Phase behaviour measurements provided the basic fluid properties
and enriched flue gas) on oil recovery. Incremental oil recovery of the reservoir oil and its mixtures with the injected gases. Core-
was observed to be sensitive to waterflood residual oil saturation floods examined the effect of different process parameters, such as
and to the process application scheme. Coreflooding results sug- slug size and number of huff-n-puff cycles, on oil recovery effi-
gest that the huff-n-puff process may be more suitable to oil-wet ciency. Oil samples were taken from the Elswick East Midale field,
than water-wet reservoirs. located in southeast Saskatchewan. Up to December 2002, this field
had produced about 16% of the initial oil in place (IOIP)(7). The
water cut has currently exceeded 85%; therefore, EOR methods are
needed for increasing the recovery factor.
Introduction
Various technologies have been applied in tertiary oil recovery
processes, such as gas miscible/immiscible injection and chem- Sample Preparation and Experimental
ical flooding. Among these enhanced oil recovery (EOR) methods,
the huff-n-puff process has been reported to be economic at an oil Set-Up
price of less than US$20/STB and CO2 costs of US$40/ton(1). For Crude oil and brine samples were supplied by Nexen Canada
example, it was shown in a flue-gas huff-n-puff project(2) that oil Ltd. from the Elswick East Midale field. A sample of the oil phase
production rates stabilized, and the project proved to be cost effec- was flashed down to atmospheric pressure, and the density and vis-
tive with small investment requirements and low operating costs. cosity of the resulting dead oil were measured and are presented in
In another case(3), the CO2 huff-n-puff process was not successful Table 1. The density and viscosity of the dead oil were not mea-
in increasing incremental oil recovery. However, there were re- sured at the reservoir temperature of 68° C due to a significant
duced water-handling and electrical requirements during the injec- loss of light ends by evaporation. Consequently, these values were
tion, soak, and flow phases, which were beneficial to the project. estimated by extrapolating the low-temperature measurements,
There is increasing interest in CO2/flue-gas huff-n-puff injection and are included in Table 1. The oil has an API gravity of 30.3°,
into single wells because the process is relatively easy to apply and which places it as a medium-light oil, and has an average molec-
does not require a large initial capital outlay. The process typically ular weight of 224 kg/kg-mol.
begins with the injection of a slug of gas into a single well. This is The formation brine was represented by produced water col-
followed by a shut-in or soak period to allow the gas to dissolve lected from a battery near the oil well. It was filtered through a 2.5
into the oil, swell its volume, and reduce its viscosity. The same μm pore size filter, and analyzed for densities and viscosities at
well is then returned to production and the response is monitored. three temperatures (15, 20, and 60° C), as well as mineral content
In reservoirs with poor inter-well communication, this single-well (Table 2). The density and viscosity at 68° C and ambient pressure
approach may be one of the best ways, and sometimes the only obtained from regression were 1,080.6 kg/m3 and 0.57 mPa⋅s, re-
way, to accelerate response in underperforming wells. Since mis- spectively.
cibility between the reservoir oil and injected gas is not a require-
ment of the huff-n-puff process, it is well suited for low pressure
reservoirs and for gases with high minimum miscibility pressures
Reconstitution of Reservoir Fluid
such as flue gas. The reservoir fluid used in this study was made by recom-
The mechanisms involved in the production of oil during gas bining the sampled gas and oil to a gas-oil ratio (GOR) of 60 m3/
huff-n-puff are diverse and complex. The following mechanisms m3, which is the estimated reservoir fluid GOR from a previous
PEER REVIEWED PAPER (“REVIEW AND PUBLICATION PROCESS” CAN BE FOUND ON OUR WEB SITE)

24 Journal of Canadian Petroleum Technology

TABLE 1: Chemical and physical properties of cleaned Elswick dead oil.

Temperature (° C) 15 20 68
Density (kg/m3) at 0.1 MPa 873.8 (30.3° API) 870.2 835.6*
Viscosity (mPa•s) at 0.1 MPa 12.4 10.4 3.04*
Basic sediment and water Trace
Oil molecular weight (g/g-mol) 224
* Estimated based on regression

TABLE 2: Properties of Elswick brine (produced water). total GOR, density, swelling factor, formation volume factor, and
viscosity of the liquid phase as a function of saturation pressure.
Property Value The saturation pressure of the reconstituted oil was determined to
be 10.2 MPa from isothermal pressure-volume expansion plots as
Density (kg/m3) @ 15° C 1,106.2 well as traditional Y-function plots.
@ 20° C 1,104.2
@ 60° C 1,084.6
@ 68° C 1,080.6* Oil-CO2 Mixture Phase Behaviour
Measurement Apparatus
Viscosity (mPa•s) @ 15° C 1.52 The experimental apparatus used for this study is a traditional
@ 20° C 1.33 PVT apparatus designed for operation at typical reservoir tempera-
@ 60° C 0.65 tures and pressures. A variable-volume high-pressure cell is placed
@ 68° C 0.57* in a thermostated oven. Fluids (oil and gas) are introduced into the
cell under pressure, and then mixed to achieve equilibrium at the
Refractive index @ 25° C 1.3584 desired temperature and pressure. Viscosities are measured by run-
ning the fluids at a constant rate through a capillary viscometer.
Conductivity (mS) @ 25° C 208 Fluid densities are measured in a high-pressure oscillating tube
pH @ 20° C 6.91
densitometer (Anton Paar DMA 512 P) at reservoir temperature.
A single-stage separator flash procedure is used to obtain the
Total dissolved solids (mg/L) GOR, composition, and formation volume/swelling factors. In this
@ 120° C 161,000 procedure, a sample of fluid from the cell is flashed directly down
@ 180° C 159,000 to atmospheric conditions. The liquid and gas portions resulting
from this process are collected separately. The weight and density
Major components in brine (mg/L)
of the liquid are measured, as well as the volume and composition
Chloride (Cl-) 90,400 of the gas. The results are used to calculate the composition, GOR,
Sulfate (SO42-) 2,300
etc., of the whole sample.
Sodium (Na+) 47,300
Potassium (K+) 870
Magnesium (Mg2+) 720 Huff-n-Puff PVT Procedure
Calcium (Ca2+) 3,010
The experiments carried out in this study were designed to sim-
Iron (Fe2+) < 0.1
< 0.8
ulate the sequence that occurs during a typical huff-n-puff process.
Barium (Ba2+) As mentioned in the introduction, typical huff-n-puff operations
Manganese (Mn2+) < 0.08
begin with carbonation or pressurization of the oil during the gas
* Estimated based on regression injection stage (huff). A soak period allows the gas to dissolve
into the oil. Finally, liberation or depressurization of the oil occurs
pressure-volume-temperature (PVT) study. The following PVT during the production stage (puff). Accordingly, the oil was satu-
properties of the reconstituted fluid were measured: GOR, vis- rated with injection gas in several steps of increasing pressure up to
cosity, density, and formation volume factor (FVF). a maximum value for the huff part of the process. The pressure was
The viscosity and density were measured over a range of pres- then reduced in a number of stages down to atmospheric condi-
sures, which made it possible to determine the value at saturation tions. At each step, the oil and gas were equilibrated, and the prop-
pressure by a short extrapolation. The equilibrium reservoir fluid erties of the oil and gaseous phases were measured as described in
properties are presented in Table 3. These properties include the the previous section.
TABLE 3: Equilibrium liquid properties during CO2 huff-n-puff at 68° C.

Description Psat Total Mixture Live Oil FVF SF Mixture

(MPa) GOR Density Density (m3/m3) (m3/m3) Viscosity
(sm3/m3) at Psat at Psat at Psat
(kg/m3) (kg/m3) (mPa•s)

Recombined oil 10.2 60.4 796.6 796.6 1.1924 1 1.72

Recombined 11.5 77.3 803.9 797.9 1.240 1.040 1.43

oil–CO2 12.4 89.5 806.8 798.7 1.250 1.049 1.28
mixtures huff 13.7 111.1 809.8 800.0 1.298 1.088 1.11
cycle 15.7 133.7 815.7 801.9 1.345 1.128 1.01
Recombined 12.6 127.9 808.1 798.9 1.344 1.127 0.98
oil–CO2 10.0 104.0 811.3 796.5 1.290 1.082 1.12
mixtures puff 7.1 73.3 815.3 793.6 1.225 1.027 1.39
cycle 3.0 31.7 824.7 789.7 1.128 0.946 -
Dead oil 0.1 0.0 837.4 - 1.000 - 2.95

February 2006, Volume 45, No. 2 25

 FIGURE 1: Diagram of huff-n-puff coreflood apparatus.

an inexpensive EOR process. However, the operating costs could

Coreflooding Apparatus be further reduced by using flue gas instead of CO2. Therefore, in-
A schematic diagram of the huff-n-puff coreflooding appa- jection gases used in the experiments consisted of pure CO2 and a
ratus is shown in Figure 1. Four transfer vessels were used to in- synthetic enriched flue gas (30% CO2 + 70% N2) to allow compar-
ject the cleaned dead oil, recombined oil, injection gas, and brine. ison of the gases’ oil recovery efficiency.
Cores, measuring around 30 cm in length and 5 cm in diameter, Each test began with evacuating the mounted core, then satu-
were placed in a lead sleeve mounted within a core holder, and rating it with brine to determine the pore volume. Brine floods at
submitted to a triaxial hydraulic pressure simulating overburden different flow rates were then carried out to measure the brine per-
pressure. In traditional corefloods, one end of the core is used ex- meability. To establish the initial (or connate) water saturation, the
clusively for injection of fluids, and the other end for producing core was flooded with two or three pore volumes of the cleaned
them. However, in the huff-n-puff mode, both gas injection and oil dead oil, which was then displaced with recombined oil until the
production are from the same end since this is a single-well pro- produced oil formation volume factor matched that of injected oil.
cess. Thus, the plumbing of the apparatus was flexibly designed The pressure and temperature were maintained at reservoir condi-
to enable switching from linear flooding to huff-n-puff modes. A tions (11 MPa and 68° C, respectively) while the core was aged for
sight-glass visual cell was used to monitor that no oil flowed out an extended period. Live oil was injected periodically during the
of the core during gas injection (huff cycle), and also to model ageing stage to see if more water was produced. This ageing pro-
the portion of the reservoir beyond the wellbore vicinity and to cess establishes the core’s wettability in a manner similar to that
provide pressure support during the production stage (puff cycle). occurring in reservoirs. The endpoint permeability of the core to
Pressure bleeding during the puff cycle was controlled by using the oil in the presence of formation brine was determined based on
a gas flow control meter connected to a backpressure regulator the injection rate of the oil and the pressure drop.
(BPR). The volume of produced liquids was recorded by collec- An initial waterflood was carried out after initial oil saturation
tion in a graduated glass cylinder. The produced gas was recorded was established and the ageing process completed. The initial wa-
by a gasometer, and its composition was analyzed by means of an terflood was terminated at a designed water injection volume. An
in-line gas chromatograph. Pressures at both production and injec- injection rate of 10 cm3/hr was used for all the waterfloods, which
tion ends were monitored throughout the flooding process. Sayegh translates to an approximate propagation rate of 0.6 m/day.
and Maini(8) described a comparable set-up, while Shayegi et al.(5)
used a floating piston cell to represent an aquifer. Thereafter, the huff-n-puff process was performed by injecting
pure CO2 or synthetic flue gas from a compressed gas cylinder
connected to the production end of the core (designated as Port “2”
Huff-n-Puff Coreflooding Procedure in Figure 1), allowing production from the injection end (Port “1”).
The huff-n-puff coreflood tests were conducted with recom- The injected gas volume was designed so that only brine was dis-
bined reservoir fluid and produced brine. Berea cores were used placed during the huff stage. Upon completion of the gas injection
rather than reservoir cores since the former are known to be rea- (huff cycle), the core was shut in for a three- or four-day soak pe-
sonably uniform, hence more suitable for a mechanistic study. In riod. The puff cycle was then carried out with the production at the
addition, Berea cores can be easily manufactured with larger pore core production end (on Figure 1, from Port “1” to “2”), driven by
volumes than reservoir core plugs, which improves the accuracy brine injection at the core injection end, or by the pressure support
of the results. All the floods were carried out at the reservoir tem- from the sight glass vessel.
perature of 68° C. The second huff-n-puff cycle was carried out immediately after
It was reported that using a small amount of CO2 with nitrogen oil production from the first puff had ceased, and followed a sim-
not only recovers more oil than does pure CO2, but is also cheaper ilar procedure to the first huff-n-puff. The same procedures were
and less corrosive(9). The single-well huff-n-puff process is already repeated in the third huff-n-puff cycle. The test was then followed
26 Journal of Canadian Petroleum Technology
 FIGURE 2: Gas solubilities in Elswick recombined oil-CO2
huff-n-puff.
FIGURE 3: FVF and SF in Elswick recombined oil-CO2 huff-n-
puff.
by a four-stage blowdown. The ultimate oil recovery was subse-
quently calculated by volumetric material balance.
higher than those during the huff cycle for pressures down to about
6 MPa. At lower pressures than that, the swelling factors are less
than unity because they are referenced to the live oil and not the
Results and Discussion dead oil. A somewhat different definition of swelling factor should
be formulated for the huff-n-puff process to take into account that
Phase Behaviour Study this is an open system and all the gases are evolving as the pressure
The huff-n-puff equilibrium properties of CO2 with the reservoir is depleted. This makes the concept of oil swelling confusing if the
fluid are presented in Table 3. The total gas solubility in the oil is process involves several huff-n-puff cycles.
represented in Figure 2 as a GOR. The gas is the total amount of all The results of the oil phase density measurements are presented
the gaseous components in the system—those originally in the re- in Table 5 and plotted in Figure 4 in both the direction of increasing
combined oil plus the added CO2. The amount of dissolved gas in- (huff) and decreasing (puff) pressure. It is noted that the first line
creased almost linearly with increasing saturation pressure during in each set of data is the extrapolated value for the saturated fluid
the huff cycle. It should be mentioned that CO2 was not miscible as presented earlier in Table 3, and the remaining points (at higher
with the oil at pressures up to the maximum pressure of the experi- pressures) are for the undersaturated liquids. It is observed that the
ments (15.7 MPa); however, it had substantial solubility, reaching density of the oil-rich phase increases with increasing saturation
133.7 sm3/m3. The gas/oil ratios on the puff cycle were higher than pressure. This indicates that volume shrinkage occurs when CO2
those on the huff. A similar observation was made in the work of dissolves into the oil. When the pressure is reduced, there are two
Sayegh et al.(9). This is attributed to the oil preferentially retaining factors that determine the net change in density: the loss of gas and
carbon dioxide and releasing methane on the puff cycle. This is the compressibility of the liquid. The final result is that the density
confirmed by the fluid analysis data (Table 4) from which the mole initially decreases as the pressure decreases, then increases to the
fraction of carbon dioxide in the puff cycle is greater than that in value of the dead oil.
the huff cycle, while the reverse is true for methane. Another way The plots in Figure 4 show that the undersaturated fluid den-
of interpreting this result is to say that the volatility, or K-value, of sities increase linearly with pressure in the range considered. It
methane is higher than that of CO2. is interesting to note that the pressure-density plots of all the dif-
The dissolution of CO2 into the oil resulted in substantial oil ferent mixtures are almost parallel, even though the fluids have
swelling, as shown in Figure 3. The swelling factor rose almost varying compositions. Since the slope of the pressure-density plot
linearly, from the value of 1 for the live oil at its saturation pres- determines the fluid compressibility, it may be concluded that,
sure (10.2 MPa), to 1.13 at the maximum CO2 saturation pressure in the range studied, composition has little effect on the fluid
of 15.7 MPa. The swelling factors on the puff cycle are seen to be compressibility.
TABLE 4: Equilibrium fluid compositions during CO2 huff-n-puff at 68° C.

Composition (mol%)
Psat Equilib.
Description
(MPa) Phase N2 CO2 H2 S C1 C2 C3 iC4 nC4 iC5 nC5 C6+

Reservoir
10.2 - 1.6 0.4 0.0 19.3 6.8 7.1 1.2 4.0 1.8 2.1 55.8
fluid
Reservoir 11.5 Liquid 1.9 10.3 0.0 16.5 6.0 6.3 1.1 3.5 1.6 1.9 50.9
fluid-CO2 12.4 Liquid 1.9 17.7 0.0 14.4 5.3 5.8 1.2 3.4 1.6 1.9 46.7
mixtures 13.7 Liquid 1.5 27.6 0.0 12.1 4.6 5.2 0.9 3.1 1.4 1.7 41.7
huff cycle 15.7 Liquid 0.9 36.2 0.0 10.0 4.2 4.5 0.8 2.9 1.4 1.7 37.5
12.6 Liquid 0.9 34.9 0.0 10.0 4.3 4.6 0.9 2.9 1.4 1.7 38.5
Vapour 6.0 59.1 0.0 25.5 4.4 2.8 0.4 1.0 0.4 0.5 0.0
Reservoir 10.0 Liquid 0.4 31.2 0.0 8.6 4.4 4.9 0.9 3.0 1.4 1.7 43.5
fluid-CO2 Vapour 4.6 57.7 0.0 28.6 4.8 2.7 0.3 0.8 0.2 0.3 0.0
mixtures 7.1 Liquid 0.2 26.0 0.0 5.7 4.1 5.0 1.0 3.2 1.5 1.8 51.5
puff cycle Vapour 2.3 62.0 0.0 25.9 5.3 2.9 0.3 0.8 0.3 0.3 0.0
3.0 Liquid 0.1 13.9 0.0 1.8 3.1 4.8 1.0 3.2 1.6 1.9 68.6
Vapour 1.2 66.6 0.0 18.2 7.0 4.5 0.5 1.3 0.3 0.4 0.0

February 2006, Volume 45, No. 2 27

TABLE 5: Effect of pressure on densities of CO2-Elswick recombined oil – huff and puff cycles.
Huff cycle Puff cycle
CO2 Content Pressure Density CO2 Content Pressure Density
(mol%) (MPa) (kg/m3) (mol%) (MPa) (kg/m3)
11.50* 803.9 12.60* 808.1
15.47 808.0 14.52 810.2
17.01 809.6 16.69 812.7
10.3 34.9
19.08 811.7 19.23 815.6
20.83 813.5 21.58 818.2
22.50 815.2 - -
12.40* 806.8 10.04* 811.3
16.48 810.9 12.42 814.1
17.7 18.33 812.9 31.2 14.36 816.4
20.45 815.0 16.46 818.9
22.74 817.3 18.39 821.1
13.70* 809.8 7.10* 815.3
16.71 813.1 10.71 818.8
27.6 18.29 814.8 26.0 13.66 822.0
20.31 817.1 16.71 825.2
21.97 818.8 19.79 828.3
15.70* 815.7 3.02* 824.7
16.44 816.5 5.40 826.8
36.2 18.34 818.7 13.9 8.56 829.8
20.49 821.0 11.51 832.5
22.24 822.9 14.43 835.1
* Saturation pressure extrapolated value

FIGURE 4: Mixture densities in Elswick recombined oil-CO2 FIGURE 5: Mixture viscosities in Elswick recombined oil-CO2
huff-n-puff. huff-n-puff.

The viscosities of CO2-oil mixtures during the huff and puff from 6,822 mPa⋅s for the recombined oil down to 226 mPa⋅s for the
stages are tabulated in Table 6 and plotted in Figure 5. As with CO2-saturated oil at 6.55 MPa. This should be taken into account
Table 5, the first lines of data in each set in Table 6 are the esti- when screening reservoirs for applicability of cyclic gas stimula-
mated values for the saturated fluid, and the remaining data are tion enhanced oil recovery.
measured for the undersaturated liquids. In all cases, the viscosity An interesting phenomenon is observed when the pressure is re-
increases with increasing pressure, as is expected. As with density duced: the viscosity remains low as gas is released from the liquid
(see Figure 4), the slopes of the pressure-viscosity plots are almost phase, and is consistently lower than the viscosity of the liquid
constant and do not vary appreciably with composition. phase of the huff cycle. This is a result of the oil-rich phase re-
From Figure 5, it is observed that the saturated oil-rich phase’s leasing methane preferentially to CO2 during pressure decrease,
viscosity decreases as it absorbs more CO2 during the huff cycle. so the residual liquid is relatively rich in CO2, which is an effec-
This is one of the main benefits of the CO2 huff-n-puff process, tive viscosity-reducing component. This is reinforced by the higher
wherein the reduced viscosity renders the oil more mobile. In the GORs during the puff cycle than during the huff, as shown in
present case, the Elswick recombined oil’s viscosity was reduced Figure 2; more dissolved gas results in lower viscosities.
from 1.7 mPa⋅s for the live oil to 1.0 mPa⋅s at the maximum satura-
tion pressure of 15.7 MPa. Huff-n-Puff Core Coreflooding Results
It should be mentioned that the viscosity-reduction effectiveness Laboratory coreflooding tests focused on investigating the in-
of CO2 is more pronounced with heavier, more viscous oils(8,10). fluence of the remaining oil saturation and drive gas composition
For example, Sayegh and Maini(8) reported a viscosity reduction on oil recovery efficiency for the huff-n-puff process. In total, three
28 Journal of Canadian Petroleum Technology
TABLE 6: Effect of pressure on viscosities of CO2-Elswick recombined oil – huff and puff cycles.

Huff cycle Puff cycle

CO2 Content Pressure Viscosity CO2 Content Pressure Viscosity
(mol%) (MPa) (mPa•s) (mol%) (MPa) (mPa•s)
11.50* 1.43 12.60* 0.98
14.65 1.55 13.75 1.01
15.98 1.59 15.76 1.06
10.3 34.9
17.68 1.64 17.40 1.10
19.24 1.71 19.23 1.15
20.90 1.77 20.80 1.19
12.40* 1.28 10.04* 1.12
15.44 1.38 12.47 1.20
17.53 1.45 14.22 1.25
17.7 31.2
19.26 1.51 16.09 1.31
20.92 1.57 17.81 1.37
22.67 1.63 19.38 1.42
13.70* 1.11 7.10* 1.39
16.26 1.20 11.88 1.55
17.39 1.23 14.17 1.62
27.6 26.0
18.93 1.27 16.35 1.68
20.48 1.33 18.13 1.76
21.82 1.37 - -
15.70* 1.02
17.56 1.08
19.19 1.12
36.2
20.56 1.16
22.17 1.22
23.54 1.25
* Saturation pressure extrapolated value

TABLE 7: Summary of huff-n-puff coreflood results (saturated with Elswick live oil).

Huff-n-puff displacement test no. Run 1 Run 2 Run 3

Core data
Type of core Berea Berea Berea
Pore volume (cm3) 112 111 110
Air permeability (10-3 μm2) 181 182 187
Brine permeability (10-3 μm2) 162 97 77
Oil permeability (10-3 μm2) 66 66 59
Initial oil saturation (%PV) 55.7 58.5 57.2
Injection gas data
Type of gas injected CO2 CO2 Flue Gas**
Conditions of injected gas 68° C/11 MPa 68° C/11 MPa 68° C/15 MPa
Cycle 1 injected gas volume (PV) 0.15 0.27 0.20
Cycle 2 injected gas volume (PV) 0.54 0.15 0.08
Cycle 3 injected gas volume (PV) 0.15 0.08 0.04
Recovery data
Initial waterflood recovery (%IOIP) 62.1 40.8 36.6
Puff process Puff + 0.5 PV WF* Puff Puff
Cycle 1 (%IOIP) 1.53 4.01 6.50
Cycle 2 (%IOIP) 1.91 5.67 2.27
Cycle 3 (%IOIP) 1.53 0.15 0.30
Total tertiary oil recovery (%IOIP) 5.0 9.8 9.1
Total oil recovery (IWF + tertiary) (%IOIP) 67.1 50.6 45.7
Residual oil saturation after IWF (%PV) 21.1 34.7 36.3
Residual oil saturation after H-n-P (%PV) 18.3 28.9 31.1
S(ro-h-n-p)/S(ro-iwf) (%) 86.9 83.4 85.7
Gas utilization (sm3/m3) 3,105 881 508
Gas utilization (Mscf/STB) 17.2 4.9 2.8
* WF – waterflood ** Flue gas – 30% CO2 + 70% N2

February 2006, Volume 45, No. 2 29

coreflood runs were carried out on three Berea sandstone cores. Run 3: Flue Gas Injection to Breakthrough With No Water
The test conditions and results are summarized in Table 7. Chase in Puff Cycle
In Run 1, the core was waterflooded down to residual oil satu- This run was identical to the previous run except that synthetic
ration. This was followed by three cycles of CO2 huff-n-puff dis- flue gas (30% CO2 + 70% N2) was used instead of CO2 to investi-
placements, then an extended waterflood. Run 2 was similar to gate the efficiency of this drive gas in the huff-n-puff process. The
Run 1, except that the initial waterflood was not carried out to flood conditions and results are summarized in Table 7. It can be
completion so as to leave a higher residual oil saturation for the seen that the residual oil saturation left behind by waterflood was
CO2 huff-n-puff process. Run 3 followed the same procedure as 36.3% PV, much the same as in Run 2. The high remaining oil
Run 2, except that a synthetic flue gas (30% CO2 + 70% N2) was saturation again resulted in high additional oil recovery in Run 3,
used during the huff-n-puff process. as was observed in Run 2. It is observed that most of the tertiary
From Table 7, it is observed that the three cores had similar pore oil recovery occurred in the first cycle (6.5% IOIP). The second
volumes of 112, 111, and 110 cm3. The absolute air permeabilities cycle produced less oil (2.3% IOIP), while the third produced only
of the three cores were also very close to each other: 181 × 10-3, a trace (0.3% IOIP).
182 × 10-3, and 187 × 10-3 μm2, respectively. On the other hand,
the absolute brine permeabilities of these cores showed some vari-
ability and were measured to be 162 × 10-3, 97 × 10-3, and 77 × 10-3 Comparison of Oil Recovery Efficiency
μm2. The reason for this is not clear, but may be due to flushing From Table 7, it is observed that Run 1 had a relatively uniform
of fines out of the first core by the injected water. However, after response over all three cycles, whereas Run 2 had excellent oil re-
ageing, the cores’ initial oil saturations (55.7, 58.5, and 57.2%) and covery in the first two cycles but produced little oil in the third.
their effective permeabilities to oil (66 × 10-3, 66 × 10-3, and 59 × Run 3 had the highest response in the first puff cycle among the
10-3 μm2) compared very well. This indicates that the cores had three runs (6.5% IOIP), but only recovered a small amount of ad-
comparable properties for the subsequent huff-n-puff tests. ditional oil in the second and third cycles. The good response in
both Run 2 and Run 3 suggests that repressurization and gas rela-
Run 1: Injection of 0.15 PV CO2 With Pressure Depletion tive permeability hysteresis appear to be the important oil recovery
Then Water Chase mechanisms for the gas huff-n-puff process(8).
The cumulative tertiary recoveries of the three runs are also
Run 1 included an initial waterflood until oil production be- given in Table 7. Run 1 produced the least amount of tertiary oil
came negligible, three huff-n-puff cycles, and then a four-stage (5.0% IOIP), while Runs 2 and 3 produced comparable amounts of
blowdown. The huff-n-puff process was designed to be 0.15 pore oil (9.8 and 9.1% IOIP, respectively). Comparing Runs 2 and 3 in-
volumes (PV) of pure CO2 injected in both the first and the last dicates that, for this system, the type of gas (CO2 or flue gas) was
huff cycles, and 0.54 PV of pure CO2 injected in the second huff not an important factor for the ultimate recovery; however, it did
cycle. The three puff cycles were conducted in a similar way; pres- affect the rate of production.
sure depletion down to 11.3 MPa, then about 0.5 PV slug of brine
influx.
The waterflood residual oil saturation was determined to be Effect of Waterflood Residual Oil Saturation
21.1% PV for Run 1. Oil recovery by initial waterflood was very The coreflood results show that waterflood residual oil can be
high, up to 62.1% IOIP. Because the residual oil saturation in this recovered using huff-n-puff cyclic gas injection. The results of the
case was very low, the additional oil recoveries by huff-n-puff gas three runs (Table 7) indicate that the residual oil saturation after
injection, as listed in Table 7, were 1.5% IOIP for the first puff initial waterflood affects the oil recovery efficiency in a gas huff-n-
cycle, 1.9% IOIP for the second puff cycle, and 1.5% IOIP for the puff process. The best single-cycle oil recovery among all the runs
last puff cycle. The overall gas utilization of Run 1 was very high was 6.5% IOIP for the first huff-n-puff cycle in Run 3, which had
(3,105 sm3/m3 or 17.2 Mscf/STB) because the total tertiary oil re- an oil saturation of 36.3% pore volumes at its start. The best three-
covery was only 5.0% IOIP. cycle results, on the other hand, were for Run 2, which produced a
It was observed that the pressure drops during the first two water total of 9.8% IOIP and had 34.7% PV oil saturation at the start of
influx cycles declined continuously and were very low in the last the huff-n-puff process.
follow-up water chase. Therefore, the probable reason for poor re- CO2 was used in both Runs 1 and 2, but oil recovery was signif-
sponse in this run might be gas fingering in the core. icantly higher in Run 2 than in Run 1. This is probably due to the
higher residual oil saturation left behind by the waterflood of Run
Run 2: CO2 Injection to Breakthrough With No Water Chase 2: 34.7% for Run 2 vs. 21.1% for Run 1. The response to a second
in Puff Cycle puff cycle was also better in Run 2 with 5.7% IOIP, than Run 1
with only 1.9% IOIP for the same cycle. The higher oil recovery of
A major difference between this run and Run 1 was the wa- Run 2 over Run 1 points to the importance of having high residual
terflood residual oil saturation: in order to keep the remaining oil oil saturation prior to carrying out the enhanced oil recovery pro-
saturation in the core high, this run was flooded with only 0.2 PV cess, since the higher the oil saturation, the more mobile is the oil.
water at the initial waterflood stage. Then the core was set in the The association of higher residual oil saturation with an improved
oven under the test conditions for one day to allow redistribution of chance of success means that the earlier the huff-n-puff process is
saturations in the core. Following that, CO2 was injected to break- applied, the better. In general, high waterflood residual oil satu-
through, and then a soaking period was allowed, followed by back- ration is an indication of oil-wetness of the formation, so the re-
production by pressure depletion (puff). sults indicate that the huff-n-puff process may be more suitable to
The injected gas in Run 2 was pure CO2, the same as in Run 1. oil-wet than water-wet reservoirs. Further testing using oil-wet
The gas volumes injected in each huff cycle were determined by cores, rather than water-wet Berea, is needed to confirm this
monitoring the sight glass. As soon as the first gas bubble was ob- observation.
served in the sight glass vessel, gas injection was stopped and the
soak process was started. The puff cycles were controlled by slow
depletion of the back-pressure. The production pressure increased
Gas Utilization Efficiency
from approximately 11 MPa after the waterflood to 13.3 MPa at An important parameter for the economic success of the gas
the end of the first huff cycle. The residual oil saturation and re- huff-n-puff process is the amount of gas used per unit of produced
covery data for Run 2 are given in Table 7: the initial waterflood oil. In practice, part of the gas injected in the first cycle is recycled
recovered 40.8% IOIP, while the huff-n-puff process recovered an for the second one, and so on. To take this into account, it is as-
additional 9.8% IOIP. It is to be noted that the first two huff-n-puff sumed that only half the total injected gas is purchased. Thus, an
cycles produced comparable amounts of oil (4.0 and 5.7% IOIP, average gas utilization factor is defined as half of the total amount
respectively) while the third cycle produced very little oil (0.2% of the injected gas volume (corrected to standard conditions) di-
IOIP). vided by the additional oil recovered.
30 Journal of Canadian Petroleum Technology
 f) The pressure variation of the viscosities of the undersatu-
rated liquid phases followed similar trends, even though
they had different compositions.
2. The properties of the oil saturated with CO2 at progressively
decreasing pressures, simulating a puff cycle, indicated that:
a) The oil phase retained CO2 during the puff stage and more
readily liberated methane;
b) The oil retained its high swelling and formation volume
factor well into the puff stage;
c) The densities of the oil phase decreased initially with de-
creasing pressure, and then increased up to the dead oil
value at atmospheric pressure;
d) As observed during the huff cycle, the undersaturated mix-
tures showed almost the same rate of density increase with
pressure, i.e., had similar isothermal compressibilities;
e) The oil phase retained its low viscosity well into the puff
FIGURE 6: Comparison of gas concentrations between PVT gas cycle, probably due to the retention of CO2 and high
cap and coreflood produced gas. GORs; and,
f) The pressure variation of the viscosities of the undersatu-
The overall gas utilization factors are included in Table 7 for the rated liquid phases followed similar trends, even though
three runs: Run 1 consumed the most gas (3,105 sm3/m3), while they had different compositions.
Runs 2 and 3 consumed considerably less gas (881 and 508 sm3/ 3. The recovery of additional oil by huff-n-puff gas injection
m3, respectively). Thus, the most efficient gas utilization was using improved with increasing residual oil saturation.
the flue gas huff-n-puff process, and the least efficient one was by 4. Different application schemes of the huff-n-puff process re-
injection of a relatively large CO2 slug followed by pressure deple- sulted in very different amounts of gas utilized to produce
tion and chase water. As mentioned earlier, the high gas utilization the oil; ranging from 3,105 sm3/m3 (17.2 Mscf/STB) down to
in Run 1 may be attributed to the low oil saturation prior to en- 508 sm3/m3 (2.8 Mscf/STB).
hanced oil recovery, which makes the tertiary oil less mobile. 5. High enhanced oil recoveries were associated with high ini-
tial waterflood residual oil saturations.
Produced Gas Composition 6. Enriched flue gas performed as well as or better than pure
carbon dioxide.
The composition of the gas produced during the puff cycles was
determined by gas chromatography. During the huff-n-puff core-
flood, the pressures of the first puff cycle were reduced from about
14 MPa to approximately 11 MPa (note that the saturation pressure
Acknowledgements
of the live oil was 10.2 MPa). The ranges of pressure reduction in The authors acknowledge financial support from the Petroleum
Runs 1 and 2 are similar. Therefore, the average concentrations Technology Research Centre and the participating oil companies,
of major gas components in the first puff cycle of these runs are which include Husky Energy Inc., Nexen Canada Ltd., Petrovera
compared with the results obtained at equivalent conditions in the Resources, and Talisman Energy Inc. The authors wish to express
phase behaviour study. As plotted in Figure 6, the concentrations of their thanks to Nexen Canada Ltd. for providing the oil, gas, and
the PVT puff process are approximated by the gas cap at a satura- brine samples. Thanks are also extended to H. Jonasson of Nexen
tion pressure of 12.6 MPa, which is between the starting and final Canada Ltd. for providing data and discussions, to B. Schnell, R.
pressures of the puff cycles. The concentrations of the major gas Exelby, and R. Shi for carrying out the experimental measure-
components are fairly similar in all three cases. This indicates that ments, and to B. Tacik in the preparation of the manuscript. Finally,
the oil preferentially retains CO2 and releases methane on the puff we appreciate the permission granted by Saskatchewan Research
cycle in the dynamic coreflood situations, as was previously ob- Council to publish this paper.
served in the phase behaviour cells.
NOMENCLATURE
BPR = Back-pressure regulator
Conclusions EOR = Enhanced oil recovery
Based on the results of this study, the conclusions can be sum- FVF = Formation volume factor
marized as follows: GOR = Gas-oil ratio
IOIP = Initial oil-in-place
1. The properties of the oil saturated with CO2 at progressively
increasing pressures, simulating a huff cycle, showed that: IWF = Initial waterflood
Psat = Saturation pressure
a) CO2 dissolved readily in the oil, reaching about 36.2 mol%
PV = Pore volume
at 15.7 MPa, the maximum CO2 saturation pressure;
PVT = Pressure volume temperature
b) The formation volume factor and swelling factor of the SF = Swelling factor
oil-CO2 mixture increased almost linearly with saturation
S(ro-h-n-p) = Oil saturation after huff-n-puff process
pressure, reaching values of 1.35 and 1.13 m3/m3, respec-
S(ro-iwf) = Oil saturation after initial waterflood
tively, at 15.7 MPa;
WF = Waterflood
c) The density of the saturated mixture increased with in-
creasing saturation pressure;
REFERENCES
d) The undersaturated mixtures showed almost the same rate
of density increase with pressure, i.e., had similar iso- 1. SIMPSON, M.R., The CO2 Huff ’n’ Puff Process in a Bottomwater-
thermal compressibilities; Drive Reservoir; Journal of Petroleum Technology, Vol. 27, No. 7, pp.
887–893, 1988.
e) As a result of CO2 dissolution into the oil, the viscosities 2. JOHNSON, H.R., SCHMIDT, L.D., and THRASH, L.D., A Flue Gas
of the saturated mixtures decreased with saturation pres- Huff ’n’ Puff Process for Oil Recovery From Shallow Formations;
sure from 1.7 mPa⋅s for the recombined oil down to 1 paper SPE 20269, presented at the SPE/DOE 7th Symposium on En-
mPa⋅s at 15.7 MPa; and, hanced Oil Recovery, Tulsa, OK, April 22 – 25, 1990.
February 2006, Volume 45, No. 2 31
 3. DOE Project, CO2 Huff-n-Puff Process in a Light Oil Shallow Shelf
Carbonate Reservoir; Quarterly Technical Progress Report, 3rd Qtr.,
No. DE-FC22-94BC14986, 1996.
Authors’ Biographies
4. KARIM, F., BERZINS, T.V., SCHENEWERK, P.A., BASSIOUNI, Yiping Zhang is a senior research engineer
Z.A., and WOLCOTT, J.M., Light Oil Recovery From Cyclic CO2 in the Energy Division of the Saskatch-
Injection: Influence of Drive Gas, CO2 Injection Rate, and Reservoir
ewan Research Council (SRC) in Regina,
Dip; paper SPE 24336, presented at the SPE Rocky Mountain Re-
gional Meeting, Casper, WY, May 18 – 21, 1992. Canada. He is working on several miscible/
5. SHAYEGI, S., JIN, Z., SCHENEWERK, P.A., and WOLCOTT, J.M., immiscible gas and chemical enhanced oil
Improved Cyclic Stimulation Using Gas Mixtures; paper SPE 36687, recovery projects. Before joining SRC, he
presented at the SPE Annual Technical Conference and Exhibition, was a reservoir engineering consultant at
Denver, CO, October 6 – 9, 1996. United Oil & Gas Consulting Ltd. As part
6. MELEAN, Y., BUREAU, N., and BROSETA, D., Interfacial Effects of his master’s thesis research at the Uni-
in Gas-Condensate Recovery and Gas Injection Processes; paper SPE versity of Calgary and later at the Petro-
71495, presented at the SPE Annual Technical Conference and Exhi- leum Recovery Institute, Calgary, he was
bition, New Orleans, LA, September 30 – October 3, 2001.
involved in the study of foamy oil behaviour in cold heavy oil pro-
7. Saskatchewan Industry and Resources, Misc. Rept. 2003-1: Reser-
voir Annual 2002, Geo-Publications, Regina, SK, 2003. duction. Mr. Zhang is a member of APEGS, SPE, the Petroleum
8. SAYEGH, S.G. and MAINI, B.B., Laboratory Evaluation of the CO2 Society, and the Canadian Heavy Oil Association.
Huff-n-Puff Process for Heavy Oil Reservoirs; Journal of Canadian
Petroleum Technology, Vol. 23, No. 3, pp. 29-36, May – June 1984. Selim Sayegh is a principal research en-
9. SHAYEGI, S., SCHENEWERK, P.A., and WOLCOTT, J.M., En- gineer with the Saskatchewan Research
hancement of Residual Oil Recovery Using a Mixture of Nitrogen Council. He obtained his M.Eng. and
or Methane Diluted With Carbon Dioxide in a Single-Well Injection Ph.D. degrees in phase behaviour thermo-
Process; United States Patent 5,725,054, March 10, 1998. dynamics from McGill University. He has
10. DYER, S.B., HUANG, S.S., FAROUQ ALI, S.M., and JHA, K.N.,
accumulated over 20 years of industrial
Phase Behaviour and Scaled Model Studies of Prototype Saskatch-
ewan Heavy Oils With Carbon Dioxide; Journal of Canadian experience at the Petroleum Recovery In-
Petroleum Technology, Vol. 33. No. 8, pp. 42-48, 1994. stitute, as an independent consultant in Cal-
11. FONG, D.K.S. and NGHIEM, L.X., A Viscosity Model for Reservoir gary, at Saudi Aramco’s R&D Centre, and
Fluids; Computer Modelling Group Research Report R7.02, March at SRC. His areas of expertise include the
1980. measurement and modelling of the phase
behaviour of oil and gas fluids, improved oil recovery, and gas
hydrates. Mr. Sayegh has published over 20 refereed papers in the
Provenance—Original Petroleum Society manuscript, Laboratory technical literature, and has co-authored 50+ papers at conferences
Investigation of Enhanced Light-Oil Recovery by CO2/Flue Gas Huff- and workshops worldwide. Mr. Sayegh is a member of APEGS, the
n-Puff Process (2004-021), first presented at the 5th Canadian International Petroleum Society, and SPE.
Petroleum Conference (the 55th Annual Technical Meeting of the Petroleum
Society), June 8 - 10, 2004, in Calgary, Alberta. Abstract submitted for Sam Huang is the manager of the En-
review November 13, 2003; editorial comments sent to the author(s) April hanced Recovery Technologies section,
25, 2005; revised manuscript received June 27, 2005; paper approved for with the Energy Division of the Saskatch-
pre-press June 30, 2005; final approval January 19, 2006. ewan Research Council (SRC) in Regina.
He is also an adjunct professor with the
Faculty of Graduate Studies and Research
at the University of Regina. Before joining
SRC, he held a research scientist position
with Gulf Canada in the Sheridan Park Re-
search Centre, Ontario, where he was in-
volved in several hydrocarbon and CO2
miscible EOR projects. Prior to this, he was a research engineer in
the Chemical Research Labs of CIL Inc. and also with D.B. Rob-
inson’s group at the University of Alberta. Dr. Huang did under-
graduate work in chemical engineering at the Taipei Institute of
Technology in Taiwan, and holds a Ph.D. degree in physical chem-
istry from Marquette University in Wisconsin. He has published
more than 40 technical papers on petroleum phase behaviour, en-
hanced oil recovery, and electron and ionic transport behaviour in
hydrocarbons and noble gases.

Mingzhe Dong is an associate professor

of petroleum systems engineering in the
Faculty of Engineering at the University
of Regina. He holds a B.A.Sc. degree from
Northwestern University, Xi’an, an M.A.Sc.
degree from the University of Petroleum,
Beijing, and a Ph.D. degree from the
University of Waterloo, Ontario, all in
chemical engineering. His research interests
include multiphase flow in porous media,
enhanced oil recovery, phase behaviour in
gas injection, reservoir simulation, and surface phenomena in oil
recovery processes. He is a member of the Petroleum Society, SPE,
ACS, and APEGS, and is a registered professional engineer in the
province of Saskatchewan.

32 Journal of Canadian Petroleum Technology