SPE-190285-MS

Immiscible Nitrogen Flooding in Bentheimer Sandstones: Comparing Gas

Injection Schemes for Enhanced Oil Recovery

Martijn T. G. Janssen, Fardin Azimi, and Pacelli L. J. Zitha, Delft University of Technology

Copyright 2018, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Improved Oil Recovery Conference held in Tulsa, Oklahoma, USA, 14-18 April 2018.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Gas injection is a widely applied enhanced oil recovery method. However, poor vertical and areal sweep
efficiency result in inefficient oil displacement. For improving gas mobility control, Water-Alternating- Gas
injection has often been applied. The goal of this study was to compare several immiscible nitrogen injection
schemes and to investigate how rock-fluid and fluid-fluid interactions control the immiscible flooding
process. Well-controlled core-flood experiments were performed in Bentheimer sandstone cores. Nitrogen
was injected into cores saturated with n-hexadecane at connate water saturation at constant pressures (5 and
10 bar) and while varying backpressure (5 to 60 bar). Nitrogen was also injected at residual oil to waterflood
and a Water-Alternating-Gas injection scheme was assessed. Coreflood results clearly demonstrated the
beneficial effects of Water-Alternating-Gas injection over continuous gas injection. The findings in this
study suggest that a) an increase in pressure favours oil recovery slightly during continuous nitrogen
injection at connate water saturation, b) residual oil saturation for immiscible nitrogen flooding is lower
under three-phase flow compared to two-phase flow and c) the relatively high oil recovery, i.e. lower
ultimate residual oil saturation, by Water-Alternating- Gas injection is most likely related to an increase in
trapped gas saturation.

Introduction
The global oil demand will most likely outrun global oil production in the near future as the overall oil
demand increases annually whilst oil production from many giant, mature, oil fields worldwide is declining
(Hook et al., 2009). In addition to the less likely probability of discovering new large oil fields, this
underlines the importance of developing novel techniques that enhance oil production from mature fields:
enhanced oil recovery (EOR) techniques.
One of the most widely applied EOR processes is gas flooding (Orr, 2007). In this technique natural or
non-natural gas (e.g. nitrogen (N2) or carbon dioxide (CO2)) is injected into a (mature) oil field in order to
increase oil recovery over water flooding. The process might be miscible or immiscible, depending on type
of injected gas, hydrocarbon components of the oil in place (OIP) and on reservoir pressure and temperature.
Gasses that have low minimum miscibility pressures (MMP) (e.g. CO2), oils that contain mainly light
2 SPE-190285-MS

hydrocarbon components and high reservoir pressures all favour miscibility between injected gas and OIP.
During miscible flooding injected gas and displaced oil form a single
fluid phase, leading to oil swelling and/or viscosity reduction thus promoting oil displacement (Lake,
1989). Under immiscible conditions, gas dissolves only partly in the oil so that rock-fluid and fluid-fluid
interactions, as described by relative permeabilities and capillary pressures, control oil displacement.
Gas can be injected either in a continuous mode or in a Water-Alternating-Gas (WAG) mode. For any
continuous gas injection unfavourable mobility ratios between the injected gas and the displaced fluids is a
major concern as it results in an unstable displacement. Viscous fingering, channeling in high permeability
streaks and gravity segregation are common phenomena in gas flooding, leading to early gas breakthrough
(Rossen et al., 2006; Farajzadeh et al., 2009; Farajzadeh et al., 2010; Andrianov et al., 2011).
WAG, i.e. a process where gas slugs are alternated by water slugs, can be applied for controlling gas
mobility thus improving gas sweep efficiency (Hallam et al., 1995). By periodically shifting between slugs
of gas and water, gas breakthrough can be delayed and hydrocarbon contact time increased. Nevertheless,
gravity segregation may also occur during WAG yielding a reduced sweep efficiency (Andrianov et al.,
2011). Poor fluid injectivity, i.e. extremely high pressure drops, is another major management challenge
concerning WAG field applications (Rogers and Grigg, 2000).
Several studies have shown that immiscible gas/WAG injection is a viable EOR technique when using
air (Dong et al., 2005), CO2 (Zhang et al., 2010) or N2 (Bhoendie et al., 2014) as a gas phase. However, a
knowledge gap exists regarding the displacement mechanisms for immiscible gas flooding methods. The
goal of this study was to investigate how rock-fluid and fluid-fluid interactions control oil displacement
in immiscible gas flooding processes. To this end several well-controlled core-flood experiments were
conducted using various injection schemes: a) continuous nitrogen injection and b) WAG injection. X-
ray Computed Tomography (CT) images were taken during water and gas propagation to map the phase
saturation distributions over time and to gain insight in the responsible displacement mechanism. The paper
proceeds with the following sections: Materials and Methods, Results and Discussion and Conclusions.

Materials and Methods

Chemicals
The oil used to conduct the core-floods was n-hexadecane (CH3(CH2)14CH3, ≥ 99% purity from Merck) with
a density of 0.775±0.001 g/cm3 and a viscosity of 3.37±0.06 cP (at 20 °C). A red colorant (C26H24N4O, from
Sigma-Aldrich) was added to the oleic phase (<0.006 wt%) for visualisation purposes. Brine was prepared
by dissolving sodium chloride (NaCl, ≥ 99% purity from Merck) in demineralized water (3 wt%). Its density
and viscosity equalled 1.020±0.001 g/cm3 and 1.07±0.05 cP (at 20 °C) respectively. The gas used in the
core-flood experiments was N2, with a density and viscosity of 1.165×10− g/cm3 and 1.76×10−2 cP (at 20
°C and atmospheric pressure) respectively. In two experiments the oleic and aqueous phases were doped
with 1-iododecane (CH3(CH2)9I, 98% purity from Sigma- Aldrich) and potassium-iodide (KI, ≥ 99% purity
from Sigma-Aldrich) respectively for enhancing the CT contrast. Brine doped with 7.5 wt% and 15 wt% KI
had densities of 1.078±0.001 g/cm3 and 1.142±0.001 g/cm3 respectively. Their viscosities were 0.97±0.08
cP (7.5 wt% KI) and 1.01±0.06 cP (15 wt% KI). The model oil was doped with either 5 wt% or 7.5 wt%
1-iododecane. Their respective densities and viscosities were 0.790±0.001 g/cm3 and 3.22±0.07 cP (5 wt%
1-iododecane) and 0.798±0.001 g/cm3 and 3.19±0.06 cP (7.5 wt% 1-iododecane).

Core samples
Bentheimer outcrop samples are often used for laboratory studies as model sandstone as they are rougly
homogeneous, fairly clean (>91 wt% Quartz) and have relatively high permeabilities (2.6±1.2 Darcy) (Peksa
et al., 2015). Table 1 presents the physical properties of the Bentheimer sandstone cores used in this study.
The porosities were measured using gravimetric methods, an Ultra Pycnometer and by processing obtained
SPE-190285-MS 3

CT data. The discrepancy in core length is due to CT limitations as 40.0 cm cores do not fit vertically in the
medical CT scanner. Core samples were drilled out of outcrop blocks and subsequently dried in an oven at
60 °C for 48 hours. Afterwards, the cutted cores were casted in an Expoxy resin to avoid bypassing flow
alongside the core. Five equidistant holes were drilled in all the cores for pressure(drop) measurements.

Table 1—Properties of Bentheimer sandstone cores used in the core-flood experiments. Five different experiments
were conducted: a) continuous gas injection at varying pressure (CGI VP), b) continuous gas injection at a
constant pressure of 5 bar (CGI CP 5 bar), c) continuous gas injection at a constant pressure of 10 bar (CGI CP
10 bar), d) gas flooding subsequent to water flooding (WF + CGI) and e) Water-Alternating-Gas (WAG) injection.

1-CGI VP 2-CGI 3-CGI CP 4-WF + CGI 5-WAG

CP 5 bar 10 bar
Porosity (%) 24.40 ± 0.10 24.30 ± 0.10 22.70 ± 0.10 23.10 ± 0.10 23.60 ± 0.10
Permeability
2.88 ± 0.04 2.96 ± 0.09 2.68 ± 0.08 1.94 ± 0.14 2.30 ± 0.18
(Darcy)
Length (cm) 40.00 ± 0.10 40.00 ± 0.10 17.00 ± 0.10 17.00 ± 0.10 17.00 ± 0.10
Diameter (cm) 3.90 ± 0.10 3.90 ± 0.10 3.90 ± 0.10 3.90 ± 0.10 3.90 ± 0.10
Pore volume
116.59 ± 6.87 116.11 ± 6.84 46.10 ± 2.89 46.91 ± 2.94 47.93 ± 3.00
(cm3)

Experimental set-up. Figure 1 presents schematically the set-up used to conduct the core-flood
experiments. The cores were placed in a core-holder made of polyether ether ketone (PEEK) that exhibits
low X-ray attenuation, and were exposed to a confining pressure that equalled the inlet pressure. The core-
holder was placed in line with either a single cylinder syringe pump (1000D Syringe Pump, Teledyne ISCO
series) or, for the WAG experiment, with a dual cylinder liquid pump (Quizix QX-1500 HC). The ISCO
pump was used for injecting both the aqueous and oleic phases, whereas the Quizix pump was only used
for injecting aqueous solutions. Nitrogen gas was supplied from a 200 bar cylinder. A mass flow controller
(Bronkhorst, EL-FLOW) was used to regulate the gas flowrate. A backpressure regulator was connected to
the outlet to control and maintain the outlet pressure during the experiments. CO2, used for initial flushing
of the core, was provided by a 200 bar cylinder. A fraction collector (GE Akta Frac-920) was used to sample
the effluents at the outlet at predetermined time intervals. A total of four differential and two absolute
pressure transducers were placed to monitor the pressure along the core length. A thermocouple was used
to monitor the temperature. The pressure transducers, and one thermocouple, were connected to a data
acquisition system (National Instruments) which recorded pressure and temperature data at a 10 s interval.
For each injection phase (gas, water and oil) a separate injection port was installed on the set-up. Since the
experiments were performed using gravity stable conditions, i.e. core placed vertically, two sets of valves
were installed which allows for changing the injection direction (from top to bottom and from bottom to
top). The experiments were performed at room temperature (20 °C).
4 SPE-190285-MS

Figure 1—Schematic of the experimental set-up. The arrows indicate flow from
top to bottom. Note that the flow direction can be changed by switching valves 9.

CT scanning. For two experiments, cores were CT scanned to gain insight into phase saturation
distributions during the displacement processes. To this end a Siemens SAMATOM Definition CT scanner
was used. One CT scan contained 12 slices of 4 mm thick, each containing 512×512 pixels with a pixel
size of 0.2×0.2 mm. Scanning was done in sequential mode using an energy of 140 KeV (250 mA). The
acquired data was analyzed using both ImageJ and Matlab™ software. Table 2 gives an overview of the
CT settings used.

Table 2—Settings CT scanner.

Specification

Tube voltage (KeV) 140

Tube current (mA) 250

Slice thickness (mm) 4

Pixel size (mmxmm) 0.2*0.2

Scan mode Sequential

Experimental procedure. Table 3 presents an overview of the gas injection schemes investigated in this
study. During the first three experiments (continuous gas injection at varying pressure (CGI VP), at a
constant pressure of 5 bar (CGI CP 5 bar) and at a constant pressure of 10 bar (CGI CP 10 bar)) pressure
conditions were varied to study the effect of system pressure on oil recovery by immiscible N2 flooding.
SPE-190285-MS 5

In the next two experiments N2 was injected in a continuous manner after water flooding (WF + CGI) and
in a WAG mode (WAG).
Note that for experiments 4 and 5 there is a difference in the added amounts of dopant to both the oleic
and aqueous phase (Table 3). Based on the results of the dopant calibration and the CT imaging observations
made during core-flood 4 (WF + CGI), it was decided to use dopant concentrations of 5 wt% and 15 wt
% for the oleic and aqueous phases respectively in experiment 5 (WAG). The applied change in dopant
concentrations revealed better contrasts between the various phases on CT images.

Table 3—Overview of the experiments conducted. In the first three experiments N2 flooding was initiated
at connate water saturation whereas in the fourth experiment gas flooding was applied at residual
oil to waterflood. Water-Alternating-Gas (WAG) injection took place at connate water saturation.

Experiment Procedure Gas flow rate Liquid flow WAG ratio Backpressure (bar) CT
(cm3/min) rat (cm3/min) (water-gas ratio)
0, 5, 10, 20,
1-CGI VP N2 flooding vary* – – No
40 and 60
2-CGI CP 5 bar N2 flooding 1.0 – – 5 No
3-CGI CP 10 bar N2 flooding 0.5 – – 10 No
Yes Oil – 7.5
Water flooding wt% dopant
4-WF + CGI 0.5 2.0 – 5
N2 flooding Brine – 7.5
wt% dopant
Yes Oil – 5
Water-Alternating- wt% dopant
5-WAG 0.5 2.0 1:6 5
Gas injection Brine – 15
wt% dopant

*In experiment 1 the N2 injection rate was held constant at 5 mljmin, yielding local flowrates varying from 5 ml/min (at atmospheric pressure) to 0.083 ml/
min (at 60 bar).

The experimental procedure (Table 4) started with flushing the core with CO2 for two hours at an injection
pressure of 5 bar to remove all the air inside the system. Afterwards approximately 10 pore volumes (PV)
of 3 wt% NaCl brine was injected. During brine saturation the backpressure was increased to 25 bar to
ensure complete dissolution of CO2 in brine. By varying brine injection rates the average permeability of
the core used was computed using Darcy's law (Darcy, 1856). Subsequently, primary drainage was initiated
by injecting approximately 6 PV of n-hexadecane. To obtain the oil end-point relative permeability (kro) the
flowrates were varied during the last PV of oil injection. At this moment connate water saturation (Swc) was
reached; starting point for the EOR processes. Following primary drainage, for experiment 1 (CGI VP), N2
was injected in a continuous mode at atmospheric pressure and subsequently at 5, 10, 20, 40 and 60 bars
once steady state pressure conditions were observed. For experiments 2 (CGI CP 5 bar) and 3 (CGI CP
10 bar) N2 was injected, at Swc, at constant pressures of 5 and 10 bars respectively. During test 4 (WF +
CGI) N2 flooding took place following water flooding (6 PV) whilst in core-flood 5 (WAG) a total of 12
WAG cycles (each cycle consisted of injecting a N2 slug (1.30 PV) followed by a water slug (0.22 PV))
were injected at Swc. In order to operate under gravity stable conditions the core was placed vertically and
water was injected from bottom to top whilst oil and gas were injected from top to bottom (for WAG both
water and gas injection took place from top to bottom). For each experiment the EOR process continued
until no more oil was produced. The corefloods conducted are analyzed in terms of oil recovery, pressure
data, CT images (if available) and saturation profiles.
6 SPE-190285-MS

Table 4—Sequence of experimental procedure.

Step Experiment Description Backpressure Flow rate Injection Flow direction

(bar) (cm3/min) pressure (bar)

1 1,2,3,4 and 5 CO2 flushing – – 5 Down

Brine saturation
2 1,2,3,4 and 5 25 2.0 – Up
(3 wt% NaCl)

3 1,2,3,4 and 5 Oil injection 0 2.0 Test 4: 5.0 – Down

Water flooding
4 4 0 2.0 – Up
(3 wt% NaCl)

Test 1:0, 5, 10, 20,

Test 1: vary* Test 2:
5 1,2,3 and 4 Gas flooding 40 and 60 Test 2 – Down
1.0 Test 3 and 4: 0.5
and 4: 5 Test 3: 10

WAG injection
6 5 5 Gas: 0.5 Water: 2.0 – Down
(3 wt% NaCl)

*In experiment 1 the N2 injection rate was held constant at 5 mijmin, yielding local flowrates varying from 5 ml/min (at atmospheric pressure) to 0.083 ml/
min (at 60 bar).

CT scans were taken during core-flood experiments 4 and 5 to visualize and gain insight into the
distribution of phase saturations. The acquired CT scanning attenuation data were expressed in Hounsfield
units. Porosity and two-phase saturations were computed from the measured Hounsfield units using the
formulae presented in Table 5 (Sharma et al., 1997).

Table 5—Equations used for processing CT scan images. CTwet, CTdry, CTbrine, CTair and CTswc represent the CT response in
Hounsfield units of the brine saturated core, dry core, water phase, air phase and oil saturated core respectively. Swc is the
connate water saturation based on material balance calculations. The equations were adopted from Sharma et al. 1997.

Parameter Formula

Porosity

Two-phase oil
saturation (So)

Results and discussion

The main results of the performed core-flood experiments are shown in Table 6. The obtained end-point
relative permeabilities are in good agreement with the findings of Treiber et al. (1972) for consolidated
water-wet porous media, although slight differences between individual experiments (i.e. cores) were
observed.
This section proceeds with the following sections: primary drainage and forced imbibition, continuous N2
flooding at connate water saturation, continuous N2 flooding at residual oil to waterflood, Water-Alternating-
Gas (WAG) injection at connate water saturation and oil recovery. Pressure drop profiles, oil recovery plots
and CT data are analyzed in order to assess the various injection schemes.
SPE-190285-MS 7

Table 6—Summary of the core-flood experiments conducted. kro, krw, Swc, Soi, Sor_WF, Sor_GF, RFWF and RFGF represent
the oil end-point relative permeability, water end-point relative permeability, connate water saturation,
initial oil saturation, residual oil saturation to waterflood, residual oil saturation to N2 flood/WAG, recovery
factor corresponding to water flooding and recovery factor corresponding to N2 flooding/WAG respectively.

Experiment 1 2 3 4 5

kro* 0.53 ± 0.03 0.53 ± 0.03 0.48 ± 0.05 0.65 ± 0.07 0.60 ± 0.05

krw* – – – 0.14 ± 0.01 –

Swc 0.22 ± 0.02 0.21 ± 0.02 0.30 ± 0.02 0.24 ± 0.02 0.26 ± 0.02

Soi 0.78 ± 0.02 0.79 ± 0.02 0.70 ± 0.02 0.76 ± 0.02 0.74 ± 0.02

Sor_WF – – – 0.49 ± 0.02 –

RFWF (% of OIIP) – – – 36 ± 4 –

0.42 ± 0.02
Sor_GF (atmospheric P) 0.39 0.40 ± 0.02 0.35 ± 0.02 0.36 ± 0.02 0.30 ± 0.02
± 0.02 (increasing P)

46 ± 4 (atmospheric
RFGF (% of OIIP) P) 50 ± 4 49 ± 4 50 ± 4 53 ± 4 59 ± 4
(increasing P)

Primary drainage and forced imbibition

This section discusses the pressure drop readings and corresponding oil saturation profiles for the primary
drainage and forced imbibition stages in experiments 4 and 5, which are representative for all the
experiments.
Pressure drop. The pressure drop profiles obtained during primary drainage (tests 4 and 5) and forced
imbibition (test 4) are shown in Figure 2. Although there was a difference in oil injection rate (5.0 cm3/
min in test 4 and 2.0 cm3/min in test 5), both pressure profiles for primary drainage are very similar. Oil
breakthrough happened at approximately 0.70±0.02 PV (test 1), 0.65±0.02 PV (test 2), 0.79±0.02 PV (test
3), 0.76±0.02 PV (test 4) and 0.78±0.02 PV (test 5). Once the oleic phase reached the core inlet, a sharp
increase in pressure drop was observed due to the capillary entry pressure; equivalent to approximately
55±10 mbar for both experiments. Next, the pressure drop increased gradually until it reached a maxiumum
value of 543±5 mbar (experiment 4) and 200±5 mbar (experiment 5) at the respective oil breakthrough
times. Note that the oil injection rate in experiment 4 was more than twice as high as in experiment 5 (Table
4). After oil broke through, the pressure drops decreased towards a steady state value of 370±10 mbar
(experiment 4) and 140±10 mbar (experiment 5) which corresponds to oil flow at Swc.
After injecting oil for approximately 6.71±0.02 PV, water flooding was initiated in experiment 4 at 2.0
cm3/min. Firstly, the pressure drop increased until it reached a maximum value of 202±5 mbar at water
breakthrough (0.45±0.02 PV after water flooding started). Subsequently pressure drop declined slightly to a
steady state value of 180±5 mbar (pressure gradient of 1.06 bar/m), corresponding to solely water production
from the core.
8 SPE-190285-MS

Figure 2—Total pressure drop profiles for experiment 4 (top) and experiment 5 (bottom) prior to gas injection.
Oil represents the primary drainage stage whereas Water refers to water flooding. Oil injection, at 5.0 cm3/
min (test 4) or at 2.0 cm3/min (test 5), was continued for several pore volumes before varying the flowrate
to obtain ko. Next, in experiment 4, water flooding was initiated (at 2.0 cm3/min) after approximately 6.71
PV of oil injected. Once 5.0 PV of water was injected the flowrate was varied in order to compute krw*.

CT images. The CT images and analogous oil saturation (So) profiles for both primary drainage and water
flooding, in experiment 4, are presented in Figure 3. Figure 4 shows related CT images and So profiles during
primary drainage in experiment 5. Note that, due to the vertical position of the core-holder on the couch
of the CT scanner, particularly high beam hardening artifacts were present, in various degrees, in all CT
images taken. The acquired CT data was corrected in order to obtain two-phase saturation distributions. For
each timestep the center slice was loaded into Matlab™, cropped and averaged. Subsequently, the averaged
slice was corrected for the CT artifacts present before calculating two-phase saturation distributions.
During primary drainage water (red/green) was displaced by oil (blue) in a piston-like manner (Figure
3A). At the end of primary drainage, the relatively low So near the outlet region (i.e. bottom) is most-likely
a consequence of the capillary end effect. It implies the accumulation of water (wetting phase) near the
outlet in order to try to satisfy the zero capillary pressure condition at the outlet boundary. It is expected
that, when scanning the core horizontally, the accumulation of the wetting phase is more distinctly present
on CT images (i.e. without any correction applied for beam hardening- and cross-artifacts). Corresponding
So profiles show a sharp shock front region. However, most probably due to the correction applied, a
typical rarefaction wave upstream of the shock front is hard to recognize. Eventually an averaged initial oil
saturation (Soi) of 0.76±0.02 was reached (Figure 3B).
SPE-190285-MS 9

Figure 3—Displacement profiles during primary drainage and water flooding in experiment 4: A) CT
images and B) oil saturation profiles. PV=0 corresponds to the start of the particular injection phase. A:
water (red/green) is displaced by oil (blue) during primary drainage and during water flooding oil (blue)
is produced by water (red/green). B: continuous profiles present primary drainage phase whilst dashed
lines correspond to water flooding. Note that oil was injected from the top and water from the bottom.

CT scan images taken during water flooding show the displacement of the oleic phase (blue) by water
(red/green) (Figure 3A). Corresponding So profiles show a similar trend compared to those for primary
drainage. Nonetheless, the displacement front is less sharp due to capillary forces. After 5.43 PV a residual
oil saturation to waterflood (Sor_WF) of 0.46±0.04 was achieved, which is slightly lower than the value
obtained through material balance (Table 6). The displacement of oil by water is assumed to be piston-like
with an end-point mobility ratio (=(krw*/μw)/(kro*/μo)) of 0.71.
The CT images and So profiles related to primary drainage in experiment 5 (Figure 4) are very similar
to those presented in Figure 3. However, due to changes in dopant concentrations (Table 3), the difference
between the oil and water attenuation coefficients increased. Due to the beam hardening- and cross-artifacts
correction applied the expected relatively low So at the outlet and a rarefaction wave upstream of the shock
10 SPE-190285-MS

front are not observable. Eventually an averaged Soi of 0.84±0.01 was reached, which is nearly 10% higher
than the Soi obtained through material balance (Table 6).

Figure 4—A: CT images taken during primary drainage in experiment 5. PV=0 corresponds
with the start of the oil injection. Water (red/green) is displaced by oil (blue) in a stable
manner. B: related oil saturation profiles. Note that oil was injected from the top.

Continuous N2 flooding at connate water saturation

After primary drainage N2 was injected under gravity stable conditions, i.e. through the top of the core
(experiments 1, 2 and 3 in Table 6). Gas flooding took place at increasing pressures (test 1), at a constant
pressure of 5 bar (test 2) and at a constant pressure of 10 bar (test 3). As corresponding pressure drop profiles
for the three experiments are similar, only the one for experiment 3 is discussed in more detail.
Pressure drop. Figure 5 presents the total pressure drop profile during immiscible gas flooding for
experiment 3. As aforementioned, continuous gas injection was done under gravity stable conditions (i.e.
from the top of the core). The estimated critical injection velocity was approximately 2.0 ft/day, which
equals the used injection rate (Dietz, 1953). As soon as gas entered the core, oil was displaced and produced
from the outlet (i.e. bottom). Due to the production of oil by the moving gas front, the total pressure drop
increased gradually to a maximum value of 97±3 mbar and then it decreased to steady state values of
approximately 85±5 mbar. This value is equivalent to a pressure gradient of 0.50 bar/m. Gas breakthrough
was observed after injecting 0.51±0.02 PV of gas, evident from the fluctuations in pressure drop due to
gas production. In experiments 1 and 2 gas breakthrough occurred at approximately 0.54±0.02 PV and
0.45±0.02 PV respectively. The observed breakthrough times for secondary gas flooding are consistend
with those found by Naylor and Frørup (1989) who studied gravity stable oil displacement by N2 (using a
superficial velocity of approximately 1.0 ft/day) in fairly clean water-wet Aeolian sandstones.
SPE-190285-MS 11

Figure 5—Total pressure drop profile during gas injection at gravity stable conditions in
experiment 3. Gas was injectied at 0.5 cm3/min and the total core length equalled 17.0 cm.

Continuous N2 flooding at residual oil to waterflood

In experiment 4 (WF + CGI) first the core was brought to Sor_WF, then N2 was injected from the top in a
continuous manner at a backpressure of 5 bar. This section elaborates on the related pressure drop profile.
Pressure drop. Figure 6 shows the total pressure drop profile over the entire core during gas flooding (at
0.5 cm3/min) for the first two PV injected. As soon as gas touched the core a gradually increasing trend in
total pressure drop was observed, indicating downward movement of the gas front through the core. During
this stage both oil and water were produced. Once gas breakthrough occurred, at 0.51±0.02 PV, the total
pressure drop reached a maximum of 91±5 mbar after which it slightly reduced to 82±7 mbar (equivalent
to a pressure gradient of 0.48 bar/m). The observed gas breakthrough time is in good agreement with the
findings of Hustad and Holt (1992) who studied the gravity stable displacement of oil by hydrocarbon gas
after water flooding (using superficial velocities of 0.2 and 0.3 ft/day for water and gas respectively).

Figure 6—Total pressure drop profile during N2 injection at 0.5 cm3/min in

experiment 4. Gas was injected from the top to ensure gravity stable conditions

Water-Alternating-Gas injection at connate water saturation

In experiment 5 gas was injected at Soi through a WAG scheme to investigate gas mobility control. In this
core-flood both gas and water were injected from the top. The injected water during the EOR scheme is
similar to the model brine. A total of 12 cycles were injected, each cycle started with approximately 1.34
PV of N2 followed by 0.23 PV of water. The WAG ratio (1:6) selected may be considered to be dry, i.e.
continuous gas injection in which small slugs of water were injected to improve mobility control (Dyer and
12 SPE-190285-MS

Farouq Ali, 1994). The backpressure used equalled 5 bar. In this chapter related pressure drop profile and
CT images are analyzed.
Pressure drop. Figure 7 presents the total pressure drop profile during WAG injection in experiment 5.
Each cycle started with the injection of N2 at 0.5 cm3/min followed by water injection at 2.0 cm3/min. As N2
injection was initiated the same total pressure drop behaviour was observed as in experiment 3, an increase
in pressure drop from 62±2 mbar to 103±2 mbar followed by a slight decrease to 80±5 mbar (i.e. a pressure
gradient of 0.47 bar/m). Gas breakthrough occurred at 0.49±0.02 PV. After injecting approximately 1.34
PV of gas, 0.23 PV of water was injected. The shift from gas to water injection caused the total pressure
drop to increase to 363±2 mbar. This increase is most likely due to a combination of increasing the injection
rate and the increase in trapped gas saturation. At the start of the second WAG cycle (N2 injection) the total
pressure drop decreased to a plateau value of 102±10 mbar, equivalent to a pressure gradient of 0.60 bar/
m. Note that this value is slightly higher than the steady state pressure drop during the first N2 injection
phase. This is due to the combination of trapped gas and newly introduced water (injected at the end of the
previous cycle). Once water was injected near the end of the second cycle the total pressure drop reached
a peak of 709±2 mbar. The increase in pressure drop during water injection in all subsequent cycles is in
good agreement with the work of Dong et al. (2005). Most probably the free (i.e. flowing) gas saturation
decreased whilst the trapped (i.e. non-movable) gas saturation increased, resulting in decreasing gas and
water relative permeabilities.

Figure 7—Total pressure drop profile during WAG injection in experiment 5. Both gas and water
were injected from the top. For the first four cycles the N2 and water injection phases are reported.

CT images. Figure 8 shows several CT images taken during gas flooding in the first WAG cycle. It is
evident from the images that oil (red/green) is being displaced by gas (blue). During this period only oil was
produced from the core. The presence of beam hardening- and cross-artifacts makes it very hard to identify
the propagating gas front. It is for that reason that the acquired CT data can only be used for qualitative
purposes, not for quantifying the three-phase saturation distribution.
SPE-190285-MS 13

Figure 8—CT images taken during WAG injection in experiment 5. PV=0 corresponds
to the start of the first WAG cycle. Oil (red/green) is displaced by gas (blue) which
is injected from the top. Gas breakthrough happened at approximately 0.49±0.02 PV

Oil recovery
This section presents the oil recovery profiles for all the core-floods performed in this study. Note that
the presented profiles were constructed by deducting the inlet and outlet dead volumes (i.e. tubing) from
the produced quantities. Hence the actual recovery profiles slightly differ from the ones shown here.
Nonetheless, the trend and final recovery factors (RF) are representative for each experiment. The RF profiles
for experiments 1,2 and 3 are shown in Figure 9. During test 1 N2 flooding was first performed at atmospheric
pressure. Once no more oil was produced the backpressure was set to 5 bar (at 14.92±0.02 PV in Figure 9)
and subsequently to 10, 20, 40 and 60 bar. The corresponding recovery profiles show that the increase in
pressure yielded a slight jump in recovery. Although conditions are still very well below the expected MMP
of approximately 350 bar (Sebastian and Lawrence, 1992), by increasing the pressure the system gets closer
to its MMP. This may result in a lowering of the gas-oil interfacial tension and therefore an increase in oil
recovery. The RF profiles related to core-floods 2 and 3 are almost identical and no significant difference
was observed. The profiles show that approximately half of the ultimate oil recovery was achieved before
gas breakthrough. The ultimate RF values for experiments 1, 2 and 3, after 16 PV injected, are respectively
50±4%, 49±4% and 50±4% of the oil initially in place (OIIP). Corresponding residual oil saturations (Sor_GF
in Table 6) equalled 0.39±0.02, 0.40±0.02 and 0.35±0.02 for tests 1, 2 and 3 respectively.
14 SPE-190285-MS

Figure 9—Oil recovery factor profiles for the first three core-flood experiments performend: a) continuous gas
flooding at increasing backpressure (red), b) continuous gas flooding at 5 bar backpressure (green) and c)
continuous gas flooding at 10 bar backpressure (black). Note that N2 flooding was initiated at SoJ in all the three
experiments; subsequent to primary drainage. The oil recovery factors shown are fractions with respect to
the OIIP. Eventuallly all three core-floods show similar recoveryfactors after 16 PV injected: 50±1% of the OIIP.

Figure 10—Oil cut, water cut and oil recovery profiles for experiment 4 (A) and experiment 5 (B). 0 PV
corresponds with the start of water injection (A) or with the start of the first WAG cycle (B). All values are
with respect to the OIIP. Both experiments were conducted at 5 bar backpressure. Note that only the first 16
PV are shown (i.e. the first 10 WAG cycles). Water and gas refer to respectively water- and gas flooding (A).

Figure 10 presents the oil cut, water cut and RF profiles for experiment 4 (A) and experiment 5 (B). For
plot A, only oil was produced until water breakthrough occurred. After breakthrough only limited amounts
of oil was produced, increasing RF from 31% to 36% of the OIIP (RFWFin Table 6). Other studies reported
SPE-190285-MS 15

recovery factors for water flooding (at 0.5 cm3/min), when using the same rock type and oil as in this
work, of 44±2% (Simjoo, 2012) and 46±6% (Janssen et al., 2018) of the OIIP. One possible reason for
the relatively low RFWF found here might be the rather high injection rate used (2.0 cm3/min). Probably
some oil was bypassed by the water front and hence not produced. Another important factor is the relatively
low absolute permeability of the core used (Table 1). The smaller pore sizes yield higher capillary forces,
compared to the other four cores, that prevent the oil from being produced. The overall RF after gas flooding
was 53±4% of the OIIP, which implies an incremental oil recovery of 17±8% of the OIIP. This value could
be an overestimation since RFWF was fairly low. We suspect that some mobile oil might have been present
in the core once gas flooding started.
Figure 10B shows corresponding plots for experiment 5. As soon as the first WAG cycle started (through
the injection of N2) only oil was produced up to gas beakthrough. At the end of the first WAG cycle a RF of
approximately 35% of the OIIP was achieved. This result is presumably due to a good contact between the
OIP and injected water and gas phases. After injecting cycles 2 and 3 the RF was increased to respectively
53% and 57% of the OIIP. At the end of the experiment, after 12 WAG cycles were successfully injected,
RF reached 59±4% of the OIIP; equivalent to a Sor_GF of 0.30±0.02 (Table 6). It is evident from the plot
that the first two WAG cycles contributed the most to the ultimate RF. Afterwards the RF-PV injected slope
tends to decrease.

General discussion
In this section the results of our study are re-examined in the light of a conceptual model for the immiscible
oil displacement by gas. We shall develop our arguments relying upon the ternary saturation diagrams as
they provide a schematic way to depict the different gas injection schemes investigated. Figure 11 shows the
saturation paths for experiments 1, 2 and 3. The three saturation paths are rather similar. N2 injection took
place at practically constant Swc, indicating that gas injection essentially resulted in two-phase gas-oil flow.
However, since the used gas was not humidified it is possible that a small amount of water was vaporized
into the gas phase; this seems to be negligible though. Oil recovery through continuous N2 flooding in all
the three core-floods show characteristic features of an immiscible displacement of oil by the propagating
gas front. The difference in pressure appears not to have a significant impact on the oil recovery as all
three experiments show similar recovery factors. In terms of oil displacement mechanisms for the case
of continuous N2 injection, it is expected that oil swelling and oil viscosity reduction are negligible as
the system operated well below its MMP. One potential displacement mechanism responsible for the oil
displacement here is film flow (Khorshidan et al., 2016). Oil may be drained out from the pore throat corners
and consequently spread across the gas- water interface. For film flow to happen the gas phase should be
able to enter a pore throat (i.e. capillary pressure should be high enough) and the spreading coefficient should
exceed a critical value as well. Figure 13 presents a schematic overview of the film flow phenomenon.
Figure 12 shows saturation paths corresponding to experiments 4 (WF + GF) and 5 (WAG). Let us first
focus on experiment 4. After reaching a Soi of 0.76±0.02, water flooding reduces So to 0.49±0.02 (Sor_WF
in Table 6). Only limited amounts of oil were produced after water breakthrough occurred. The ultimate
recovery factor for water flooding (RFWF) was 36±4% of the OIIP. The relatively low RFWF observed might
be due to the rather high injection rate used (2.0 cm /min); allowing fractions of OIP to be bypassed by
the injected water. Higher RFWF values are expected when injecting with a lower superficial velocity and
applying a bump flood (i.e. increasing injection rate with a factor 10 to overcome the capillary end effect)
at the end of water flooding (Simjoo, 2012; Bhoendie et al., 2014; Janssen et al., 2018). This will most
likely result in a more realistic Sor_WF, increasing RFWF. Finally, N2 injection caused RF to increase to a value
of 53±4% of the OIIP, reducing Sor to 0.36±0.02. These observations advocate that Sor_GF is lower, i.e.
RFGF is higher, under three-phase flow conditions compared to two-phase flow conditions. Water was only
produced at the start of gas flooding (Figure 10A), after which its saturation remained constant at a value
16 SPE-190285-MS

slightly higher than Swc. The fairly high incremental oil recovery by gas flooding may also be influenced by
the hypothesis that water flooding did not reached true Sor_WF but a saturation somewhat higher than that.

Figure 11—Saturation paths for experiment 1 (top left), 2 (top right) and 3 (bottom). Sw, So,
Sg, Swc and Sor_GF represent the water, oil, gas, connate water and residual oil to gas flooding
saturations respectively. In all three core-floods N2 was injected in a continuous mode at Swc.

The results of core-flood 5 (right plot in Figure 12) support the abovementioned interpretation that a
lower Sor_GF under three-phase flow compared to two-phase flow is an important oil recovery mechanism for
N2 WAG. As seen in the plot, most of the oil was produced during the first two WAG cycles (RFcycle2=53±4%
of the OIIP). Eventually, after 12 WAG cycles, a final RF of 59±4% of the OIIP was achieved (equivalent
to a Sor_WAG of 0.30±0.02). Most likely the trapped gas saturation (total gas saturation=free gas saturation
+trapped gas saturation) increased drastically during water injection in the first two WAG cycles. This may
yield a relative reduction in free gas saturation, thus a reduced gas relative permeability. A reduced gas
relative permeability favours the oil displacement by gas (more favourable gas-oil mobility ratio) which
most probably accounts for the high RF observed. As more cycles were injected, the trapped gas saturation
increased which resulted in a lower So. The findings for the N2 WAG scheme correspond well with the
studies of Zhang et al. (2010) and Fatemi et al. (2013).
SPE-190285-MS 17

Figure 12—Saturation paths for experiments 4 (left) and 5 (right). Sw, So, Sg, Swc, Sor_WF, Sor_GF, Sor_cyde2 and SorWAG represent
the water, oil, gas, connate water, residual oil to waterflood, residual oil to gasflood, residual oil to the 2nd WAG cycle
and residual oil to WAG saturations respectively. Note that only the first two WAG cycles are shown in the right
diagram. The purple triangle in the right plot represents the three-phase saturation distribution after 12 WAG cycles.

Figure 13—Schematic overview of the film flow process based on Khorshidian et al. (2016). The sketched situation
considers a water-wet system where gas was injected at SoJ (Swc) from the top to ensure gravity stable conditions
(similar to core-floods 1, 2 and 3). At a certain moment the gas phase has entered the pore body (middle sketch)
and once the vertical distance (H) between the gas front and the center of the pore throat is sufficient, the capillary
pressure is high enough for the gas to enter the pore trough the center of its throat (Pc = Pc_front + ΔρgH). If the spreading
coefficient is high enough, oil may spread out over the gas-water interface and can be drained from the throat (right
sketch). Note that capillary continuity is required for oil films to exist. A small pore throat may yield a highly curved
gas-oil interface which might terminate film flow and result in residual oil (right pore throat in the right sketch).

Conclusions
An experimental study of immiscible N2 flooding was conducted using Bentheimer sandstone cores and
a light model oil. Several gas injection schemes were investigated in controlled core-flood experiments
including: 1) continuous N2 injection at increasing backpressures, 2) continuous N2 injection at 5 bar
backpressure, 3) continuous N2 injection at 10 bar backpressure, 4) water flooding followed by continuous
N2 injection, and 5) water-alternating-gas injection. From this study it can be concluded that:
1. The WAG injection scheme provided the largest recovery factor of 59±4% of the OIIP. Most likely the
trapped gas saturation increased drastically at the end of a WAG cycle; during water injection. This
might reduce the free, mobile, gas saturation and therefore the gas relative permeability. Eventually
resulting in a more favourable mobility ratio which most probably accounted for a lower oil saturation
and a higher oil recovery.
2. The obtained results are consistent with the hypothesis that an increment in oil recovery arises mostly
from the fact that the residual oil to gas flooding is lower in three-phase flow conditions compared
to two-phase flow.
18 SPE-190285-MS

3. An increase in pressure might slightly favour oil recovery during continuous nitrogen flooding at
connate water saturation since pressures get closer to the minimum miscibility pressure of the system.
4. Continuous nitrogen injection gave rise to higher ultimate recoveries than water flooding at connate
water saturation.
5. In gravity stable injection conditions, gas does not necessarily breakthrough earlier than water.

Acknowledgement
The core-floods experiments presented in this work were performed at the Reservoir Engineering
Laboratory of Delft University of Technology. The authors acknowledge the technical support of Marc
Friebel (TU Delft), Ellen Meijvogel-de Koning (TU Delft), Michiel Slob (TU Delft) and Jens van der Berg
(TU Delft).

Nomenclature
CT = CT response of the core, HU
CTair = CT response of the gas phase, HU
CTbrine = CT response of the water phase, HU
CTdry = CT response of the dry core, HU
CTswc = CT response of the core at connate water saturation, HU
CTwet = CT response of the brine saturated core, HU
g = gravitational acceleration, L/t2
H = vertical distance between gas front and pore center, L
kro* = oil end-point relative permeability
krw* = water end-point relative permeability
OIIP = oil initially in place, L
OIP = oil in place, L
Pc = capillary pressure, P
Pc_front = capillary pressure at the front, P
PV = pore volume, L3
RF = recovery factor
RFGF = recovery factor to gas flooding
RFWF = recovery factor to water flooding
Sg = saturation of the gas phase
So = saturation of the oil phase
Soi = initial oil saturation
Sor = residual oil saturation
Sor_GF = residual oil saturation to gas flooding
Sor_WF = residual oil saturation to water flooding
Sw = saturation of the water phase
Swc = connate water saturation
Δρ = density difference, m/L
μo = oil phase viscosity, P*t
μw = water phase viscosity, P*t

References
Andrianov, A.I., Liu, M.K., and Rossen, W.R. 2011. Sweep Efficiency in CO2 Foam Simulations With Oil. SPE 142999,
Proc. Of the SPE EUROPEC/EAGE Annual Conference and Exhibition, Vienna, Austria.
SPE-190285-MS 19

Bhoendie, K.S., Moe Soe Let, K.P., Li, T., and Zitha, P.L.J. 2014. Laboratory Evaluation of Gas-Injection EOR for the
Heavy-Oil Reservoirs in Suriname. Society of Petroleum Engineers. doi: 10.2118/171038-MS.
Darcy, 1856. Les fontaines publiques de la ville de Dijon: exposition et application. Victor Dalmont.
Dietz, D.N. 1953. A Theoretical Approach to the Problem of Encroaching and by-passing Edge Water. Akad. van
Wetenschappen, Amsterdam. Proc. V.56-B: 83.
Dong, M., Foraie, J., Huang, S., and Chatzis, I. 2005. Analysis of Immiscible Water-Alternating-Gas (WAG) Injection
Using Micromodel Tests. Journal of Canadian Petroleum Technology, 44(2).
Dyer, S.B., and Farouq Ali, S.M. 1994. Linear Model Studies of the Immiscible CO2 WAG process for Heavy-Oil
Recovery. SPE Reservoir Engineering, 9(2), 107–111.
Farajzadeh, R., Andrianov, A., Bruining, J., and Zitha, P. 2009. Comparative Study of CO2 and N2 Foams in Porous Media
at Low and High Pressure-Temperatures. Ind. Eng. Chem. Res., 48(9), 4542–4552.
Farajzadeh, R., Andrianov, A., and Zitha, P. 2010. Investigation of Immiscible and Miscible Foam for Enhancing Oil
Recovery. Ind. Eng. Chem. Res., 49(4), 1910–1919.
Fatemi, M.S., and Sohrabi, M. 2013. Experimental Investigation of Near-Miscible Water-Alternating-Gas Injection
Performance in Water-Wet and Mixed-Wet Systems. Society of Petroleum Engineers. doi:10.2118/145191-PA.
Hallam, R.J., Ma, T.D., and Reinbold, E.W. 1995. Performance Evaluation and Optimization of the Kuparuk Hydrocarbon
Miscible Water-Alternating-Gas Flood. Geological Society, London, Special Publications, 84(1), 153–164.
Hook, M., Hirsch, R., and Aleklett, K. 2009. Giant oil Field Decline Rates and Their Influence on World oil Production.
Energy Policy, 37(6), 2262–2271.
Hustad, O.S., and Holt, T. 1992. Gravity Stable Displacement of Oil by Hydrocarbon Gas After Waterflooding. Society
of Petroleum Engineers. doi:10.2118/24116-MS.
Janssen, M.T.G., Zitha, P.L.J., and Pilus, R.M. 2018. Oil Recovery by Alkaline-Surfactant-Foam (ASF) Flooding: Effect
of Drive Foam Quality on Oil Bank Propagation. Society of Petroleum Engineers, SPE-190235-MS.
Khorshidan, H., James, L.A., and Butt, D.S. 2016. The Role of Film Flow and Wettability in Immiscible Assisted Gravity
Drainage. International Symposium of the Society of Core Analysts, Colorado, USA.
Lake, LW. 1989. Enhanced Oil Recovery. Prentice-Hall, Englewood Cliffs, New Jersey.
Naylor, P., and Frerup, M. 1989. Gravity-Stable Nitrogen Displacement of Oil. Society of Petroleum Engineers.
doi:10.2118/19641-MS
Orr, F.M. 2007. Theory of Gas Injection Processes. Tie-line Publications, Copenhagen, Denmark.
Peksa, A.E., Wolf, K.H.A., and Zitha, P.L.J. 2015. Bentheimer Sandstone Revisisted for Experimental Purposes. Marine
and Petroleum Geology, 67, 701–719.
Rogers, J.D., and Grigg, R.B. 2000. A Literature Analysis of the WAG Injectivity Abnormalities in the CO2 Process. Society
of Petroleum Engineers, doi:10.2118/59329-MS.
Rossen, W.R., Van Duijn, C.J., Nguyen, Q.P., and Vikingstad, A.K. 2006. Injection Strategies to Overcome Gravity
Segregation in Simultaneous Gas and Liquid Injection into Homogeneous Reservoirs. Society of Petroleum Engineers,
doi:10.2118/99794-PA.
Sebastian, H.M., and Lawrence, D.D. 1992. Nitrogen Minimum Miscibility Pressures. SPE and US Department of Energy,
SPE/DOE 24134.
Sharma, B., Brigham, W.E., and Castanier, L.M. 1997. A Report on CT-imaging Techniques for Two-Phase and Three-
Phase In-Situ Saturation Measurements. Prepared for US Department of Energy, Stanford University.
Simjoo, M. 2012. Immiscible Foam for Enhancing Oil Recovery. PhD dissertation. Delft University of Technology.
Treiber, L.E., Archer, D.L., and Owens, W.W. 1972. A Laboratory Evaluation of the Wettability of Fifty Oil Producing
Reservoirs. SPE Journal, 12(6), 531–540.
Zhang, Y.P., Sayegh, S.G., Luo, P., and Huang, S. 2010. Experimental Investigation of Immiscible Gas Process
Performance for Medium Oil. Journal of Canadian Petroleum Technology, 49(2).