nanomodel visualization of fluid injections in tight ... visualization of...and the environmental,...
TRANSCRIPT
TSpace Research Repository tspace.library.utoronto.ca
Nanomodel visualization of fluid injections in
tight formations
Junjie Zhong, Ali Abedini, Lining Xu, Yi Xu, Zhenbang Qi, Farshid Mostowfi, and David Sinton
Version Post-print/accepted manuscript
Citation (published version)
Zhong, J., Abedini, A., Xu, L., Xu, Y., Qi, Z., Mostowfi, F., & Sinton, D. (2018). Nanomodel visualization of fluid injections in tight
formations. Nanoscale. Doi: 10.1039/c8nr06937a
How to cite TSpace items
Always cite the published version, so the author(s) will receive recognition through services that track
citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace because you cannot access the published version, then cite the TSpace version in addition to the published
version using the permanent URI (handle) found on the record page.
This article was made openly accessible by U of T Faculty.
Please tell us how this access benefits you. Your story matters.
Journal Name
ARTICLE
This journal is © The Royal Society of Chemistry 20xx J. Name. , 20xx , 00 , 1-6 | 1
Please do not adjust margins
Please do not adjust margins
Received 00th January 20xx,
Accepted 00th January 20xx
DOI: 10.1039/x0xx00000x
www.rsc.org/
Nanomodel visualization of fluid injections in tight formations†
Junjie Zhonga, Ali Abedinib, Lining Xua, Yi Xua, Zhenbang Qia, Farshid Mostowfic, David Sintona,*
The transport and phase change of a complex fluid mixture under nanoconfinement is of fundamental importance in
nanoscience, and limits the recovery efficiency from tight oil reservoirs (< 10%). Herein, through experiments and supporting
theory we characterize the transport and phase change of a nanoconfined complex fluid mixture. Our nanofluidic platform,
nanomodel, replicates shale reservoirs in terms of mean pore size (~100 nm), permeability (~µD) and porosity (~10%). We
screen conditions for the most promising shale EOR strategies, directly quantifying their pore-scale efficiency and underlying
mechanisms. We find that immiscible gas (N2) flooding presents a prohibitively large capillary pressure threshold (~ 2MPa).
Miscible (CO2) gas flooding eliminates this threshold leading to film-wise stable oil displacement with high recovery
efficiency. Strong capillary forces present barriers as well as opportunities for recovery strategies unique to nanoporous
reservoirs by transitioning from a miscible to an immiscible condition locally within the reservoir. These results quantify the
fundamental transport and phase change mechanisms applicable to nanoconfined complex fluids, with direct implications
in unconventional oil as well as nanoporous media more broadly.
1. Introduction
Transport and phase change in nanoporous media are of funda-
mental importance in a range of applications including fuel
cells1, water treatment2, solar steam production3, electrocata-
lytic conversion of CO24, as well as natural gas and oil in nanopo-
rous reservoirs5. Tight oil and shale gas recovered from shale
formations, have reshaped the global energy market6 (e.g., tight
oil accounts for ~50% of cumulative US 2017 domestic oil pro-
duction)7, with broad economic and environmental impacts as
well as geopolitical implications worldwide. Because of the mas-
sive scale of this process, the impacts on CO2 are globally signif-
icant and diverse. For instance, increased production of shale
gas has led to a reduction in more carbon intensive coal-based
electricity production, with significant climate benefit. The CO2
emitted from natural gas is only 45% of the CO2 emissions of
coal in generating the same amount of electrical energy8, and is
thus regarded as a bridge energy to future carbon-free energy
by slowing the rise in global CO2 emissions. At the same time,
the sheer volume of shale resources – including both natural gas
and oil – represents a massive CO2 source and associated cli-
mate threat. The extensive water and pumping energy used in
hydraulic fracturing, as well as methane leakage from aban-
doned wells represent additional environmental concerns9, 10,
particularly as the current energy return is well below potential.
For these reasons, the field is in search of enhanced oil recovery
(EOR) strategies. One approach of particular interest is CO2 in-
jection, which can in principle achieve a combination of more
efficient resource recovery while sequestering CO2. This process
is well-established in conventional oil and gas operations, and
constitutes the single largest global industrial use of CO2 which
would otherwise be emitted11. Compared to conventional res-
ervoirs, however, shale reservoirs typically have nanometer
pores (< 100 nm), low porosity (~10%) and low permeability
(nano Darcy to micro Darcy)12, which significantly limits well
productivity5. At such scales, conventional experimental meth-
ods for porous media are not applicable, and there is a critical
need for methods and fundamental understanding of these
complex highly confined fluid systems.
Numerical simulations13, core flood tests14 and field pilot
tests15 are currently the primary means of studying EOR strate-
gies in unconventional reservoirs. Although these studies have
provided insights into tight oil EOR, there are limitations for
these approaches. For the numerical simulation, a fundamental
challenge lies in the non-applicability of classical fluid mechan-
ics and thermodynamics at relevant nanoscales16 – a challenge
that can only be addressed with experimental validation. In con-
ventional core testing, the resolution of the experimental re-
sults is limited by the opacity of the system (i.e., it is not possible
to directly observe nanoscale fluid behavior within a shale core
sample). The extremely small pore sizes exacerbate this defi-
ciency. A full field pilot is the ultimate test of an EOR strategy,
but this option is very expensive and slow, with effectively low
resolution. In addition, the influence of the strategy under test
can be confounded by external factors within the reservoir.
a. Department of Mechanical and Industrial Engineering, University of Toronto, To-
ronto, ON, M5S3G8 Canada. b. Interface Fluidics Limited, Edmonton, AB, T6G1V6 Canada c. Schlumberger-Doll Research, Cambridge, Massachusetts, 02139 USA. *[email protected]
†Electronic Supplementary Information (ESI) available: [Supplementary information
contains details of the following: Fabrication and characterization of the nano-model, Experimental setup and procedures, Capillarity calculation and Lenormand
phase diagram quantification, Quantification of the miscible film displacement
length and diffusion length in shale reservoirs, N2 flooding at 12 MPa, CO2 diffusion
in the nanomodel during the huff-and-puff injection]. See DOI: 10.1039/x0xx00000x
Journal Name
ARTICLE
2 | J. Name. , 20xx , 00 , 1-5 This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Please do not adjust margins
Field pilots should be reserved as final validations for strategies
optimized through rapid iteration with informative, high resolu-
tion, controlled tests. Recent progress in micro/nanofluidics has
allowed accurate fluid analysis down to a few nanometers, such
as hydrocarbon phase behavior measurements17, 18 and liquid
transport19. While these works provide some fundamental in-
sight into the primary production in shale reservoirs, such meth-
ods have not been applied to the critical challenge of assessing
EOR strategies in unconventional oil recovery – a system with
complex fluid mixture transport and interactions under nano-
confinement.
In this study, a nanofluidic approach is adopted to quantify
transport and phase change of a nanoconfined complex model
hydrocarbon mixture. A nanomodel with 106 2-D nanoarrays are
fabricated, defining a nanoporous system with a pore size of 60 nm
(close to the major pore size in a majority of unconventional oil for-
mations20), while matching the low porosity (here ~14%) and perme-
ability (here ~ 10 µD) of shale. The nanomodel allows direct in-situ
observation of nanopore-scale complex fluid behaviors, enabling
much more rapid evaluation of EOR mechanisms and strategies (a
few hours) compared to core (days) or field tests (months). Using this
approach we screen the most promising EOR strategies in shale res-
ervoirs and detail the fundamental transport mechanisms in each
case. We assess the feasibility of these strategies and quantify per-
formance as a function of injected fluids and running conditions. Im-
portantly, we find a significant initiation pressure threshold for im-
miscible gas flooding that is unique to operations at this scale. Misci-
ble gas flooding can reduce the threshold and improve the recovery
efficiency. In the huff-and-puff processes, the initial injection gas
pressure strongly affects the final recovery efficiency via both gas sol-
ubility and miscibility. A sufficient gas pressure drawdown rate dur-
ing the recovery stage is also essential to limit gas diffusion and gen-
erate recovery via gas breakout. The nanomodel approach here can
also be applied to other nanoporous media applications in energy
and the environmental, including fuel cells , water treatment, oil-re-
pellent membranes21 and CO2 electrocatalysis.
Fig. 1 Nanomodel approach to screening candidate EOR strategies for tight oil: (a) Schematic of horizontal wells and hydraulic fractures for primary tight oil production. (b) On-
chip physical nanomodel of tight oil formations. The scale bar represents 500 µm. (c) SEM and AFM characterizations of the nanomodel. The nanopore dimensions are ~75 nm × 50
nm. The nanomodel porosity and permeability are ~14% and ~10 µD. The scale bars represent 5 µm and 500 nm, respectively. (d) Schematic of nanomodel-based gas flooding with
an injection and a production well. (e) Schematic of nanomodel-based huff-and-puff where gas is first injected into the reservoir (i.e., huff cycle) followed by soak period and
production cycle from same well (i.e., puff cycle). (f) Example of image processing displaying change from original experimental image to the color scale recovery history plot. The
color scale indicates the first instant that a given pore was cleared, and the scale bar represents 100 µm.
Puff
A - A
A - A
(a)
(d)
(b)
(c)
Flooding
Oil
Ga
s
Huff
Gas
Ga
s
Oil
Oil
Soaking
(f)
Tim
e
RGB to BW Stack
(e)
“Shale”
“Fractures”
“Fractures”
0 50 100 1500
30
60
Dep
th (
nm
)
Position (nm)
Journal Name
ARTICLE
This journal is © The Royal Society of Chemistry 20xx J. Name. , 20xx , 00 , 1-5 | 3
Please do not adjust margins
Please do not adjust margins
2. Experimental section
Fractured shale reservoirs have dense matrices of nanopores
connected to large fractures (Fig. 1a). To physically model the key
elements of this scenario, we fabricated a fluidic device with a dense
nanoporous media (1 mm × 1mm, Fig. 1b) connected by two much
larger microchannels (200 µm in width and depth). The hydraulic di-
ameter of the nanopore is ~60 nm (Fig. 1c), representing the typical
oil-rich pore geometry in many shale reservoirs. For example, the
peak pore volume occurs at a pore size of ~100 nm (e.g., Bakken, Ni-
obrara, Wolfcamp and Utica)5. The porosity and permeability of the
nanoporous media are ~14% and ~10 µD (Fig. 1c, see ESI†, section 1),
respectively, which closely match the shale fluid transport properties
after primary production22. Due to the extremely low permeability of
shale, water-based EOR strategies accrue high pumping energy costs
and are generally not commercially viable. Gas-based injection strat-
egies are more applicable in enhanced tight oil recovery23. In the gas-
based flooding process (Fig. 1d), high-pressure gas is injected into the
reservoir from a horizontal well (injection well) and finally into frac-
tures. Oil is then expected to be displaced toward and recovered
from neighbouring wells. In the huff-and-puff process (Fig. 1e), high-
pressure gas is first injected into the reservoir from multiple wells
(huff) and the wells are subsequently sealed for long periods to soak
(soak period). After reaching a local dissolution equilibrium, wells are
reopened and the reservoir pressure drives the resulting oil-gas mix-
ture to production (puff).
To directly verify and quantify the efficiency of these approaches
in reservoir-relevant nanoconfinement, we replicate these processes
within the nanomodel. For gas flooding tests, the fabricated nanopo-
rous media was initially filled with light crude oil (Texas Crude)
through one microchannel (oil channel on the right, Fig. 1b). The res-
ervoir pressure was then controlled through the oil channel during
testing. Gas was injected from the other microchannel (gas channel
on the left) at a pressure higher than the oil reservoir pressure, driv-
ing the oil trapped in the nanoporous media towards the oil channel.
In huff-and-puff testing, the nanoporous media was first saturated
with light oil through both microchannels. Then, the two microchan-
nels were cleaned with air simultaneously, until all oil was removed
(as verified with fluorescence-based quantification of any residual
crude oil). After vacuuming the air, the target gas was then injected
at the testing pressure on both sides of the nanoporous media (huff),
and the entire system was sealed to reach equilibrium. During pro-
duction (puff), the gas pressure in both microchannels was reduced
(see ESI† section 2). For all tests, the nanofluidic chip was at a uni-
form temperature of 323 K. Time-lapse images were captured using
an optical microscope, and analysed to evaluate the displacement
dynamics of the oil (dark) and gas (light) phases, as shown in Fig. 1f.
The image sequence was processed to track the gas-oil interactions
and to provide a measure of recovery efficiency as a function of op-
erating conditions and treatment (Fig. 1f).
3. Results and discussion
3.1 Immiscible gas flooding
Nitrogen injection is an EOR strategy that is commonly applied in
conventional reservoirs. Nitrogen is relatively low cost, non-corro-
sive and can provide excellent performance24. For these same rea-
sons, N2 injection in nanoporous shale is of interest, but the perfor-
mance, or potential, of N2 injection in nanoconfinement is largely un-
known25. While information on the topic is scarce, Nitrogen pilots
have been reported on wells in Appalachian basin where nitrogen gas
was used as a fracturing fluid resulting in 28% improvement in esti-
mated ultimate recovery compared to a foam-based method26. Here,
after filling the nanomodel with oil, the oil reservoir pressure was set
to 5 MPa (simulating near-wellbore depleted reservoir pressure fol-
lowing primary recovery). N2 was then injected at pressures ranging
from 5 MPa to 11 MPa. The injection pressure was kept constant for
the duration of each test, and tests were run at 5, 5.5, 6, 6.5, 7, 9,
and 11 MPa. Due to a high N2–oil minimum miscibility pressure (~40
MPa)27, all N2 experiments were immiscible floods, in keeping with
the expected conditions for N2 in many tight oil reservoirs (e.g.
Bakken formation)28. The initiation of N2 flooding happened first at 7
MPa, and no detectable N2 flooding occurred in lower pressure tests.
With the oil side reservoir pressure maintained at 5 MPa, these re-
sults indicate a significant pressure difference threshold (2 MPa) re-
quired for immiscible flooding in nanoconfinement that has no ana-
log in conventional, microporous reservoirs. This threshold is a prod-
uct of the strong capillarity effects in nanoconfinement. The strong
capillarity at the nanoscale is known to significantly affect fluid
transport29 and phase transition30, as were carefully characterized
with single nanochannel previously. The capillary pressure generated
at the N2–oil interface is predicted to be ~1.4 MPa (see ESI†, section
3). When the pressure difference between N2 and oil is below this
threshold, the system simply reaches a pressure equilibrium at the
liquid-gas interface, that is, the fracture-nanoporous media interface.
This pressure requirement is a barrier to injection and ultimately re-
covery in these systems. In conventional reservoirs, the pore size is
at the micrometer or even millimeter scale, and thus the capillary
force is reduced by more than 1000 times (~ 1 kPa), and is thus neg-
ligible in the context of injection pressures and viscous pressure
drops in conventional operations.
Journal Name
ARTICLE
4 | J. Name. , 20xx , 00 , 1-6 This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Please do not adjust margins
Fig. 2 Nanomodel screening of immiscible N2 flooding in tight oil reservoirs. (a)-(c) Oil recovery history of immiscible N2 flooding at 7, 9 and 11 MPa. At test pressures below 7
MPa, the gas was prohibited from flooding the nanomodel due to capillary force. The scale bar represents 200 µm. (d) Cumulative oil recovery changes with time. The inset figure
shows the final oil recovery percentage (FOR) at different N2 pressures. The error bar represents the standard deviation of three independent nanoporous media tests. The oil
reservoir pressure on chip was held constant at 5 MPa.
Fig. 2a-c shows the composite spatio-temporal advancement of
nitrogen into the oil phase with the color bands showing relative tim-
ing during the injection process (that is, the color reflects the first
instant that a given pore was cleared), at 7, 9 and 11 MPa. For clarity,
greyscale image sequences are provided in each case at left (Fig. 2a-
c). At these injection conditions the N2-oil systems exhibited capillary
fingering behavior reflective of the low Ca (~10-8) and mobility ratio
(~10-3) zone in the Lenormand phase diagram31 (see ESI†, section 3).
As the gas pressure was increased from 7 to 11 MPa, the gas fingers
become smaller and exhibit a higher spatial density leading to higher
recovery in keeping with the classical theory31. While the fingering
effect is an inherent limitation in immiscible flooding, increasing the
injection pressure can improve EOR productivity in nanoconfine-
ment. In this test, the ultimate oil recovery in the nanomodel in-
creases from 48% to 86% by increasing the injection pressure from 7
to 11 MPa (Fig. 2d).
In terms of operations, these immiscible N2 flooding results
offer some insights. While it is typical for micro- and nanomodel
results to overestimate recovery in general, the difference here
in the side-by-side comparison of nanomodels indicates the sig-
nificance of the initial gas pressure on the performance of im-
miscible gas flooding in nanoporous systems. The nanomodel
results also clearly indicate limitations in terms of the ultimate
applicability of immiscible EOR in shale. Specifically, the smaller
pore throats in shale formations can be a few nanometers20, in
which case the capillary pressure influence observed here is ex-
pected to be further increased to tens of MPa. This pressure is
on the same order as initial reservoir pressure and effectively
prohibits immiscible gas access into these pores. Generating gas
flow in nanoporous media already introduces significant viscous
pressure drop (scaling with the square of pore size), and the ad-
ditional overhead of this capillary pressure is likely to make im-
miscible gas flooding impractical for many shale reservoirs.
More broadly, the significant capillary barrier observed here
is at work in a range of nanoporous material applications. For
example, electrocatalytic flow cell and fuel cell gas diffusion
electrodes with finely structured with nanopores could permit
larger pressure discrepancies up to MPa compared to kPa in mi-
croporous electrode32, presenting a much broader range of op-
erating conditions and potential for the removal of produced
liquid phase. Particularly in CO2 electrocatalytic reduction cells,
strong capillary pressure in hydrophobic nanoporous gas diffu-
sion electrode is essential to prevent flooding the CO2 gas cham-
ber33, with the production of highly wetting multicarbon prod-
ucts, such as ethanol. Nanomodel testing as demonstrated here
can inform on the response of these critical nanoporous layers.
0 50 100 1500
20
40
60
80
100
Oil
Reco
very
Perc
en
tag
e (
%)
Time (s)
N2
0
100
60
20
40
80
Unit (s)
0
25
15
5
10
20
Unit (s)
0
25
15
5
10
20
Unit (s)
7 MPa (N2)
9 MPa (N2)
11 MPa (N2)
(a) (b)
(c) (d)
7 9 110
50
100
FO
R (
%)
(7 MPa N2) (9 MPa N2)
(11 MPa N2)
(5 MPa Oil) (5 MPa Oil)
(5 MPa Oil)
Journal Name
ARTICLE
This journal is © The Royal Society of Chemistry 20xx J. Name. , 20xx , 00 , 1-5 | 5
Please do not adjust margins
Please do not adjust margins
3.2 Miscible gas flooding
CO2 is another injection gas candidate for enhanced oil re-
covery. EOR using CO2 can not only improve oil recovery effi-
ciency, but can also serve a carbon storage goal. CO2 flooding is
widely applied in conventional medium to heavy oil reservoirs34,
and has recently been piloted in the Jilin tight Oil field (China),
with reports of the oil recovery factor increasing 4.5 % with 0.26
million tons of CO2 been injected underground after 6 years of
operation35. Compared to nitrogen, the minimum miscibility
pressure of CO2 in light oil is significantly lower due to light com-
ponent extraction from the oil into CO2 at relatively low pres-
sures. The minimum miscibility pressure of CO2 with the light oil
in this study (i.e., West Texas crude) at 323 K was ~10 MPa36.
Here, we found the minimum miscibility pressure in the nano-
model to approximate this bulk value. The miscible CO2 flooding
experiments were conducted under two conditions, at (i) low
initial oil pressure well below the minimum miscibility pressure
(5 MPa), and (ii) high initial oil pressure where CO2-oil miscibility
is expected (10 MPa). These two cases simulate EOR in shale
formations with low and high reservoir pressures, respectively.
The CO2 was injected at 11 MPa to maintain initial miscible
flooding conditions for both cases (see ESI† section 2 for de-
tailed experimental procedures).
The spatio-temporal advancement of the CO2 phase into the
oil for both the low and high initial reservoir pressures is shown
in Fig. 3a and 3b, respectively. For the low reservoir pressure
case (5 MPa), oil displacement was first observed as gas break-
out close to the oil reservoir side (right). A significant film-wise
displacement followed, (i.e., miscible flooding) initiating from
the gas channel (left), ultimately resulting in a clean sweep of
the trapped oil in the nanoporous media. This early gas break-
out is attributed to a change in miscibility conditions across the
nanomodel. Specifically, on the injection side CO2 is miscible
with the oil phase and enters readily, however, as the CO2 mi-
grates across the nanomodel, the local pressure drops below
the minimum miscibility pressure (~10 MPa) and CO2 gas
emerges as a distinct phase. The strong capillary forces sur-
rounding the gas phase (noted earlier as a significant flooding
barrier) serve a powerful recovery role here, when generated
within the nanoporous media. The overall flooding result in
terms of oil recovery percentage (93%, Fig. 3c) is notably better
than that in the immiscible N2 flooding case (86%), with the
same gas pressure (11 MPa).
Relating these results to nanoporous tight oil and shale res-
ervoirs, the gas break-out mechanism observed here (Fig 3A)
can be expected if minimum miscibility pressure is between
Fig. 3 Screening conditions for miscible gas flooding (11 MPa CO2) in a shale/tight oil
reservoir at 323 K, for (a) a relatively low pressure reservoir (5 MPa) and (b) a high
pressure reservoir (10 MPa). The scale bar represents 200 µm. (c) Cumulative oil recov-
ery plotted vs. time. The inset figure shows the final oil recovery percentage (FOR) at
different reservoir pressures. The error bar represents the standard deviation from three
independent nanoporous media tests.
the injecting CO2 pressure and that of the reservoir, provided
the diffusion of the injected fluid significantly leads the displace-
ment of the fluid front. We quantify the relation between the
displacement CO2-oil film front and the diffusion length of CO2
into the nanoporous system with a simplified model (full details
in ESI†, section 4). As shown in Fig. 4, the displacement CO2-oil
film front (𝐿𝑔𝑎𝑠 ) can be expressed as:
0 20 40 60 800
20
40
60
80
100
Oil
Re
co
ve
ry P
erc
en
tag
e (
%)
Time (s)
(a)
(b)
0
50
30
10
20
40
Unit (s)
5 MPa (oil)
10 MPa (oil)
(c)
5 1090
95
100
FO
R (
%)
0
75
45
15
30
60
Unit (s)
CO2
(11 MPa CO2) (5 MPa Oil)
(11 MPa CO2) (10 MPa Oil)
Journal Name
ARTICLE
6 | J. Name. , 20xx , 00 , 1-5 This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Please do not adjust margins
𝐿𝑔𝑎𝑠 = 𝐿𝑅 − √𝐿𝑅2 −
𝜑𝑟2 (𝑃𝑔𝑎𝑠 −𝑃𝑜𝑖𝑙)
4𝜇𝑜𝑖𝑙𝜏𝑡 (1)
where 𝐿𝑅 is the total reservoir length, 𝑃𝑔𝑎𝑠 is the injection pres-
sure, 𝑃𝑜𝑖𝑙 is the reservoir oil pressure, 𝜑 is the porosity of the
reservoir, 𝑟 is the mean pore radii, 𝜏 is the reservoir tortuosity,
𝜇𝑜𝑖𝑙 is the oil viscosity, 𝑡 is the injection time. The diffusion
length of injected fluid into the nanoporous media (𝐿𝑑𝑖𝑓𝑓 ) fol-
lows Fick’s law:
𝐿𝑑𝑖𝑓𝑓 = 2√𝐷
𝜏𝑡 (2)
Where the gas diffusivity into oil is 𝐷, here is 5 × 10−9 m2 /s. The
difference (∆𝐿) between 𝐿𝑑𝑖𝑓𝑓 and 𝐿𝑔𝑎𝑠 is thus:
∆𝐿 = 𝐿𝑑𝑖𝑓𝑓 − 𝐿𝑔𝑎𝑠 = 2√𝐷
𝜏𝑡 − 𝐿𝑅 + √𝐿𝑅
2 −𝜑𝑟2 (𝑃𝑔𝑎𝑠 −𝑃𝑜𝑖𝑙)
4𝜇𝑜𝑖𝑙𝜏𝑡 (3)
For ∆𝐿 ≥ 0, the gas injection time is within a range (i.e., 0 < 𝑡 <
𝑡𝑠 ), and:
𝑡𝑠 =16𝐷𝐿𝑅
2
𝜏[4𝐷+𝜑𝑟2(𝑃𝑔𝑎𝑠−𝑃𝑜𝑖𝑙)
4𝜇𝑜𝑖𝑙𝜏]
2 (4)
At time less than 𝑡𝑠 , gas diffusion ahead of the front is dominant
and there is potential for gas breakout within the media upon
drawdown. For the nanomodel in this work, 𝐿𝑅 = 10−3 m, 𝜑 =
0.13, 𝑟 = 30 nm, 𝜏 = 1.5 and 𝑃𝑔𝑎𝑠 − 𝑃𝑜𝑖𝑙 = 6 MPa, the time range is
calculated to be 𝑡𝑠 = 23 s. This timescale corresponds to that ob-
served in the experiments, where the film front reached the gas
breakout zone at ~20 s (Fig. 3a). Translating these results to a
reservoir, 𝐿𝑅 is on the order of 10 m, 𝜑 is on the order of 0.1, 𝑟
is on the order of 100 nm, 𝜏 is on the order of 1 and 𝑃𝑔𝑎𝑠 − 𝑃𝑜𝑖𝑙 is
on the order of 10 MPa. The corresponding reservoir relevant
time constant, 𝑡𝑠 = 107 s (~4 months), which is a practical time-
scale for operations.
Fig. 4 Model schematic of film-wise displacement with gas diffusion in the reservoir.
The diffusion length of CO2 into the reservoir is longer than
the displacement CO2-oil film front within the first few months
of the production period (due to the extreme low permeability
of shale hindering the pressure driven film-wise displacement).
In this scenario, CO2 diffused deeply into the reservoir, far away
from the film front, can exceed solubility locally. Gas phase can
then break out into fingers as observed here, providing addi-
tional oil recovery through a mechanism akin to local immiscible
flooding but without the initial capillary pressure barrier. This
recovery mechanism is unique to nanoporous reservoirs and is
a product of both nanoconfinement and thermodynamics. A
trade-off of this strategy would be the relatively high CO2-to-oil
ratio in the produced fluids, which can be remedied by surface-
based thermal separation and reinjection of CO2.
For the high pressure reservoir case (10 MPa), the oil displace-
ment process is uniform film-wise throughout, exhibiting highly effi-
cient oil recovery with only 1 MPa driving pressure difference (~100%
cumulative oil recovery percentage, Fig. 3c). For miscible flooding,
the interfacial tension between gas (here CO2) and oil is significantly
reduced (approaching zero). The oil capillarity in nanoporous media
is thus minimized compared to immiscible flooding. For an additional
comparison, we also tested the efficiency of N2 flooding at 11 MPa,
with oil reservoir pressure kept at 10 MPa. We found that there was
no flooding observed, in keeping with the earlier finding that a pres-
sure difference of ~ 2 MPa is required for nitrogen to overcome the
capillary pressure inherent to the nanoporous media. Further in-
creasing the N2 pressure to 12 MPa achieved flooding, however, the
cumulative oil recovery percentage was only 31% (see ESI†, section
5), which is much lower than all CO2 miscible flooding results. In ad-
dition to reducing the capillary force, the diffusion of CO2 into the oil
phase also eases the flow of oil by reducing the oil viscosity36 (the
viscosity of supercritical CO2 being two orders of magnitude lower
than that of the oil).
Together, these N2 and CO2 flooding results indicate the im-
portance of engineering nanoporous EOR methods to ensure misci-
bility, or partial miscibility in these reservoirs. The minimum miscibil-
ity pressure value measured at bulk scale is reasonably representa-
tive at the ~ 60 nm pore scale used here, and can serve as a general
guide to understanding EOR effectiveness in these systems. While
miscible injection is generally helpful in conventional reservoirs
mainly by reducing the oil viscosity, the elimination of capillary force
barriers (several to several tens of MPa) is paramount in unconven-
tional, nanoporous reservoirs. The strength miscible flooding can be
additionally leveraged in a hybrid miscibility strategy in low pressure
unconventional oil reservoirs, where the gas is injected above MMP
such that gas breakout occurs within the pore space, driving oil to
the producer. More generally, the miscible flooding results here
demonstrate enhancement in nanoscale mass transport achieved
with miscible supercritical fluid injection. At supercritical condition,
the interfacial energy is effectively weakened, and both convective
and diffusive transport modes are enhanced. These mass transport
Gas
inje
ctio
n
Oil
pro
du
ctio
n
LR
Lgas
Pgas
Ldiff
Poil
Journal Name
ARTICLE
This journal is © The Royal Society of Chemistry 20xx J. Name. , 20xx , 00 , 1-5 | 7
Please do not adjust margins
Please do not adjust margins
enhancements are welcome in nanosystems where confinement
generally limits transport and mass exchange.
3.3 Huff and Puff
Traditional gas flooding requires continuous gas injection
into the reservoir, with associated costs in pumping energy, fa-
cilities, and gas usage. Flooding also necessitates the use of at
least two wells for injection and production. Huff-and-puff is an
iterative, single-well alternative to traditional flooding where
gas is injected for a time, allowed to soak into the reservoir, and
then the mixture is produced from the same well. This method
is gaining traction with tight oil and shale producers, with pilot
field tests in the Bakken formation37. However, the overall re-
covery performance has not been significant to date due to gas
leakage issues. The potential for huff-and-puff in shale is thus of
both academic and industrial interest23. We reproduce this pro-
cess in the nanomodel with the configuration shown in Fig. 1b.
After filling the nanoporous media with oil and cleaning the ad-
joining microchannels, we injected CO2 symmetrically at the tar-
get pressure (5, 7, 9 or 11 MPa) and sealed the device for 1 hr
to allow sufficient CO2 to diffuse into the oil. The system was
then depressurized (~20 s) to 1 MPa to produce oil from the
nanomodel. The pressure conditions chosen here are similar to
previous reported shale core tests38.
Fig. 5 Screening conditions for CO2 huff-and-puff in a shale nanomodel. CO2 was initially injected into the nanomodel (filled with oil) at (a) 7, (b) 9 and (c) 11 MPa, respectively
(huff). The scale bar represents 200 µm. After soaking for one hour, the system pressure was rapidly reduced to 1 MPa (Puff). Oil was produced from the nanomodel during the
production process, with (d) accumulative oil recovery changing with time plotted. The inset figure shows the final oil recovery percentage (FOR) at different injecting CO2 pressures.
The error bar represents the standard deviation from three independent nanoporous media tests.
Production results using initial injection pressures of 7, 9,
and 11 MPa in the huff-and-puff strategy are shown in Fig. 5a-
c. All three cases show significant production, with the miscible
(11 MPa) injection case showing the highest sweep efficiency
with more interconnected and film-wise displacement, espe-
cially near the injection channels. The cumulative oil recovery
(Fig. 5d) is increased from 17% at 7 MPa to 77% at 11 MPa. In
all cases gas breakout and expansion within the reservoir is a
primary recovery mechanism. The miscible case outperforms
due to a combination of increased CO2 input and associated vis-
cosity reduction, as well as significant initial direct extraction of
light hydrocarbon components during injection. These meas-
urements are in very close agreement with a previous core test
(Eagle Ford formation), where miscible CO2 injection at 10 MPa
led to ~70% oil recovery efficiency, and immiscible CO2 injection
at 6 MPa was only ~20%39.
While the cases shown in Fig. 5 all show recovery, it is
equally informative to consider running conditions that were in-
effective. First, lower injection pressure (5 MPa) resulted in neg-
ligible production, and second, lower pressure drawdown rates
(~ 200s) resulted in negligible production at all test pressures.
At lower injection pressures, the main issue is capillary pressure
0 1 2 3 4 5
0
20
40
60
80
100
Oil R
ec
ov
ery
Pe
rce
nta
ge
(%
)
Time (s)
0
5
3
1
2
4
Unit (s)
0
3.0
1.8
0.6
1.2
2.4
Unit (s)
0
2.5
1.5
0.5
1.0
2.0
Unit (s)
(a) (b)
(c) (d)
7 MPa (CO2)
9 MPa (CO2)
11 MPa (CO2)7 9 11
0
50
100
FO
R (
%)
CO2
(7 MPa CO2) (9 MPa CO2)
(11 MPa CO2)
Journal Name
ARTICLE
8 | J. Name. , 20xx , 00 , 1-6 This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Please do not adjust margins
in nanoconfinement decreasing the CO2 solubility, and ulti-
mately reducing the density of CO2 in the oil below that needed
to form a gas bubble. Specifically, at 5 MPa the oil pressure was
actually 4 MPa due to capillary pressure drop across the inter-
face (ESI†, section 3), and the CO2 concentration in the oil
reaches a maximum of 25 kg/m3.36 Even with significant pres-
sure drawdown this density is insufficient to overcome nano-
confinement capillary force and form bubbles. In contrast, with
CO2 at 7 MPa, the internal oil pressure would be 6.3 MPa, bring-
ing the CO2 concentration to 103 kg/m3, which is sufficient to
drive bubble formation within the nanoporous media upon
drawdown (detailed calculations in ESI†, section 6). Lower
drawdown rates also proved ineffective at all injection pres-
sures. Reducing the pressure between injection and production
stages over 200 s (vs. 20 s as shown in Fig. 5), resulted in no
observable oil recovery. The key mechanism here is gas diffu-
sion40 out of the nanoporous media during drawdown. If pres-
sure reduction is slow, CO2 diffusion out of the oil can reduce
the concentration to below that required to generate a gas bub-
ble internally (calculations in ESI†, section 6). In short, huff and
puff in nanoconfinement relies on an asymmetry: during the in-
jection phase one must provide sufficient time and gas pressure
for deep diffusion of CO2, and during production, one must pro-
vide insufficient time for diffusion.
The mechanisms identified here are relevant to nanoporous
reservoirs in the field, but must be adjusted for much larger
lengths and timescales. Specifically, the distance between wells
in shale reservoirs is >10 m41, and the time to approach the
equilibrium dissolved gas concentration via diffusion through
nanoporous media at this scale is a prohibitively long 109 s, or
~32 years. Thus, in practice CO2 reach into the reservoir is finite,
and optimizing this process requires a balance in injection pres-
sure and injection/soak time to achieve sufficient CO2 density
near the well. In this case, the nanomodel results clearly show
the importance of having a sufficient drawdown rate in order to
‘beat’ local diffusion of CO2 out of the nanoporous matrix. These
results resolve a general debate regarding drawdown rates
(low42 vs. high43) with both physical evidence and key mecha-
nisms resolved. Another example of field-scale utility of the na-
nomodel testing is in measuring, and mitigating, the many im-
pacts of nanoporous capillary pressure – a field relevant varia-
ble replicated at scale in our device. The huff-and-puff results
here show the dual role of capillary pressure in limiting solubil-
ity and increasing the barrier to gas phase formation unique to
nanoporous reservoirs. Likewise, results from bulk scale solubil-
ity tests and microporous EOR pilots are likely to overestimate
the total amount of gas dissolved in nanoporouus reservoirs,
and must be reconciled with the many implications of nanopo-
rous capillary pressure clearly resolved in the nanomodel.
More generally, the huff-and-puff test results here highlight
the impact of the much higher superheat required for gas bub-
ble nucleating within nanoporous media. This effect is present
in wide range of physical systems. For example in membranes,
oxygen solubility in meta-stable water has been found to double
within nanoconfined sub-saturated liquid water (relative humil-
ity ~0.55, pore diameter ~1.4 nm)44 without bubble nucleation.
The modification of gas solubility at nanoscale is fundamentally
relevant to corrosion of porous metals, as well as chemical re-
actions in meta-stable water.
In terms of limitations, the nanomodel approach here
shares some deficiencies with all two-dimensional representa-
tions of porous media, or micromodels45. Most significantly, the
planar geometry enables direct optical access, but also funda-
mentally limits transport as compared to three-dimensional
natural materials. This influence is less significant in EOR pro-
cesses that tend to proceed as a front (or a line in a 2D plane),
but remains a limitation. Other key limitations of the nano-
model approach are the lack of geological heterogeneity in-
cluded, and the lack of variation in pore size. The regular, mesh-
like, nanoporous media used here does not fully represent the
complex geometry of a given shale system, but instead provides
a means to directly visualize fluid interactions at a nanopore
scale representative of major shale resources. It is also worth
noting that nanopores at extreme scales (1-10 nm, sub-10 nm)
will lead to fluid transport properties deviating from bulk fluid
(e.g., molecular diffusivity reduced by a few orders of magni-
tude46, 47). The sub-10 nm pores in tight oil reservoirs are not
dominant in terms of pore volume contribution, however, sub-
10 nm transport physics could play a role in pore throats in con-
necting and potentially gating larger pores of interest. Recent
studies have elucidated the fundamental role of pore size on the
fluid mechanics and thermodynamics of pure hydrocarbon sys-
tems and multicomponent mixtures in channels ranging from
sub-10 nm to 1 μm29, 30. While these studies report a rich, size-
dependent physics, the onset of deviations from bulk liquid be-
havior is generally at sub-10 nm, and thus smaller than the main
pore volume contributors of interest here. The long history of
micromodel development provides some guidance on how to
mitigate these limitation by, for instance, incorporating miner-
als and clay particles, and multi-depth micro/nanofabrication.
Conclusions
In this work, we characterize the transport and phase change of
a nanoconfined complex fluid mixture using a nanofluidic approach.
A 2-D nanoporous media was fabricated matching the main pore size,
permeability and porosity in unconventional shale formations. We
studied gas flooding and huff-and-puff strategies under miscible and
Journal Name
ARTICLE
This journal is © The Royal Society of Chemistry 20xx J. Name. , 20xx , 00 , 1-6 | 9
Please do not adjust margins
Please do not adjust margins
immiscible conditions. In immiscible gas flooding (N2), the inherent
capillary pressure in nanopores presented a significant initiation
pressure threshold not present in conventional reservoirs with mi-
cro/millimeter pores. In contrast, immiscible gas flooding will gener-
ate capillary fingers in nanoporous media (shale) similar to the dy-
namics observed in microporous media (conventional reservoirs). By
selecting gas species with a low minimum miscibility point in oil (e.g.,
CO2), at the same gas pressure condition one can achieve miscible
gas flooding with a much lower initiation pressure threshold, and
meanwhile significantly improve the recovery efficiency through a
stable film-wise displacement.
In contrast to flooding where injection pressure is the main op-
erational variable, the huff-and-puff strategy additionally hinges on
injection duration and pressure drawdown rate. The nanomodel
uniquely informs the role of these controllable variables in terms of
recovery effectiveness and the underlying reservoir transport phe-
nomena. Specifically, for low injection pressures, the main issue is
capillary pressure in nanoconfinement decreasing the CO2 solubility,
and ultimately reducing the density of CO2 in the oil below that
needed to form the gas bubbles that drive recovery. The formidable
capillary pressure plays a dual role here, limiting CO2 solubility at in-
jection as well as presenting a barrier to bubble formation during
production. For the injection period and drawdown rate, we find that
huff and puff in nanoconfinement relies on an asymmetry: during in-
jection phase one must provide sufficient time and gas pressure for
deep diffusion of CO2, and during production, one must not provide
sufficient time for diffusion in order to force productive gas breakout.
Effective deployment of such strategies into nanoporous
systems hinges on the informed combination of controllable pa-
rameters, with nanoporous capillary pressure, diffusion and dis-
placement playing dominant – and in some cases multifaceted
- roles. Physically resolving these processes in a nanomodel with
representative geometries, complex multicomponent fluids,
and relevant pressures is essential for developing effective op-
erations for shale reservoirs, and more generally for nanopo-
rous systems.
Conflicts of interest
D.S. and A.A. have a financial interest in Interface Fluidics, and
A.A. is an employee of Interface Fluidics, a start-up company
that tests EOR strategies for the oil and gas industry.
Acknowledgement
The authors gratefully acknowledge support from Schlum-
berger Canada Ltd., Alberta Innovates-Energy and Environment
Solutions, and from the Natural Sciences and Engineering Coun-
cil of Canada through a Collaborative Research and Develop-
ment Grant, as well as on-going research funding through the
Discovery Grants program, Discovery Accelerator Supplement
and the Canada Research Chairs program. In addition, infra-
structure funding provided by the Canada Foundation for Inno-
vation and Ontario Research Fund is gratefully acknowledged.
References
1. J. Lee, R. Banerjee, M. George, D. Muirhead, P. Shrestha, H. Liu, N. Ge, S. Chevalier and A. Bazylak, J. Electrochem. Soc., 2017, 164, F1149-F1157.
2. M. M. Pendergast and E. M. Hoek, Energy Environ. Sci., 2011, 4, 1946-1971.
3. L. Zhou, Y. Tan, D. Ji, B. Zhu, P. Zhang, J. Xu, Q. Gan, Z. Yu and J. Zhu, Sci. Adv., 2016, 2, e1501227.
4. Y. Peng, T. Wu, L. Sun, J. M. Nsanzimana, A. C. Fisher and X. Wang, ACS Appl. Mater. Interfaces, 2017, 9, 32782-32789.
5. J. Zhao, Z. Jin, Q. Hu, Z. Jin, T. J. Barber, Y. Zhang and M. Bleuel, Sci. Rep., 2017, 7, 15413.
6. J. D. Hughes, Nature, 2013, 494, 307. 7. J. J. Conti, P. D. Holtberg, J. Beamon, A. M. Schaal, J. Ayoub
and J. T. Turnure, US Energy Information Administration, 2018.
8. Kerr, R. A., Science 2010, 328, 1624-1626. 9. R. W. Howarth, A. Ingraffea and T. Engelder, Nature, 2011,
477, 271. 10. R. D. Vidic, S. L. Brantley, J. M. Vandenbossche, D.
Yoxtheimer and J. D. Abad, Science, 2013, 340, 1235009. 11. K. S. Lackner, Science, 2003, 300, 1677-1678. 12. M. A. Sayed, G. A. Al-Muntasheri and F. Liang, J. Pet. Sci.
Technol. 2017, 157, 164-186. 13. A. Phan, D. R. Cole, R. G. Weiß, J. Dzubiella and A. Striolo,
ACS Nano, 2016, 10, 7646-7656. 14. L. Li, Y. Zhang and J. J. Sheng, Energy Fuels, 2017, 31, 3856-
3867. 15. J. J. Sheng, J. Nat. Gas Sci. Eng. 2015, 22, 252-259 16. Z. Jin, Fluid Phase Equilib., 2018, 458, 177-185. 17. J. Zhong, J. Riordon, S. H. Zandavi, Y. Xu, A. H. Persad, F.
Mostowfi and D. Sinton, J. Phys. Chem. Lett. 2018, 9, 497-503.
18. A. Jatukaran, J. Zhong, A. H. Persad, Y. Xu, F. Mostowfi and D. Sinton, ACS Appl. Nano Mater. 2018, 1, 1332-1338.
19. S. Liu, Y. Ma and B. Bai, Microfluid. Nanofluid. 2018, 22, 39-50.
20. P. H. Nelson, AAPG Bull., 2009, 93, 329-340. 21. S. Zhang, G. Jiang, S. Gao, H. Jin, Y. Zhu, F. Zhang and J. Jin,
ACS Nano, 2018, 12, 795-803. 22. R. Sander, Z. Pan and L. D. Connell, J. Nat. Gas Sci. Eng.,
2017, 37, 248-279. 23. J. J. Sheng, J. Pet. Sci. Eng. 2017, 159, 654-665. 24. D. Denney, J. Pet. Technol. 2001, 53, 55-56.
Journal Name
ARTICLE
10 | J. Name. , 20xx , 00 , 1-6 This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins
Please do not adjust margins
25. Y. Yu, X. Meng and J. J. Sheng, J. Unconvent. Oil Gas Resour. 2016, 15, 56-65.
26. Wozniak, G., Vactor, R. T., Hina, D. S., SPE East. Reg. Meeting, Morgantown, West Virginia, USA. 10, 2010.
27. S. Li and P. Luo, Petroleum, 2017, 3, 79-86. 28. S. H. Hejazi, Y. Assef, M. Tavallali and A. Popli, Fuel, 2017,
210, 758-767. 29. M. A. Alibakhshi, Q. Xie, Y. Li and C. Duan, Sci. Rep., 2016,
6, 24936. 30. J. Zhong, Y. Zhao, C. Lu, Y. Xu, Z. Jin, F. Mostowfi and D.
Sinton, Langmuir, 2018, 34, 9927-9935. 31. R. Lenormand, E. Touboul and C. Zarcone, J. Fluid Mech.
1988, 189, 165-187. 32. W. Dai, H. Wang, X.-Z. Yuan, J. J. Martin, D. Yang, J. Qiao
and J. Ma, Int. J. Hydrogen Energy, 2009, 34, 9461-9478. 33. D. M. Weekes, D. A. Salvatore, A. Reyes, A. Huang and C. P.
Berlinguette, Acc. Chem. Res., 2018, 51, 910-918. 34. F. Gozalpour, S. Ren and B. Tohidi, Oil Gas Sci. Technol.,
2005, 60, 537-546. 35. B. Ren, S. Ren, L. Zhang, G. Chen and H. Zhang, Energy,
2016, 98, 108-121. 36. A. Sharbatian, A. Abedini, Z. Qi and D. Sinton, Anal. Chem.,
2018, 90, 2461-2467. 37. L. Wang, Y. Tian, X. Yu, C. Wang, B. Yao, S. Wang, P. H.
Winterfeld, X. Wang, Z. Yang and Y. Wang, Fuel, 2017, 210, 425-445.
38. J. Ma, X. Wang, R. Gao, F. Zeng, C. Huang, P. Tontiwachwuthikul and Z. Liang, Fuel, 2015, 154, 35-44.
39. Gamadi, T., Sheng, J., Soliman, M., Menouar, H., Watson, M., Emadibaladehi, H., SPE improv. Oil Recovery Symposium, Tulsa, Oklahoma, USA, 04, 2014.
40. R. Raccis, A. Nikoubashman, M. Retsch, U. Jonas, K. Koynov, H.-J. r. Butt, C. N. Likos and G. Fytas, ACS Nano, 2011, 5, 4607-4616.
41. W. Yu, H. R. Lashgari, K. Wu and K. Sepehrnoori, Fuel, 2015, 159, 354-363.
42. L. Li and J. J. Sheng, J. Pet. Sci. Eng., 2017, 153, 36-46. 43. H. Jia and J. J. Sheng, Petroleum 2017, 4, 7-14. 44. P. Lidon, S. C. Marker, J. J. Wilson, R. M. Williams, W. R.
Zipfel and A. D. Stroock, Langmuir, 2018, 34, 12017-12024. 45. A. Anbari, H. T. Chien, S. S. Datta, W. Deng, D. A. Weitz and
J. Fan, Small, 2018, 14, 1703575. 46. J. Zhong, S. Talebi, Y. Xu, Y. Pang, F. Mostowfi and D. Sinton,
Lab Chip 2018, 18, 568-573. 47. G. Bruno, R. Nicola Di Trani, E. Zabre, C. S. Filgueira, G.
Canavese, P. Jain, Z. Smith, D. Demarchi, S. Hosali and A. Pimpinelli, Nat. Commun. 2018, 9, 1682.