EWnergy

energies
Article
Recovery Mechanisms for Cyclic (Huff-n-Puff) Gas
Injection in Unconventional Reservoirs:
A Quantitative Evaluation Using
Numerical Simulation
B. Todd Hoffman and David Reichhardt *
Petroleum Engineering Department, Montana Technological University, Butte, MT 59701, USA;
thoffman@mtech.edu
* Correspondence: dreichhardt@mtech.edu; Tel.: +1-406-496-4887

Received: 24 August 2020; Accepted: 15 September 2020; Published: 21 September 2020
Abstract: Unconventional reservoirs produce large volumes of oil; however, recovery factors are
low. While enhanced oil recovery (EOR) with cyclic gas injection can increase recovery factors
in unconventional reservoirs, the mechanisms responsible for additional recovery are not well
understood. We examined cyclic gas injection recovery mechanisms in unconventional reservoirs
including oil swelling, viscosity reduction, vaporization, and pressure support using a numerical flow
model as functions of reservoir fluid gas–oil ratio (GOR), and we conducted a sensitivity analysis of
the mechanisms to reservoir properties and injection conditions. All mechanisms studied contributed
to the additional recovery, but their significance varied with GOR. Pressure support provides a
small response for all fluid types. Vaporization plays a role for all fluids but is most important for
gas condensate reservoirs. Oil swelling impacts low-GOR oils but diminishes for higher-GOR oil.
Viscosity reduction plays a minor role for low-GOR cases. As matrix permeability and fracture
surface area increase, the importance of gas injection decreases because of the increased primary oil
production. Changes to gas injection conditions that increase injection maturity (longer injection
times, higher injection rates, and smaller fracture areas) result in more free gas and, for these cases,
vaporization becomes important. Recovery mechanisms for cyclic gas injection are now better
understood and can be adapted to varying conditions within unconventional plays, resulting in better
EOR designs and improved recovery.
Keywords: EOR; cyclic gas injection; unconventional; recovery mechanisms; vaporization;

oil swelling; numerical simulation; huff-n-puff
1. Introduction
Unconventional reservoirs such as the Eagle Ford, Bakken, and others have had tremendous
success over the last decade. Production from these low-permeability, liquid-rich reservoirs has
increased to over 8 million stock tank barrels per day (STB/day) in the US, which accounts for almost
65% of US oil production [1] and, in North America, these types of reservoirs have cumulatively
produced over 10 billion barrels of oil. This success has been due to improvements in two important
areas of the petroleum industry: extended-reach horizontal wells and multi-stage hydraulic fracturing.
Possibly even more important than the high rates are the volumes of in-place oil for these unconventional
reservoirs. In each formation, there are hundreds of billions of barrels of oil in place [2–4], which exceeds
even the largest conventional reservoirs. These unconventional reservoirs served as the source rocks for
many conventional reservoirs, and while much oil migrated out to fill higher-permeability reservoirs,
even more has remained at the source.
Energies 2020, 13, 4944; doi:10.3390/en13184944 www.mdpi.com/journal/energies

Energies 2020, 13, 4944 2 of 24
However, there are still major issues to be addressed. The wells may start producing at very
high rates, but the rates drop off quickly, and the wells have low recovery factors, generally less
than 10% [5,6]. Even as longer horizontal wells and larger hydraulic fractures are implemented,
the initial rates tend to get higher, but the general behavior remains the same—rapid decline and
low overall recovery. Therefore, the need to develop methods to improve overall recovery is evident.
Some increases from infill drilling and improvements in hydraulic fracturing have the potential to
increase the recovery factors to some extent, but they are expected to continue to be much less than
20%. To achieve recovery factors of 20% and higher, some form of fluid injection is likely needed.
The value of enhanced oil recovery (EOR) for unconventional reservoirs like the Bakken and
Eagle Ford was noted early in the production of these reservoirs [7,8], and there has been significant
research over the last decade on this topic (summaries in Alfarge et al. [9] and Kanfar et al. [10]).
Much of this early work was done in the laboratory (e.g., Jin et al. [11] and Tovar et al. [12]) or with
numerical models (e.g., Li et al. [13] and Gala and Sharma [14]), where they looked at questions of
overall recovery, the type of gas, soak time, vaporization, and utilization.
There have been a few field trials, but most data from them are held within the companies.
The limited public data from these field trials were published for the Bakken [15] and the Eagle Ford [16],
but many of the data were not available, such as bottom hole flowing pressures, injection rates/pressures,
and producing gas–oil ratios. Therefore, it was difficult to validate the results from the laboratory
experiments or the numerical models. Then, in 2013, a company started having success by injecting
natural gas in the Eagle Ford in a cyclic manner where the injection wells and producing wells were
the same, often called huff-n-puff gas injection. In May 2016, the company reported their results [17],
which showed incremental recovery for the process to be between 30% and 70%.
Since that time, other companies have tried this process and found similar increases in oil recovery,
but the reasons for the additional recovery have not been extensively reported. Understanding
how fluid injection can help increase oil recovery from unconventional reservoirs will require a
multi-pronged approach to fully comprehend the issues and develop solutions that work. This paper
focuses on the recovery mechanisms for huff-n-puff natural gas injection projects to determine why
additional oil is recovered.
Recovery mechanisms are important for understanding what is happening in the reservoir and
for improving the overall recovery and efficiency. If the process is better understood, procedures
can be developed to maximize the recovery efficiency; thus, more oil can be produced with less
gas injected. Without understanding the recovery mechanisms, improving the process becomes an
expensive trial-and-error endeavor.
There have been numerous recovery mechanisms proposed for gas injection into liquid-rich
unconventional reservoirs. Nine are shown below, which should be considered a thorough but not
exhaustive list.
• Vaporization;
• Viscosity reduction;
• Secondary solution gas drive;
• Interfacial tension reduction;
• Wettability changes;
• Oil swelling (expansion);
• Pressure support;
• Rock/fluid interactions;
• Injection-induced fracturing.
The first five mechanisms (vaporization, swelling, secondary solution gas drive, viscosity reduction,
and pressure support) are the most often cited and appear to be the most significant from a literature
review. This work focuses on these, but it does not mean that the other mechanisms are not important.
Energies 2020, 13, 4944 3 of 24
While there has been significant work with enhanced oil recovery (EOR) in unconventional
reservoirs over the last decade, there has been little focus on the recovery mechanisms. Tovar et al. [18]
used laboratory
Energies 2020, 13, x data to discuss
FOR PEER REVIEW the importance of vaporization and declared that it is the main recovery 3 of 24
mechanism for gas injection EOR in organic rich shales. Hamdi et al. [19] discussed the complex
complexoffunction
function of mass
mass transfer andtransfer andprocesses
extraction extractionrelated
processes related to
to a number of adifferent
numberpotential
of different potential
mechanisms.
mechanisms.
Hoffman Hoffman[20]
and Rutledge andcompared
Rutledgethe [20] compared
role theand
of diffusion role of diffusion
advection and
for the advectionmechanism
oil-swelling for the oil-
swelling
using mechanism
analytical using
models. analytical
While these aremodels. While to
good papers these
startare good papers
discussing to start they
mechanisms, discussing
either
mechanisms, they either focus on a single mechanism or have only a qualitative
focus on a single mechanism or have only a qualitative comparison of different mechanisms. A fully comparison of
different mechanisms.
systematic and quantitativeA fully systematic
evaluation and quantitative
of different mechanisms evaluation
is needed, ofwhich
different
thismechanisms
paper intendsis
needed,
to address. which this paper intends to address.
This effort
This effort isisinspired
inspiredby byand
andbuilds
buildsupon
uponwork
workdone
donetrying
tryingto tounderstand
understand the the importance
importance of of
variousrecovery
various recoverymechanisms
mechanisms inin thermal
thermal recovery/steam
recovery/steam injection
injection methods
methods by Wu by[21].
Wu [21].
BurgerBurger
et al. et al.
[22]
[22] published
published a charta chart showing
showing howhow thethe relative
relative significance
significance of of themechanisms
the mechanismsfor forthermal
thermalrecovery
recovery
changedas
changed asaafunction
functionof ofthe
theAmerican
AmericanPetroleum
PetroleumInstitute
Institute(API)
(API)gravity
gravityof ofthe
theoil
oil(Figure
(Figure1).
1).
Figure1.1.Contributions
Figure Contributionsofofdifferent
differentmechanisms
mechanisms to improving oil recovery
to improving by a by
oil recovery heated fluid fluid
a heated Used Used
with
permission from [22] Editions Technip: Paris, France.
with permission from [22] Editions Technip: Paris, France.
Heavy
Heavy (low
(low API)
API) oils
oils are
are influenced
influenced by by viscosity
viscosity reduction,
reduction, while
while thermal
thermal expansion
expansion and
and
vaporization
vaporization are more significant for lighter oils. While this is a qualitative view of recovery
are more significant for lighter oils. While this is a qualitative view of recovery asas aa
function
function of
of the
the oil
oil density,
density,aamore
morequantitative
quantitativeevaluation
evaluationwas wasconducted
conducted forforaamedium-weight
medium-weight oiloil
(22
(22 API) as a function of pore volumes of steam injected [23]. For this oil, all mechanismscontribute
API) as a function of pore volumes of steam injected [23]. For this oil, all mechanisms contribute
significantly;
significantly;this
thiswork
workalso
alsoshowed
showedthatthatthe
therelative
relativeimportance
importance of of
thethe
mechanisms
mechanisms changes slightly
changes as
slightly
aasfunction of pore
a function volumes
of pore volumesinjected or time
injected (Figure
or time 2). 2)
(Figure
While huff-n-puff gas injection into unconventional reservoirs is different to thermal recovery,
a similar approach can be used to evaluate the mechanisms, including determining the recovery
as a function of fluid properties. The in-situ fluid properties can change dramatically across an
unconventional basin. The Eagle Ford is a good example, where, from northwest to southeast, the fluid
goes from a low gas–oil ratio (GOR) (<500 standard cubic feet per stock tank barrel (scf/STB)) black oil
to a dry gas, as shown in Figure 3.
vaporization are more significant for lighter oils. While this is a qualitative view of recovery as a
function of the oil density, a more quantitative evaluation was conducted for a medium-weight oil
(22 API) as a function of pore volumes of steam injected [23]. For this oil, all mechanisms contribute
significantly; this work also showed that the relative importance of the mechanisms changes slightly
Energies 2020, 13, of
as a function 4944
pore volumes injected or time (Figure 2) 4 of 24
Energies 2020, 13, x FOR PEER REVIEW 4 of 24
Figure 2. Relative importance of mechanisms contributing to incremental oil recovery for steamflood
over waterflood reprinted by permission [23] Taylor & Francis Ltd, http://www.tandfonline.com .
While huff-n-puff gas injection into unconventional reservoirs is different to thermal recovery,
a similar approach can be used to evaluate the mechanisms, including determining the recovery as a
function of fluid properties. The in-situ fluid properties can change dramatically across an
unconventional basin. The Eagle Ford is a good example, where, from northwest to southeast, the
fluid goes from a low gas–oil ratio (GOR) (<500 standard cubic feet per stock tank barrel (scf/STB))
Figure 2. Relative importance of mechanisms contributing to incremental oil recovery for steamflood
blackover
oil to a dry gas, as shown in Figure 3.
waterflood reprinted by permission [23] Taylor & Francis Ltd, http://www.tandfonline.com.
Figure 3. Initial
Initial gas-to-oil
gas-to-oil ratios
ratios (GORs) of Eagle
Eagle Ford
Ford wells.
wells. Source: U.S.
U.S. Energy Information
Energy Information
Administration [24].
Unlike conventional
Unlike conventional reservoirs,
reservoirs, deeper
deeper zones
zones have
have more
more gasgas due
due to the higher temperatures
temperatures
This is
needed for gas generation. This is called
called anan inverted
inverted petroleum
petroleum system.
system. Huff-n-puff
Huff-n-puff gas injection has
occurred from the black low-GOR oils to high-GOR volatile oils and
from the black low-GOR oils to high-GOR volatile oils and the gas the gas condensate
condensateregions of the
regions of
Eagle Ford. Therefore, we evaluated the recovery mechanisms as a function of
the Eagle Ford. Therefore, we evaluated the recovery mechanisms as a function of GORs rangingGORs ranging from
500 scf/STB
from to 20,000
500 scf/STB scf/STB.
to 20,000 scf/STB.
In addition
addition to the study of recovery mechanisms
mechanisms with
with respect
respect to GOR,
GOR, the sensitivity
sensitivity of the
recovery mechanisms
mechanismsto reservoir properties
to reservoir and huff-n-puff
properties operational
and huff-n-puff parametersparameters
operational is also investigated.
is also
We focus on the
investigated. Wecase with
focus onathe
GOR of with
case 1200 ascf/STB
GOR of and evaluate
1200 scf/STBa wide range of areservoir
and evaluate andof
wide range operational
reservoir
parameters
and to understand
operational parametershow the recoveryhow
to understand mechanisms are mechanisms
the recovery impacted. are impacted.
2. Reservoir
2. Reservoir Model
Model
For this
For this study,
study, reservoir
reservoir models were constructed
models were constructed to
to simulate
simulate the
the production
production from
from aa liquid-rich
liquid-rich
unconventional reservoir undergoing a huff-n-puff type gas injection process. The models
unconventional reservoir undergoing a huff-n-puff type gas injection process. The models were were used
used
to assess
to assess (1)
(1) the
the contributions from various
contributions from production mechanisms
various production mechanisms to to the
the incremental
incremental production
production
resulting from the gas injection, and (2) the sensitivity of various reservoir and gas injection
parameters to the production mechanisms.
Two reservoir models were used for the study: a larger-scale single-well sector model and a
smaller-scale quarter-fracture model based on the single well model. The single-well model was used
for the initial history match and investigation of the influence of the GOR on the EOR mechanisms.
Energies 2020, 13, 4944 5 of 24
resulting from the gas injection, and (2) the sensitivity of various reservoir and gas injection parameters
to the production mechanisms.
Two reservoir models were used for the study: a larger-scale single-well sector model and a
smaller-scale quarter-fracture model based on the single well model. The single-well model was used
for the initial history match and investigation of the influence of the GOR on the EOR mechanisms.
This model
Energies 2020, was alsoPEER
13, x FOR usedREVIEW
to study the effects of hydraulic fracture density on the EOR mechanisms.
5 of 24
The smaller quarter-fracture model was used to investigate the effects of additional reservoir and
EOR injection/production
EOR injection/production parameters
parameters on
on the
the EOR
EORmechanisms.
mechanisms. The
The development
development of of the
the models
models isis
discussedbelow.
discussed below.
2.1.
2.1. Single-Well
Single-Well Model
Model
The
Thesingle-well
single-wellmodel
modelsimulated
simulatedaa hydraulically
hydraulicallyfractured fracturedhorizontal
horizontalwellwellin
in aa low-permeability
low-permeability
liquid-rich
liquid-richunconventional
unconventionalreservoir.
reservoir.ItItrepresents
representsaatypical typicalwell
wellsuch
suchas asthose
thosecurrently
currentlybeing
beingdrilled
drilled
in the Bakken, Eagle Ford, Permian, or other source rock basins. Because
in the Bakken, Eagle Ford, Permian, or other source rock basins. Because most of the gas injection most of the gas injection
projects
projects ininunconventional
unconventional reservoirs
reservoirs occurred
occurred in in thethe Eagle
Eagle Ford,
Ford, typical
typical properties
properties from
from thethe Eagle
Eagle
Ford
Fordwere
wereused.
used. Furthermore,
Furthermore,the themodel
modelwas wasvalidated
validatedby byhistory-matching
history-matchingititto toan
anEagle
EagleFord
Fordcyclic
cyclic
gas
gasinjection
injectionfield
fieldpilot
pilotdataset.
dataset.
The
The model
model isisrepresentative
representative of ofaasingle
singlewell wellin inananeight-well-per-section
eight-well-per-sectionpattern.
pattern. ItIt is
is 10,550
10,550 ftft
(~2
(~2 miles)
miles)long
longandand650 feet
650 wide.
feet wide.TheThemodel
model is 72
is ft72thick, which
ft thick, is a typical
which valuevalue
is a typical for thefor
Lower Eagle
the Lower
Ford.
EagleTheFord.grid
Theblocks
grid are
blocks × 50
50 ftare 50ftft×× 850ft;ftthus, there
× 8 ft; thus, arethere
13 inare
the13x-direction, 211 in the211
in the x-direction, Y-direction,
in the Y-
and nine inand
direction, thenine
z-direction. There is aThere
in the z-direction. singleis9550a singleft long
9550horizontal well through
ft long horizontal the middle
well through of the
the middle
model (z = 5), leaving 500 feet on each end (Figure
of the model (z = 5), leaving 500 feet on each end (Figure 4). 4).
Figure 4. The horizontally gridded reservoir model represents a single well in an unconventional
Figure 4. The horizontally gridded reservoir model represents a single well in an unconventional
reservoir. A dual co-located grid system is used where the hydraulic fractures are in one grid and the
reservoir. A dual co-located grid system is used where the hydraulic fractures are in one grid and the
matrix properties are in the other grid and a transfer function allows interaction between the two grids.
matrix properties are in the other grid and a transfer function allows interaction between the two
The fractures are 250 feet apart from each other.
grids. The fractures are 250 feet apart from each other.
There are 40 hydraulic fractures along the wellbore modeled as planar, parallel, equidistant
There are 40 hydraulic fractures along the wellbore modeled as planar, parallel, equidistant
features 250 feet apart (Figure 4).While hydraulic fractures are often considered to have complexity
features 250 feet apart (Figure 4).While hydraulic fractures are often considered to have complexity
with fractures occurring in multiple directions, for gas injection, the fracture surface area in contact
with fractures occurring in multiple directions, for gas injection, the fracture surface area in contact
with the matrix is one of the most important attributes for recovering additional hydrocarbons. In this
with the matrix is one of the most important attributes for recovering additional hydrocarbons. In
model, simple planar fractures are used that have the same surface area as the more complex ones;
this model, simple planar fractures are used that have the same surface area as the more complex
thus, they can be considered as reasonable approximations [20]. For this model, the total surface area
ones; thus, they can be considered as reasonable approximations [20]. For this model, the total surface
of the fractures is 3.7 million ft2 . In a2 recent study, a core was taken from a well near a previous well
area of the fractures is 3.7 million ft . In a recent study, a core was taken from a well near a previous
that was hydraulically fractured [25]. In that study, they estimated the amount of fracture surface
well that was hydraulically fractured [25]. In that study, they estimated the amount of fracture surface
area in a modern multistage fracture stimulation, and it is comparable to the value used in this study.
area in a modern multistage fracture stimulation, and it is comparable to the value used in this study.
The permeability of the hydraulic fractures was determined through the history-matching process.
The permeability of the hydraulic fractures was determined through the history-matching process.
The hydraulic fractures are modeled as discrete fractures that interact with the matrix and wells
The hydraulic fractures are modeled as discrete fractures that interact with the matrix and wells
through the non-neighbor connections using a dual permeability formulation. The matrix properties
through the non-neighbor connections using a dual permeability formulation. The matrix properties
were assumed to be homogeneous throughout the model and set to 8% porosity and 100 nanodarcys
were assumed to be homogeneous throughout the model and set to 8% porosity and 100 nanodarcys
(nd) [26]. The initial water saturation was set at 37.5% which is the range of water saturations observed
(nd) [26]. The initial water saturation was set at 37.5% which is the range of water saturations
for the Eagle Ford [26]. Since, in many places, little water is produced from the Eagle Ford, we assumed
observed for the Eagle Ford [26]. Since, in many places, little water is produced from the Eagle Ford,
that the initial water was also the irreducible water saturation. This allowed the water pore volume to
we assumed that the initial water was also the irreducible water saturation. This allowed the water
pore volume to be removed from the model, improving the simulation performance and having
negligible effects on the results.
The pore volume of the fractures was controlled by setting the porosity values for the fracture
gridblocks (∅ ) to very small values using Equation (1).
∅ = ∅ , (1)
Energies 2020, 13, 4944 6 of 24
be removed from the model, improving the simulation performance and having negligible effects on
the results.
The pore volume of the fractures was controlled by setting the porosity values for the fracture
gridblocks (∅ gb ) to very small values using Equation (1).
w gb ∅ gb = w f ∅ f , (1)
where w
Energies gb is
2020, 13,the width
x FOR ofREVIEW
PEER the gridblocks, which in this case was 50 ft, and wf is the actual width of 6 ofthe
24
fracture, which was assumed to be 1/8 of inch in this case. A 24% fracture porosity (∅ f ) was assumed.
Therefore, the porosity, ∅ ∅gb ,, used
used in
in the
the fracture
fracture gridblocks
gridblocks was 0.00005 (0.005%) resulting in a 50 ft
gridblock representing a 1/8-inch fracture.
fracture.
Pressure-dependent permeability
Pressure-dependent permeability is is often
often used
used to simulate changes in permeability resulting
from pressure depletion in unconventional reservoirs
reservoirs [7,27–29].
[7,27–29]. This accounts for the micro and meso
fractures (less
fractures (less than
than 50
50 ft)
ft) in the
the reservoir
reservoir that
that close
close and
and reduce
reduce permeability
permeability asas pressure
pressure isis reduced.
reduced.
pressure-dependent permeability
Therefore, a pressure-dependent permeability function
function was was included
included inin the
the model
model for the matrix
blocks (Figure 5) and used as a history-matching
history-matching parameter. The The fracture
fracture gridblocks
gridblocks did not have
pressure-dependent permeability since it is so much greater than the matrix permeability that it had
insignificant impacts on the overall results. The curve shown in Figure 5 is the final history-matched
pressure-dependent permeability
pressure-dependent permeability function.
function. AsAsgasgasisis injected
injected back
back into
into the
the reservoir,
reservoir, it
it will likely
reopen these weak planes and increase permeability; thus, this pressure-dependent permeability is
also reversible.
Pressure Dependent Permeability

1.0
Permeability reduction factor
0.8
0.6
0.4
0.2
0.0
6000 5000 4000 3000 2000 1000 0
Pressure (psi)
Figure 5. As the pressure

pressuredrops
dropsininunconventional
unconventionalreservoirs,
reservoirs, permeability
permeability tends
tends to decrease
to decrease duedue to
to the
the closing
closing of meso-
of meso- andand microfractures.
microfractures. TheThe permeability
permeability reduction
reduction factor
factor is aisunitless
a unitless parameter
parameter that
that is
is multiplied
multiplied by by
thethe initial
initial absolute
absolute permeability
permeability to the
to get getpermeability
the permeability
for a for a given
given pressure.
pressure. This
This curve
represents
curve the values
represents found found
the values after history-matching.
after history-matching.
Figure 66shows
Figure showsthe relative
the permeability
relative for both
permeability the matrix
for both and the
the matrix fractures.
and Straight-line
the fractures. relative
Straight-line
permeability
relative is used for
permeability the fractures,
is used whereaswhereas
for the fractures, typical Corey-type curves are
typical Corey-type used
curves forused
are the for
matrix.
the
The matrix gas relative permeability in Figure 6a includes changes needed to history-match
matrix. The matrix gas relative permeability in Figure 6a includes changes needed to history-match the gas
injection
the process.process.
gas injection The capillary pressurepressure
The capillary was set was
to zero
set for this gas–oil
to zero for this system.
gas–oil system.
The initial reservoir pressure was 5500 pounds per square inch (psi). The producing bottom
hole pressure (BHP) was constrained to 500 psi, as we assumed that the well is operating efficiently.
During the injection cycles, the well was constrained to a gas injection rate of 1000 Mscf/day. The BHP
pressure increased during injection, but never exceeded 6500 psi. For the base model, which is a
1200 scf/STB GOR fluid, the bubble point pressure (Pb) was 3200 psi.
2.2. Fluid Model

For this research, the fluid model is one the most important aspects, and a significant amount of
effort was spent on the fluid description. Initially, we planned to create a compositional fluid model,
Figure 6. Relative permeability curves for (a) matrix and (b) fracture; kro is the oil relative
permeability and krg is the gas relative permeability.
The initial reservoir pressure was 5500 pounds per square inch (psi). The producing bottom hole
Pressure (psi)
Figure 5. As the pressure drops in unconventional reservoirs, permeability tends to decrease due to
the closing of meso- and microfractures. The permeability reduction factor is a unitless parameter that
is multiplied by the initial absolute permeability to get the permeability for a given pressure. This
Energies 2020, 13, 4944 7 of 24
curve represents the values found after history-matching.
Figure
but that type6 of
shows
modelthe relative
is not able permeability formechanisms
to evaluate the both the matrix and the
of viscosity fractures.
reduction orStraight-line
oil swelling
relative permeability
(expansion) is used
on recovery. for the fractures,
A properly whereas
constructed typical
black-oil Corey-type
model, however,curves
couldare used for
evaluate the
these
matrix. The matrix gas relative permeability in Figure 6a includes changes needed to
mechanisms along with vaporization and pressure support. Therefore, a black-oil formulation with history-match
the gas injection
dissolved process.
gas in the liquidThe capillary
phase pressure oil
and vaporized wasinset
thetogas
zero for this
phase wasgas–oil
used. system.
Figure 6.6.Relative
Relativepermeability
permeability curves
curves formatrix
for (a) (a) matrix
and (b)and (b) fracture;
fracture; krorelative
kro is the oil is the permeability
oil relative
permeability
and krg is theand
gas krg is thepermeability.
relative gas relative permeability.
Whitson and
The initial Sunjerga
reservoir [30] published
pressure was 5500the mole fraction
pounds per squarecompositions for plausible
inch (psi). The producingEagle
bottomFord in
hole
situ hydrocarbons
pressure (BHP) wasfor constrained
a wide rangeto of 500
GORs from
psi, which
as we we selected
assumed that three different
the well base fluid
is operating types for
efficiently.
our fluidthe
During models: a 1200
injection scf/STB
cycles, GORwas
the well black oil, a 4000to
constrained scf/STB volatile oil,
a gas injection and
rate of a1000
10,000 scf/STB The
Mscf/day. gas
condensate.
BHP pressure Compositions for the
increased during three different
injection, but never reservoir fluids
exceeded 6500were
psi.imported into model,
For the base a reservoir fluid
which is
property simulator,
a 1200 scf/STB GORPVTi [31],
fluid, thealong with
bubble the data
point from(Pb)
pressure differential
was 3200liberation
psi. (DL), constant composition
expansion (CCE), and constant volume depletion (CVD) experiments that were obtained from Whitson
and Sunjerga [30]. Each fluid was then matched to its GOR and bubble-point pressure (Pb) or dew-point
pressure (Pd), and the black-oil simulation properties were exported for live oil and wet gas using
the Whitson and Torp method [32]. To get a larger range of fluid types, the initial GOR was increased
and/or decreased for each base fluid, and additional cases were run for each of those. Table 1 shows
the GOR and respective bubble-point and dew-point pressures for all 12 cases.
Table 1. Initial gas–oil ratios (GORs) and bubble-point pressures (Pb) or dew-point pressures (Pd) for
the 12 different fluid models; 1200 standard cubic feet per stock tank barrel (scf/STB) is the base case.
Pb is for black oil and volatile oil fluids, whereas Pd is for gas condensate reservoirs.
Black Oil Volatile Oil Gas Condensate

GOR (scf/STB) 500 1000 1200 1400 1800 2400 3100 4000 6200 10,000 14,500 20,000
Pb/Pd (psi) 1900 2900 3200 3550 4100 4650 5100 5500 4450 4000 3300 2650
2.3. History-Matching Single-Well Model

The single-well model was history-matched to data from a gas injection huff-n-puff pilot well
in the Eagle Ford. This provided confidence that the properties used in the model were reasonable.
Parameters changed in the matching process included matrix permeability, hydraulic fracture
permeability, pressure dependent permeability, and relative permeability. The final history-match is
shown in (Figure 7) and it is very close to the field data.
The early peak production from the model was a little high. Early in the life of the well, when it is
flowing under natural energy, the BHP may be higher than 500 psi, which could be why the model
over-predicted the oil rates at early times. Since the wells were on for a few months before peak
production (when the model was started), the cumulative recovery is within 5% error. Most importantly,
the oil production during the gas injection phase matched very closely.
Table 2 lists the final properties used, which are the same as those mentioned in the previous
subsection. These values were used for all the models that evaluated the recovery mechanisms.
2.3. History-Matching Single-Well Model
The single-well model was history-matched to data from a gas injection huff-n-puff pilot well in
the Eagle Ford. This provided confidence that the properties used in the model were reasonable.
Parameters changed in the matching process included matrix permeability, hydraulic fracture
permeability,
Energies pressure dependent permeability, and relative permeability. The final history-match
2020, 13, 4944 is
8 of 24
shown in (Figure 7) and it is very close to the field data
Figure 7. Model
Model results
results (solid
(solid dark
dark green
green line)
line) history-matched
history-matched to to Eagle
Eagle Ford
Ford huff-n-puff
huff-n-puff gas injection
field pilot data from Hoffman
Hoffman [16][16] shown
shown with
with light
light green
green line
line and
andsymbols.
symbols.
Table 2. Reservoir
The early peak production model
from the properties.
model BHP, high.
was a little bottomEarly
hole pressure.
in the life of the well, when it
is flowing underProperty
natural energy, the BHP may be higher than Property
Value 500 psi, which could be why the model
Value
Matrix Permeability 100 nd Initial Pressure 5500 psi
Matrix Porosity 8% Model x-Dimension 650 ft
Number fractures 40 Model y-Dimension 10,550 ft
Fracture Half-length, xf 325 ft Model Thickness (z) 72 ft
Fracture Permeability 5 md Well Length 9550 ft
Fracture Spacing 250 ft Well Producing BHP 500 psi
Water Saturation, Sw 37.5% Injection Rate 1 MMcf/d
2.4. Quarter-Fracture Model

The single-well model was reconstructed into a quarter-fracture model to represent one side of
a half-fracture. This allowed for smaller gridblocks and enabled higher-resolution simulations that
were required for much of the sensitivity analysis. A fracture half-length of 325 ft was used from the
single-well model; therefore, the quarter-fracture model length was 325 ft. The distance between two
fractures in the single-well model was 250 feet; thus, the width of the quarter-fracture model was half of
that or 125 feet. The gridblocks parallel to the fracture were 1 ft wide; hence, there were 125 gridblocks.
Figure 8 shows the new higher-resolution quarter-fracture model. All the other parameters in the small
model were the same as the larger model.
The new model was run on primary production and with the same gas injection scenarios as the
single-well model except for the 1/160 the injection rate—to mimic the smaller scale. The production
from the small model was multiplied by 160 to get a rate that could be compared to the single-well
model. Figure 9 shows the production from the gas injection case for both models and the measured
data from the field trial that were used for history-matching the single-well model. The small model
produced slightly less oil than the large model, especially early on, but the small model continued to
match the field data well, even with no other changes to the model.
As an additional check on the model, the saturation and pressure changes in the near-fracture area
were examined for the injection and production periods (Figure 10). Prior to injection, gas had come
out of solution and acted as a primary solution gas drive. Once the gas saturation exceeded its critical
value, it started flowing in the reservoir such that the producing GOR started to increase (Figure 10A).
After gas injection started, the pressure started to increase, and the gas saturation decreased as gas was
forced back into solution. Once the pressure in the matrix was above the bubble-point pressure, then all
the gas was back into the oil phase (Figure 10B). When the well was put back on production, the pressure
dropped below the bubble-point pressure again, and a secondary solution gas drive mechanism was
Fracture Permeability 5 md Well Length 9550 ft
Fracture Spacing 250 ft Well Producing BHP 500 psi
Water Saturation, Sw 37.5% Injection Rate 1 MMcf/d
Energies
2.4. 2020, 13, 4944
Quarter-Fracture Model 9 of 24
The single-well model was reconstructed into a quarter-fracture model to represent one side of
ainitiated (FigureThis
half-fracture. 10C). As the next
allowed cycle ofgridblocks
for smaller injection began, the pressure
and enabled was raised, and
higher-resolution gas dissolved
simulations that
back into the oil. The process was repeated over the cycles of production and injection.
were required for much of the sensitivity analysis. A fracture half-length of 325 ft was used from the
After many
single-well model; cycles, however,
therefore, the bubble pointmodel
the quarter-fracture of the length
in-situ was
reservoir
325 ft.fluids increasedbetween
The distance becausetwo
the
mix of theingas
fractures theand oil changed
single-well model with
waseach
250cycle. Thisthe
feet; thus, is awidth
common behavior
of the describedmodel
quarter-fracture as the was
variable
half
bubble-point
of that or 125pressure
feet. The[33]. For late parallel
gridblocks cycles, some
to thefree gas remained
fracture were 1 inft the
wide;reservoir
hence, during injection
there were 125
(Figure 10D);Figure
gridblocks. this is partially
8 showsbecause the pressure
the new was not as high
higher-resolution as early injection
quarter-fracture cycles
model. Alland
thepartially
other
due to the variable
parameters bubble-point
in the small model werepressure.
the same as the larger model.
Figure 8. For grid sensitivity, a quarter-fracture model was created by cutting out a portion of the
Figure
Energies 2020, 8.
13,For grid
x FOR sensitivity, a quarter-fracture model was created by cutting out a portion of the9 of 24
PEER
original grid. The newREVIEW
model was 1/160th the size of the original model, and it was regridded in the
original grid. The new model was 1/160th the size of the original model, and it was regridded in the
x-direction down to 1 ft grid sizes. The grid sizes in the y-direction (50 ft) and z-direction (8 ft) remained
x-direction
produced down
slightly lesstooil
1 than
ft grid sizes.
the largeThe grid sizes
model, in the early
especially y-direction
on, but (50the
ft)small
and z-direction (8 ft)
model continued to
the same.
remained
match the fieldthedata
same.
well, even with no other changes to the model.
The new model was run on primary production and with the same gas injection scenarios as the
single-well model except for the 1/160 the injection rate—to mimic the smaller scale. The production
from the small model was multiplied by 160 to get a rate that could be compared to the single-well
model. Figure 9 shows the production from the gas injection case for both models and the measured
data from the field trial that were used for history-matching the single-well model. The small model
Figure 9.9.History-match
History-matchresults
results
for for
bothboth single-well
single-well modelmodel
(black)(black) and quarter-fracture
and quarter-fracture model
model (orange).
(orange).
As an additional check on the model, the saturation and pressure changes in the near-fracture
area were examined for the injection and production periods (Figure 10). Prior to injection, gas had
come out of solution and acted as a primary solution gas drive. Once the gas saturation exceeded its
critical value, it started flowing in the reservoir such that the producing GOR started to increase
(Figure 10A). After gas injection started, the pressure started to increase, and the gas saturation
decreased as gas was forced back into solution. Once the pressure in the matrix was above the bubble-
point pressure, then all the gas was back into the oil phase (Figure 10B). When the well was put back
on production, the pressure dropped below the bubble-point pressure again, and a secondary
solution gas drive mechanism was initiated (Figure 10C). As the next cycle of injection began, the
pressure was raised, and gas dissolved back into the oil. The process was repeated over the cycles of
production and injection.
solution gas drive mechanism was initiated (Figure 10C). As the next cycle of injection began, the
pressure was raised, and gas dissolved back into the oil. The process was repeated over the cycles of
production and injection.
After many cycles, however, the bubble point of the in-situ reservoir fluids increased because
the mix of the gas and oil changed with each cycle. This is a common behavior described as the
variable bubble-point pressure [33]. For late cycles, some free gas remained in the reservoir during
Energies 2020,injection
13, 4944(Figure 10D); this is partially because the pressure was not as high as early injection cycles 10 of 24
and partially due to the variable bubble-point pressure.
Figure 10. Gas saturation and pressure distribution for the quarter-fracture model at various times of
production and injection during the huff-n-puff process.
3. Methodology to Isolate Unique Recovery Mechanisms

After the model was validated through history-matching, it was used to assess the different
recovery mechanisms. This novel approach uses the flow simulator in a unique manner to model
phenomena that cannot be evaluated in real reservoirs. In the real reservoir, all of the mechanisms are
happening at the same time and it is impossible to separate the contribution of each one; however,
since the physical behaviors are simply equations in the simulator, they can be turned on and off in the
model to isolate the contribution of individual mechanisms. The paragraphs below describe in more
detail how the production for the various recovery mechanisms was isolated.
The amount of oil recovered due to swelling or expansion is equal to the change in the oil
formation volume factor. Therefore, if the oil formation volume factor term (Bo) is set to a constant
value corresponding to its average initial value for all pressures and solution GORs, then the change in
Bo is always zero, and the contribution from oil expansion is essentially turned off.
To turn off the vaporization mechanism, the vaporized oil–gas ratio term, Rv, was set to zero for
all pressures resulting in no liquids being vaporized into the gas phase and, therefore, there was no
contribution due to vaporization. When this happens, the recovery is only a function of the other
mechanisms. Similarly, the oil viscosity term (µo) can be set to a constant value that corresponds to
average initial viscosity. When gas dissolves into the oil, the viscosity in the model stays the same;
thus, there is no additional oil production from reducing viscosity.
For this evaluation, the production was simulated for a total of just over 5 years. Gas injection
occurred for the last 3 years and 2 months, mimicking the field test shown in Figure 9. The process
below outlines how the contributions from the different recovery mechanisms were determined for
varying fluid GORs. This process was also used for the sensitivity analysis.
A base case or primary model without gas injection was constructed for a given set of fluid
properties with a unique initial GOR and run on primary production for just over 5 years. In addition
to knowing the initial GOR, the initial values of Bo and µo were also known. The cumulative recovery
at the end of the model run was recorded, and a restart file was output when gas injection would have
started, in this case, 1 November 2014.
Next, a gas injection huff-n-puff case that included all recovery mechanisms and mimicked the
injection and producing times from the field trials shown in Figure 9 was run. Since the first 2 years of
production were the same as for the base case, the restart file was used to run the model starting in
November 2014. In this case, all the mechanisms contributed to the recovery. The cumulative recovery
at the end of the model run was also recorded. The difference in the production from the gas injection
case and the base case represents the total increase in production due to the gas injection.
Energies 2020, 13, 4944 11 of 24
To determine the contribution from pressure support, all the gas injection recovery mechanisms
except pressure support were shut off. This was done by setting Bo and µo to the respective constant
initial values and setting Rv to zero. The model was again restarted in November 2014 so that these
changes did not impact the primary recovery. The cumulative recovery at the end of the model run
(31 December 2018) was recorded. The difference between this cumulative recovery and the cumulative
recovery from the primary model is the contribution due to pressure support.
The contribution from the expansion mechanism was determined by setting Bo to the original
values and setting µo to a constant value and Rv to zero and running the model beginning in November
2014 using the restart file. The difference between this cumulative recovery and the cumulative recovery
from the base case run is the contribution due to oil expansion. This process was repeated for the
vaporization mechanism by setting Rv to the original values and setting Bo and µo to constant values.
For the viscosity reduction mechanism, µo was set to the original values and Bo to a constant value
and Rv to zero.
To further validate the method, the contributions from the four different mechanisms were
summed, which should be approximately equal to the difference between the full gas injection model
and the primary model. The relative importance of each mechanism can be determined by dividing
each individual contribution by the total additional recovery due to gas injection.
The results can also be checked by making three more runs. In these runs, all the mechanisms
are turned on except the one being investigated. The contribution of that mechanism is determined
by calculating the difference in cumulative recovery between that run and the run where all the
mechanisms are turned on, i.e., full gas injection case. Once all the contributions are calculated,
the relative importance of each one can be determined as before.
Due to the numerical nature of the solutions, the results should not be exactly the same, but they
should be close to each other, within 10% to 20%. If they are not, the models may have to be interrogated
more closely for potential issues.
Once the results for this set of properties or injection parameters are satisfactory, a new fluid
model is created with a different GOR and the above steps are repeated to find the relative importance
of the different mechanisms for this new model.
In these simulations, it is not possible to distinguish between the additional oil produced due to
liquid expansion versus the oil due to secondary solution gas drive. Both phenomena are related to the
oil formation volume factor and cannot be individually turned on or off. Therefore, when evaluating
the different recovery mechanisms, additional oil due to both liquid expansion and secondary solution
gas drive are reported together.
Example
An example of the process is shown below for a 1200 scf/STB GOR oil. It is based upon the initial
gas production from the field pilot test shown in Figure 9. This is also the model that was used to
history-match the field data. Figure 11 shows the results from running (1) the primary production case,
(2) the gas injection case, and (3) the injection case with the mechanisms turned off.
The difference between the primary production case (1) and the injection case with the mechanisms
turned off (3) is the contribution due to pressure support. Figure 12a shows the incremental production
due to oil swelling/solution gas drive (4), and Figure 12b shows the incremental production from
vaporization (5). A similar plot could also be created for viscosity reduction.
Additionally, three new runs were completed. This time, instead of comparing each run to
the “no mechanisms” case, they were compared to the “gas injection” case. Figure 13 shows how
these runs were used to determine the contribution of different mechanisms. Adding up these
differences is another way of determining the relative contribution of each mechanism and verifying
the previous results.
secondary solution gas drive are reported together.
Example
An example of the process is shown below for a 1200 scf/STB GOR oil. It is based upon the initial
gas production from the field pilot test shown in Figure 9. This is also the model that was used to
Energies 2020,history-match
13, 4944 the field data. Figure 11 shows the results from running (1) the primary production 12 of 24
case, (2) the gas injection case, and (3) the injection case with the mechanisms turned off.
Figure 11. Figure 11. Cumulative

Cumulative recovery recovery from flow
from basic basic simulation
flow simulation
runsruns
for for primaryproduction,
primary production, gas
gas injection,
Energies 2020, 13,injection,
x FOR PEER REVIEW
and gas injection with the mechanisms turned off. 12 of 24
and gas injection with the mechanisms turned off.
The difference between the primary production case (1) and the injection case with the
mechanisms turned off (3) is the contribution due to pressure support. Figure 12a shows the
incremental production due to oil swelling/solution gas drive (4), and Figure 12b shows the
incremental production from vaporization (5). A similar plot could also be created for viscosity
reduction.
Figure 12. Graphical explanation showing how to find the contribution of (a) oil swelling and (b)
vaporization.
Additionally, three new runs were completed. This time, instead of comparing each run to the
“no mechanisms” case, they were compared to the “gas injection” case. Figure 13 shows how these
runs were used to determine the contribution of different mechanisms. Adding up these differences
is another way of determining the relative contribution of each mechanism and verifying the previous
Figure12.
Figure 12. Graphical
Graphical explanation
explanationshowing how
showing to find
how the the
to find contribution of (a)ofoil(a)
contribution swelling and (b)and
oil swelling
results.
(b)vaporization.
vaporization.
Additionally, three new runs were completed. This time, instead of comparing each run to the
“no mechanisms” case, they were compared to the “gas injection” case. Figure 13 shows how these
runs were used to determine the contribution of different mechanisms. Adding up these differences
is another way of determining the relative contribution of each mechanism and verifying the previous
results.
Figure13.
Figure 13.AAdifferent
differentmethod
method to
to find the contribution
contributionofof(a)
(a)oil
oilswelling
swellingand
and(b)(b)
vaporization.
vaporization.
The
Thecumulative
cumulativeproduction
productionfor for each
each case and the
case and theabsolute
absoluteand andrelative
relativecontributions
contributions forfor each
each of of
thethe
mechanisms
mechanisms are shown in Table 3. The left side of the table (Method 1) shows the results whenthe
are shown in Table 3. The left side of the table (Method 1) shows the results when
mechanisms are turned
the mechanisms on one-by-one
are turned on one-by-one andandcompared
compared to the "no"no
to the mechanisms
mechanisms case.
case.ToTodetermine
determinethe
pressure support
the pressure contribution
support for these
contribution runs, the
for these runs,nothe
mechanisms
no mechanisms case wascasecompared
was compared to the to primary
the
primary
case. case.side
The Figure
right Theof right
the side of
table the table
shows the shows the comparison
additional additional comparison
(Method 2) (Method
where one2) mechanism
where one is
13. A different method to find the contribution of (a) oil swelling and (b) vaporization.
mechanism
turned off, andis the
turned off, and the
contribution wascontribution
determined was bydetermined
comparingby comparing
each eachfull
case to the case
gasto injection
the full gas case.
Toinjection
determine case.
the To determine
pressure the
support pressure
for thissupport
workflow,for this
the workflow,
primary the primary
cumulative
The cumulative production for each case and the absolute and relative contributions for each of cumulative
recovery was recovery
subtracted
wasthe
from
the subtracted
mechanisms from
gas injection,
are the then
and
shown gas injection,
3. Theand
the relative
in Table then
side ofthe
therelative
contributions
left of the
table contributions
three other
(Method of the
theresults
mechanisms
1) shows threewereother
when also
mechanisms
subtracted were
from this
the mechanisms also subtracted
arenumber. from this number.
turned on one-by-one and compared to the "no mechanisms case. To determine
theBoth
pressure
methodssupportgavecontribution
similar but for not these
exactly runs, the noresults,
the same mechanisms
which case was compared
is expected to the
with a numerical
primary case. The right side of the table shows the additional comparison (Method
solution. The general trends, however, do appear to be the same. For this low-GOR oil, the oil-swelling 2) where one
mechanismappears
mechanism is turnedtooff,beand
thethe
majorcontribution
source for was determined
the incremental by comparing
oil productioneach case
andto thetheother
full gas
three
injection case.
mechanisms makeTo determine the pressure
less meaningful support for
contributions. this workflow,
Method the primary
1 was used cumulative
for the results recovery
presented in the
wassection.
next subtracted from the gas injection, and then the relative contributions of the three other
mechanisms were also subtracted from this number.
Energies 2020, 13, 4944 13 of 24
Table 3. Results from simulation runs of 1200 scf/STB oil used to determine the relative contribution of different mechanisms.
Method 1 Method 2
Relative Relative
Cumulative Fraction of Cumulative Fraction of
Contribution Contribution
Recovery (STB) Contribution\(%) Recovery (STB) Contribution (%)
(STB) (STB)
Primary 125,700 – – Primary 125,700 – –
No Mechanisms 138,000 12,300 14.5 Pressure Support – 7600 9.0
Only Swelling 198,900 60,900 72.0 No Swelling 148,600 61,700 72.9
Only Vaporization 143,300 5300 6.3 No Vaporization 202,200 8100 9.6
Only Viscosity No Viscosity
144,100 6100 7.2 203,100 7200 8.5
Reduction Reduction
Gas Injection 210,300 84,600 100.0 Gas Injection 210,300 84,600 100.0
Both methods gave similar but not exactly the same results, which is expected with a numerical
solution. The general trends, however, do appear to be the same. For this low-GOR oil, the oil-
swelling mechanism appears to be the major source for the incremental oil production and the other
three mechanisms
Energies 2020, 13, 4944 make less meaningful contributions. Method 1 was used for the results presented
14 of 24
in the next section.
4. Results
4. Results
Results
Results from
fromboth
boththe study
the on on
study the the
impact of reservoir
impact fluid properties
of reservoir and theand
fluid properties sensitivity analysis
the sensitivity
of the reservoir properties and gas injection parameters are presented in this section.
analysis of the reservoir properties and gas injection parameters are presented in this section.
4.1. Reservoir Fluid

4.1. Reservoir Fluid Properties
Properties (GOR)
(GOR)
The process described in the previous section was completed for 12 different hydrocarbon systems
The process described in the previous section was completed for 12 different hydrocarbon
ranging from 500 scf/STB to 20,000 scf/STB. Each evaluation was carried out as explained for the
systems ranging from 500 scf/STB to 20,000 scf/STB. Each evaluation was carried out as explained for
1200 scf/STB oil. Figure 14 shows the total liquid recovery from primary production and for huff-n-puff
the 1200 scf/STB oil. Figure 14 shows the total liquid recovery from primary production and for huff-
gas injection
n-puff for the different
gas injection GORs. The
for the different casesThe
GORs. are cases
split into
are black oil, volatile
split into oil,volatile
black oil, and gasoil,
condensate
and gas
on the basis of the GOR. The incremental gas injection recovery is equal to the difference between
condensate on the basis of the GOR. The incremental gas injection recovery is equal to the difference bars
in Figure bars
between 14. in Figure 14.
Figure 14. Cumulative oil recovery from primary production and from gas injection as a function of
Figure 14. Cumulative oil recovery from primary production and from gas injection as a function of
initial GOR.
initial GOR.
Table 4 shows the percentage that each mechanism contributes to the recovery from gas injection,
Table 15A
and Figure 4 shows
showsthe
thepercentage that
same data in each mechanism
graphical form. contributes to the recovery from gas
injection, and Figure 15A shows the same data in graphical form.
Table 4. Percentage that each mechanism contributes to the recovery due to gas injection.
Table 4. Percentage that each mechanism contributes to the recovery due to gas injection.
GOR Pressure Support Oil Swelling Vaporization Viscosity Reduction
GOR Pressure Support Oil Swelling Vaporization Viscosity Reduction
500 21.9 65.8 2.5 9.8
500
1000
21.9
14.5
65.8
74.8
2.5 4.4 9.8 6.2
1000
1200 14.5
14.5 74.8
72.0 4.4 6.3 6.2 7.2
1200
1400 14.5
13.2 72.0
74.1 6.3 9.7 7.2 2.9
1800
1400 6.3
13.2 71.6
74.1 9.7 20.9 2.9 1.1
2400
1800 9.4
6.3 56.0
71.6 20.934.6 1.1 0.0
3100 4.5 55.4 40.2 0.0
2400
4000
9.4
5.3
56.0
43.2
34.651.6 0.0 0.0
3100
6200 4.5
8.5 55.4
10.6 40.280.9 0.0 0.0
4000
10,000 5.3
7.1 43.27.1 51.685.7 0.0 0.0
14,500
6200 6.7
8.5 10.66.7 80.986.7 0.0 0.0
20,000 6.3
10,000 7.1 7.16.3 85.787.5 0.0 0.0
14,500 6.7 6.7 86.7 0.0
20,000 6.3 6.3 87.5 0.0
This makes sense, but the results may be more extreme than expected. Since these are light oils,
viscosity reduction played only a minor role for low GOR oils but overall did not contribute
significantly for any case. Above a GOR of 1800 scf/STB, the pressure support contribution behaved
somewhat erratically. For the 3100 scf/STB GOR case, the pressure support contribution was less than
Oil swelling/secondary solution gas drive was most important for low-GOR oil and dropped off
5%; then,
Energies it bounced around between 5% and 9%. This may be due to the way the pressure support
13, 4944 15 of 24
as GOR2020,
increases. The opposite was true for vaporization; its importance increased as GOR increases.
contribution was calculated. At lower GORs, it contributed 10% to 20%.
This makes sense, but the results may be more extreme than expected. Since these are light oils,
viscosity reduction played only a minor role for low GOR oils but overall did not contribute
significantly for any case. Above a GOR of 1800 scf/STB, the pressure support contribution behaved
somewhat erratically. For the 3100 scf/STB GOR case, the pressure support contribution was less than
5%; then, it bounced around between 5% and 9%. This may be due to the way the pressure support
contribution was calculated. At lower GORs, it contributed 10% to 20%.
15. Contribution
FigureFigure 15. Contributionof four different
of four differentrecovery
recoverymechanisms forgas
mechanisms for gasinjection
injection
in in unconventional
unconventional
reservoirs: (A) relative percentage and (B) fraction of the incremental oil production.
reservoirs: (A) relative percentage and (B) fraction of the incremental oil production.
Oil swelling/secondary
From Figure 15A, it issolutioneasy to seegasthedrive was most
importance of oilimportant
swelling forfor low-GOR
low-GOR oil and dropped
hydrocarbons and
off as vaporization
GOR increases. The opposite
for high-GOR was15B
ones. Figure true hasforthevaporization;
same ratios for its theimportance
mechanisms, increased
but the y-axis as isGOR
the incremental
increases. This makes oilsense,
production
but the forresults
gas injection
may be over primary
more (the than
extreme difference between
expected. barsthese
Since in Figure
are light
14.). In the highest-GOR cases, there was less liquid hydrocarbon
oils, viscosity reduction played only a minor role for low GOR oils but overall did not contributeto recover for both primary and
EOR situations; thus, the incremental recovery for reservoirs with these
significantly for any case. Above a GOR of 1800 scf/STB, the pressure support contribution behaved fluid types was less.
Figure 15. Contribution
16 again hasofthe four different recovery mechanisms for gas injection in unconventional
somewhatFigure erratically. For the 3100 same ratios as
scf/STB GOR Figure 16
case, thebut now the
pressure y-axis
support is the incremental
contribution was recovery
less than
reservoirs:
due to gas(A) relative divided
injection percentageby and
the (B) fraction
primary of the incremental
recovery. This gives oil theproduction.
incremental percentage of
5%; then, it bounced around between 5% and 9%. This may be due to the way the pressure support
recovery due to huff-n-puff gas injection. Depending on the GOR, the additional recovery was
contribution
From
between
was
Figure calculated.
35%15A,and 45%,
At lower
it is easy
which to is
see GORs,
the it when
contributed
importance
reasonable
10%totofield
of oil swelling
compared
20%.
for low-GOR
case hydrocarbons
studies with 3 years of and
From Figure 15A, it is easy to see the importance of oil swelling
vaporization for high-GOR ones. Figure 15B has the same ratios for the mechanisms, but recovery
injection. In Figure 16, the relative contribution is also broken down for
by low-GOR
the four hydrocarbons
different the y-axis and
is
vaporization
mechanisms.
the incremental for high-GOR ones. Figure 15B has the same ratios for the mechanisms,
oil production for gas injection over primary (the difference between bars in Figure but the y-axis is
the
14.).incremental oil production
In the highest-GOR cases,forthere
gas injection
was lessover liquidprimary (the difference
hydrocarbon to recoverbetween bars in
for both Figure and
primary 14.).
In
EORthesituations;
highest-GOR thus,cases, there was less
the incremental liquidfor
recovery hydrocarbon
reservoirs with to recover
these fluidfor both
typesprimary
was less. and EOR
situations; thus, the incremental recovery for reservoirs with these fluid
Figure 16 again has the same ratios as Figure 16 but now the y-axis is the incremental recovery types was less.
due to gas 16
Figure again has
injection the same
divided by theratios as Figure
primary 16 but now
recovery. This thegivesy-axis
the is the incremental
incremental recovery
percentage of
due to gas injection divided by the primary recovery. This gives
recovery due to huff-n-puff gas injection. Depending on the GOR, the additional recovery the incremental percentage of recovery
was
due to huff-n-puff
between 35% andgas injection.
45%, which Depending
is reasonable on when
the GOR, the additional
compared to fieldrecovery was between
case studies with 3 35% yearsandof
45%, which is reasonable when compared to field case studies with
injection. In Figure 16, the relative contribution is also broken down by the four different recovery3 years of injection. In Figure 16,
the relative contribution is also broken down by the four different recovery mechanisms.
mechanisms.
Figure 16. Incremental recovery as a percentage of primary, further broken down by recovery
mechanisms.
While Figures 15 and 16 show a quantitative display of the results, Figure 17 represents a more
qualitative view that summarizes the impacts of the mechanisms. The y-axis (efficiency) represents
the EOR recovery divided by the primary recovery. This figure is similar to Figure 1 for thermal
recovery.
Figure 16.Incremental
Figure 16. Incrementalrecovery as aas
recovery percentage of primary,
a percentage further broken
of primary, down
further by recovery
broken down bymechanisms.
recovery
mechanisms.
qualitative view that summarizes the impacts of the mechanisms. The y-axis (efficiency) represents the
EOR recovery divided by the primary recovery. This figure is similar to Figure 1 for thermal recovery.
qualitative view that summarizes the impacts of the mechanisms. The y-axis (efficiency) represents
the EOR recovery divided by the primary recovery. This figure is similar to Figure 1 for thermal
recovery.
Energies 2020, 13, 4944 16 of 24
Figure 17. Summary of results for recovery mechanisms for huff-n-puff gas injection.
Figure 17. Summary of results for recovery mechanisms for huff-n-puff gas injection.
4.2. Sensitivity Analysis
4.2. Sensitivity Analysis
The sensitivity analysis was completed using the base case model with a 1200 scf/STB GOR fluid.
Themodel
Several sensitivity analysiscould
parameters was completed using the
have an impact on base case model
the recovery andwith
how a 1200
muchscf/STB GOR fluid.
each mechanism
Several model parameters could have an impact on the recovery and how much each mechanism
contributed. The following is a partial list of the parameters that may influence the recovery:
contributed. The following is a partial list of the parameters that may influence the recovery:
• Permeability;
 Permeability;
• Porosity;
 Porosity;
• Initial pressure;
 Initial pressure;
• Relative
 permeability;
Relative permeability;
• Critical
 Criticalsaturation;
gas gas saturation;
• Diffusion;
 Diffusion;
• Fracture
 surface
Fracture area;area;
surface
• Injection
Injection
 rate; rate;
•  Injection
Injection pressure;
pressure;
• 
Cycle Cycle
time;time;
• 
Total Total injection
injection time;time;
• 
BottomBottom
hole hole pressure;
pressure;
 Injection gas composition;
• Injection gas composition;
 Wellbore length.
• Wellbore length.
Some
Some ofof the parameters were
the parameters were notnot included
included in in the
the final analysis because
final analysis because they
they could
could notnot be
be
implemented in the software (e.g., diffusion) and other parameters had similar behavior
implemented in the software (e.g., diffusion) and other parameters had similar behavior to each other to each other
(e.g., injection
(e.g., injection rate
rate and
and injection
injection pressure),
pressure), but
but six
six different
different parameters
parameters were
were tested
tested and
and because
because some
some
parameters requiredtwo
parameters required twodifferent
different changes
changes (e.g.,
(e.g., matrix
matrix permeability);
permeability); thus, thus, there9 were
there were unique 9 models.
unique
models.To begin with, all the models were run on primary production. Next, the models were run with
To beginwith
gas injection with,allallthe
themechanisms
models were run on
turned primary
on. Figureproduction.
18 shows the Next, the models
recovery for thewere
base run
casewith
and
gas injection with all the mechanisms turned on. Figure 18 shows the recovery for the
the nine other models. The taller (blue) columns represent the total oil recovered at the end of the gas base case and
the nine other
injection, models.
and the shorterThe(red)
taller (blue) columns
columns represent
are for the the total
incremental oil recovered
recovery. at the end
The difference of the gas
between the
injection,
two bars is and
thethe shorter
amount of(red) columns
oil that wouldare forbeen
have the incremental
produced onrecovery.
primary The difference
for that case. between the
two bars is the amount of oil that would have been produced on primary for that case.
Energies 2020, 13, 4944 17 of 24
Figure 18. Total

Figure 18. Totaloil
oil(blue)
(blue) and
and incremental
incremental oil (red)
oil (red) recovered
recovered at endatofend of model
model dueinjection
due to gas to gas injection
process.
process.
Increasing the injection rate had the largest impact on overall recovery, which shows why getting
Increasing
gas into the injection
the reservoir rate had
is so important the largest
for these impact
types of on overall
processes. recovery,
Cases with which shows
more fractures why
and higher
getting gas into the had
matrix permeability reservoir is sooverall
increased important for these
production, types
but theirofincremental
processes. recoveries
Cases withwere
more notfractures
as great
and higher
because theymatrix permeability
produced a lot of oilhad increased
during overall
the primary production,
production butoftheir
stage incremental
development. Therecoveries
case that
were
ran thenot as great
longest because
also had a big they produced
incremental a lotand
increase, of oil during
the case withthe
theprimary production
smaller fracture stage of
area seems to
development.
be helped the Themostcase thatinjection,
by gas ran the longest
probablyalsobecause
had a big incremental
there increase,
was the most and the
amount case
of oil with the
remaining.
smaller
The fracture
sections belowarea seems
look to be
at how thehelped
recoverythemechanisms
most by gas are
injection,
affectedprobably because
by the various there was the
parameters.
most amount of oil remaining. The sections below look at how the recovery mechanisms are affected
4.2.1.
by theReservoir Rock Properties
various parameters.
The base case model had a matrix permeability of 100 nd. To evaluate the influence of matrix
4.2.1. Reservoir
permeability, two Rock Properties
additional models were created—one with a permeability of 1000 nd and one with
a permeability
The base case model had a are
of 20 nd. These within
matrix the expected
permeability permeability
of 100 range the
nd. To evaluate for influence
unconventional
of matrixoil
Energies 2020,
reservoirs. Figure13, x FOR PEER REVIEW 18 of 24
permeability, two 19A showsmodels
additional the cumulative productionwith
were created—one for both the primaryofand
a permeability gas
1000 ndinjection
and onecases.
with
a permeability of 20 nd. These are within the expected permeability range for unconventional oil
reservoirs. Figure 19A shows the cumulative production for both the primary and gas injection cases.
For the higher-permeability case, from the time gas injection started until the end of the run,
there was a lot of additional oil produced; however, during this time, the primary production also
continued to increase and, thus, the relative contribution was not as much. For the two lower-
permeability cases, the primary production leveled off; thus, the gas injection made a more significant
contribution. If the high-permeability cases were run longer, the gas injection would have even more
impact.
Figure 19B shows the relative influence of the different mechanisms. The two lower-permeability
cases were about the same, but the higher-permeability case was influenced more by vaporization
and oil swelling/solution gas drive and less by pressure support and viscosity reduction.
Figure
Figure 19.19.
(A)(A) Cumulative
Cumulative oiloil recovery
recovery forforcases
cases withdifferent
with different matrixpermeability
matrix permeabilityvalues,
values,and
and(B)
(B)the
the relative importance of different recovery mechanisms for models with different
relative importance of different recovery mechanisms for models with different matrix permeability. matrix
permeability.
For the higher-permeability case, from the time gas injection started until the end of the run,
4.2.2.
there wasFracture
a lot Surface Area oil produced; however, during this time, the primary production
of additional
also continued to increase
The area available and,to thus,
for gas the
interact relative
with the oilcontribution was not
in the matrix plays as much.
a significant roleFor thegas
in the two
lower-permeability cases,
huff-n-puff process; the primary
therefore, differentproduction leveled
cases were built off; thus,
to evaluate thethe gas of
impact injection made
the fracture a more
surface
area. The base case model had 40 hydraulic fractures with a total surface area of 3.7 million ft2 (40 ×
650 ft length × 72 ft height × 2 sides). Two additional cases were compared where the fracture area
was doubled and halved. This was done simply by adding fractures closer to each other (an 80-
fracture model) and removing fractures (a 20-fracture model).
Figure 20 shows the cumulative oil recovery forecasts for both primary and gas injection cases.
The larger-surface-area case produced more on primary and more during EOR as there was more
Energies 2020, 13, 4944 18 of 24
significant contribution. If the high-permeability cases were run longer, the gas injection would have
even more impact.
Figure 19B(A)
Figure 19. shows the relative
Cumulative influence
oil recovery for of thewith
cases different mechanisms.
different The two lower-permeability
matrix permeability values, and (B)
casesthe
were about the
relative same, but
importance of the higher-permeability
different case wasfor
recovery mechanisms influenced
models more by vaporization
with different matrix and
permeability.
oil swelling/solution gas drive and less by pressure support and viscosity reduction.
4.2.2. Fracture Surface Area

The area available for gas to interact with the oil in the matrix plays a significant role in the gas
huff-n-puff
huff-n-puff process;
process; therefore,
therefore, different
differentcases
cases were
were built
built toto evaluate
evaluate the
the impact of the fracture surface
area. The base case model had 40 hydraulic fractures with aa total total surface
surface area
area of
of 3.7 millionftft22 (40
3.7million (40 ××
650 ft length × 72 ft height × 2 sides). Two additional cases were compared where the fracture
ft length × 72 ft height × 2 sides). Two additional cases were compared where the fracture area area was
doubled and halved.
was doubled This was
and halved. Thisdone
was simply by adding
done simply fractures
by adding closer to
fractures eachto
closer other
each(an 80-fracture
other (an 80-
model)
fractureand removing
model) fracturesfractures
and removing (a 20-fracture model). model).
(a 20-fracture
Figure 20 shows the cumulative oil recovery forecasts for both primary primary and
and gas
gas injection
injection cases.
cases.
The larger-surface-area case produced more on primary and more during EOR as there was more
volume being accessed by the fractures. It It appeared
appeared to to increase
increase recovery
recovery atat the
the end
end ofof the
the run
run time,
time,
whereas the small fracture area case was close to its maximum recovery. recovery. This further shows, for both
primary and secondary recovery,
recovery, how
howimportant
importantititisisto
tomaximize
maximizefracture
fracturesurface
surfacearea.
area.
Cumulative recovery
Figure 20. Cumulative recovery for
for primary
primary and
and gas
gas injection
injection cases when models have different
amounts of fracture surface area.
area.
Figure
Figure 21
21 displays
displays the
the relative
relative contribution
contribution of of the
the different
different recovery
recovery mechanisms
mechanisms for for fracture
fracture
surface area. Both
surface area. Boththe
the base
base case
case andand
thethe
casecase
the the
withwith
twicetwice the fracturing
the fracturing showshow similar
similar results.
results. Since
Since recovery continued to increase, these cases did not reach maturity. For the case with half
recovery continued to increase, these cases did not reach maturity. For the case with half the fractures, the
fractures, more gas came into contact with the oil near the fractures; thus, this pattern became
more gas came into contact with the oil near the fractures; thus, this pattern became more mature, more
mature, whichthe
which caused caused the contribution
contribution due to vaporization
due to vaporization to increase
to increase some. some.
4.2.3. Gas Injection Parameters

In conventional reservoirs that are gas-flooded with either CO2 or natural gas, a common
surveillance measure is the gas utilization factor, which is the amount of gas injected per barrel of oil
produced [34]. This is a way of comparing pattern efficiency and evaluating the maturity of gas floods
across different fields. A low gas utilization factor indicates a more efficient process. The units are
often Mscf/stb and can be calculated as an instantaneous or cumulative value. For huff-n-puff gas
injection, the cumulative method is more appropriate. Figure 22 shows the gas utilization factor for the
base case gas injection model.
Energies 2020, 13, 4944 19 of 24
Figure 21. Relative importance of different recovery mechanisms for models with different amounts
of fracture surface area.

injection,
Figure
Figure the21.cumulative
21. Relative methodofof
Relativeimportance
importance isdifferent
more appropriate.
different Figure 22
recoverymechanisms
recovery mechanisms shows
forfor the
models
models gas
with
with utilization
different
different factor
of for
amounts
amounts
the base casesurface
gas
of fracture
fracture injection
surface area.model.
area.

injection, the cumulative method is more appropriate. Figure 22 shows the gas utilization factor for
the base case gas injection model.
Figure 22. Gas

Gas utilization
utilization factor
factor for
for huff-n-puff
huff-n-puff gas injection.
To drive
drive the the pressure
pressureabove
abovethethebubble
bubblepoint,
point,this
thisfield
fieldlocation
locationrequired
requireda significant amount
a significant amount of
gasgas
of forfor
fill-up;
fill-up;therefore,
therefore,the utilization
the utilizationstarted
startedout
outhigh
highandandslowly
slowlydecreased.
decreased. TheThe thick
thick black line
represents the
represents the3 3years of injection/production
years of injection/production shown in Figure
shown 7. The extended
in Figure pink thinpink
7. The extended line represents
thin line
the case with
represents the the
caselonger
with therunlonger
time. run
This caseThis
time. represents 5 more years
case represents 5 more(8years
years(8total)
yearsof huff-n-puff,
total) of huff-
with cycles of 2 months of injection followed by 2 months of production. As
n-puff, with cycles of 2 months of injection followed by 2 months of production. As with conventional with conventional gas
injection,
gas the project
injection, became
the project less less
became efficient overover
efficient time.time.
As the
thepattern
patternbecame
became more mature,
more the utilization
mature, tended tended
the utilization to increase. In conventional
to increase. reservoirs,
In conventional
a utilization factor below 10 is considered efficient by rule of thumb [34];
reservoirs, a utilization factor below 10 is considered efficient by rule of thumb [34]; however, however, of course, it can
of
vary widely Figure 22. Gas utilization factor for huff-n-puff gas injection.
course, it canon varythewidely
basis ofonthe
theproject
basis oflocation, the location,
the project cost of gas,
theetc.
costFor huff-n-puff
of gas, etc. For gas injectiongas
huff-n-puff in
unconventional
injection reservoirs, areservoirs,
in unconventional standard aefficiency
standardhas not been
efficiency hasestablished, but it maybut
not been established, be similar
it may beto
To drive the pressure above the bubble point, this field location required a significant amount
conventional gas floods.
similar to conventional gas floods.
of gas for fill-up; therefore, the utilization started out high and slowly decreased. The thick black line
The base case had a BHP of 500 psi. A second model was run where the BHP was set to 1000 psi,
represents the 3 years of injection/production shown in Figure 7. The extended pink thin line
but this had a very minor impact on all results. We were initially surprised by this small change;
represents the case with the longer run time. This case represents 5 more years (8 years total) of huff-
however, in low-permeability reservoirs, there was already a large pressure difference between the
n-puff, with cycles of 2 months of injection followed by 2 months of production. As with conventional
reservoir and the wellbore. Therefore, when the pressure drop was lowered by 500 psi, it did not seem
gas injection, the project became less efficient over time.
to make too much difference.
As the pattern became more mature, the utilization tended to increase. In conventional
During the sensitivity analysis, a model was run where the injection rate increased from
reservoirs, a utilization factor below 10 is considered efficient by rule of thumb [34]; however, of
1 MMscf/day to 2.5 MMscf/day. This caused more oil to be produced than changing any other
course, it can vary widely on the basis of the project location, the cost of gas, etc. For huff-n-puff gas
parameter in this study, which also further exemplifies how important it is to get gas into the
injection in unconventional reservoirs, a standard efficiency has not been established, but it may be
reservoir. High-injection-rate cases are common in many of the field trials and may be higher than
similar to conventional gas floods.
reservoir and the wellbore. Therefore, when the pressure drop was lowered by 500 psi, it did not
seem to make too much difference.
During the sensitivity analysis, a model was run where the injection rate increased from 1
MMscf/day to 2.5 MMscf/day. This caused more oil to be produced than changing any other
parameter
Energies in 4944
2020, 13, this study, which also further exemplifies how important it is to get gas into 20 ofthe
24
reservoir. High-injection-rate cases are common in many of the field trials and may be higher than 5
MMscf/day/well in some cases [35]. We tried increasing the injection rate further; however, due to
5limitations
MMscf/day/well
with the inmodel,
some cases [35]. We tried
2.5 MMscf/day wasincreasing
the highestthe injection
that could berate further; however, due to
achieved.
limitations with the model, 2.5 MMscf/day was the highest that could be achieved.
This also caused the vaporizing mechanism to be more significant. Its contribution went from
6% toThis
14% also
forcaused
the casethewith
vaporizing mechanism
the higher injectiontorate.
be more
In thissignificant.
case, more Itsgas
contribution
enters the went from but
reservoir, 6%
to 14% for the case with the higher injection rate. In this case, more gas enters the
only a certain amount can dissolve into the oil. The remaining gas stays in gas phase and is available reservoir, but only
atocertain
vaporizeamount can dissolve
the liquids. into the the
Subsequently, oil. contribution
The remaining from gas stays
both in gas reduction
viscosity phase andand is available
pressure
to vaporize the liquids. Subsequently, the contribution from both viscosity
support was reduced by about 4%, and oil swelling/solution gas drive contribution remained the reduction and pressure
support
same. was reduced by about 4%, and oil swelling/solution gas drive contribution remained the same.
Three
Three different
differentscenarios
scenarios were
wereevaluated
evaluatedwith withdifferent
differentinjection
injectionproduction
productioncycles.
cycles. The
The base
base
case had 2-month injection and 2-month production cycles. A case was completed
case had 2-month injection and 2-month production cycles. A case was completed that had 1-month that had 1-month
injection
injectionfollowed
followedby by22months
monthsof ofproduction,
production,and andthethethird
thirdcase
casehad
had1-month
1-monthinjection
injectionand
and1-month
1-month
production
productiontimes.times.
Figure
Figure23 23shows
showsthe theoil
oilproduction
productionresponse
responsefor forthe
thedifferent
differentscenarios.
scenarios.TheThecases
casesthat
thathad
hadequal
equal
time
time for injection and production performed similarly. The case that had less injection time than
for injection and production performed similarly. The case that had less injection time than
production
production time did not produce as much. In this case, the pressure could not be maintained in the
time did not produce as much. In this case, the pressure could not be maintained in the
model
modeland andless
lessgas
gaswas
wasable
abletotogo
goback
backinto
intosolution.
solution.
Figure 23.
Figure 23. Cycle
Cycle time
time sensitivity
sensitivity for
for huff-n-puff
huff-n-puffgas
gasinjection.
injection.
A
Aseparate
separatemodel
model was also
was constructed
also constructedto determine how longer
to determine injection
how longer times impact
injection the results.
times impact the
The newThe
results. casenew
had case
5 additional years of huff-n-puff
had 5 additional injection for
years of huff-n-puff a total of
injection for8ayears
total ofof injection,
8 years ofplus three
injection,
years of primary
plus three years production for a total offor
of primary production 11 ayears. For
total of 11the newFor
years. forecast,
the newtheforecast,
injectionthe cycle times were
injection cycle
2times
months
wereof injection
2 monthsand 2 monthsand
of injection of production.
2 months of Figure 24A shows
production. the 24A
Figure cumulative
shows the production
cumulativefor
this case.
production for this case.
As
As the
the patterns
patterns became
became more
more mature,
mature, thethe amount
amount of of gas
gas remaining
remaining in in the
thereservoir
reservoir each
each cycle
cycle
increased,
increased, as
as Figure 10 demonstrates.
demonstrates. ThisThishelped
helpedpromote
promotethe thevaporization
vaporization mechanism
mechanism such
such that,
that, as
as patterns matured, the contributions from oil swelling and
patterns matured, the contributions from oil swelling and solution gassolution gas drive mechanisms decreased
mechanisms decreased
and
and vaporization
vaporization became
became more
more important
important (Figure
(Figure 24B).
24B). Moreover,
Moreover,the thecase
casethat
thatwaswasinjecting
injectinglonger
longer
was
wasfurther
furtherinfluenced
influencedby bypressure
pressuresupport.
support.
A further model was completed where the higher injection rates were combined with the longer
run time (8 years of injection). For this case, the development strategy was 1 month of injection
followed by 2 months of production. When the higher rates were used, the best of both situations
could be achieved: a shorter injection time and, thus, less non-production time, as well as higher oil
rates during production. Figure 25 shows the oil rates for the two cases. The higher injection rates
made a significant difference.
Energies 2020, 13, 4944 21 of 24
Figure 24. (A) Cumulative oil recovery when gas injection process is extended for 5 additional years
(8 years total), and (B) relative importance of different recovery mechanisms.
couldFigure
be achieved:
Figure
24.24.(A) a shorter injection
(A)Cumulative
Cumulative time
oilrecovery
oil recovery and,
when
when gasthus,
gas lessprocess
injection
injection non-production
processisisextended
extendedtime,
for 5 as well as
5 additional
for higher oil
years
additional years
rates(8during
(8 years
years production.
total),
total), and(B)
and Figure
(B)relative 25 shows of
relativeimportance
importance the oil rates
of different
different for the
recovery
recovery two cases. The higher injection rates
mechanisms.
mechanisms.
could be achieved: a shorter injection time and, thus, less non-production time, as well as higher oil
rates during production. Figure 25 shows the oil rates for the two cases. The higher injection rates
Figure 25. Total oil production when

when combining
combining shorter
shorter injection
injection cycles
cycles with
with higher
higher injection
injection rates.
rates.
5.
5. Discussion
Discussion
These results were
These results were based
basedon ona amodel
modeland, and, as as such,
such, areare subject
subject to of
to all allthe
of potential
the potentialissuesissues
with
with
models. We tried to address some potential issues by making the model as realistic as possible,
models. We tried to address some potential issues by making the model as realistic as possible,
Figure 25. Total real
history-matching oil production
pilotwhen
data,combining shorter injectionmethods
cycles with higher injection rates.
history-matchingititto to fieldfield
real pilot and
data,using
andtwo different
using two differenttomethods determine tothe mechanisms.
determine the
However,
mechanisms. it isHowever, it is just a model, and it may not exactly represent the subsurface. This istoa
just a model, and it may not exactly represent the subsurface. This is a first attempt
5. Discussion
quantitatively
first attempt todetermine the importance
quantitatively determine of thevarious
importancemechanisms
of various for mechanisms
gas injection for in unconventional
gas injection in
These
reservoirs, results were
and, ifreservoirs, based
needed, itand, on
can be a model
improved and, as
upon such, are subject to all of the potential issues with
unconventional if needed, it can be in the future.
improved upon in the future.
models.
However, We tried to address some potential issues by making the model as realistic as possible,
However, this this model
model indicates
indicates that that there
there is is variability
variability in in the
the total
total recovery
recovery as as aa function
function of of
history-matching it
reservoir to real field parameters.
pilot data, and using two different methods to determine the
reservoir properties
propertiesand and operating
operating parameters. While While characteristics
characteristicsout out of
of our
our control,
control, such
such as as matrix
matrix
mechanisms. However,
permeability, it is justona model, and it may not exactly represent thewe subsurface. This is a as
permeability, have have aa bigbig impact
impact on the the results,
results, operating
operating conditions,
conditions, which which we can can control,
control, such
such as
first attempt to quantitatively determine the importance of various mechanisms for gas injection in
injection
injection rates
rates and and cycles
cycles times
times also
also have
have aa large
large effect
effect onon recovery.
recovery. HigherHigher injection
injection rates
rates improve
improve
unconventional reservoirs, and, if needed, it can be improved upon in the future.
production
production and
and cancan allow shorter injection cycles and longer
longerproduction cycles.
However, thisallow
modelshorter
indicatesinjection cycles
that there is and
variability production
in cycles. as a function of
the total recovery
Other recovery
Other properties mechanisms
recovery mechanisms such as changes in wettability or interfacial tensiontension
and counter current
reservoir and operatingsuch as changes
parameters. Whileincharacteristics
wettability or outinterfacial
of our control, suchand counter
as matrix
flow
currentdue to imbibition
flow due appear
to imbibition to be less important for gas injection processes, but they are likely toare
be
permeability, have a big impactappearon the to be less
results, important
operating for gas injection
conditions, which weprocesses,
can control, butsuch
theyas
significant for water or surfactant injection. We recommend further study of the recovery mechanisms
injection rates and cycles times also have a large effect on recovery. Higher injection rates improve
forproduction
these other andinjection
can allowstrategies, whichcycles
shorter injection may be anddonelonger inproduction
a similar manner
cycles. to the gas injection
mechanisms
Other recovery mechanisms such as changes in wettability or interfacial tensionreservoirs
in the current research. Furthermore, the nanopores in unconventional and counter likely
alter the phase
current flow due behavior due to the
to imbibition poreto
appear confinement effect [36],
be less important for gasandinjection
this impact shouldbut
processes, be they
addressed
are
in future work with regard to recovery mechanisms for gas injection in unconventional reservoirs.
Energies 2020, 13, 4944 22 of 24
6. Conclusions
Huff-n-puff natural gas injection has shown promise in unconventional reservoirs; for example,
pilot projects in the Eagle Ford have successfully produced more oil than primary production alone.
As gas injection EOR in unconventional reservoirs is becoming more mature, a better understanding of
the physical processes that cause the increased oil production is needed. While increased recovery
occurs, the physical mechanisms for the added production are not well understood. This paper uses a
model to show the importance of the different mechanisms and to demonstrate how the results change
as model parameters are varied.
1. This paper is a first attempt to quantitatively show the relative importance of different potential
recovery mechanisms for this process.
2. Recovery is a strong function of gas–oil ratio (GOR) with the highest total recoveries for black
oils (GOR: 500–2000 scf/STB); however, as a percentage of the primary recovery, all hydrocarbon
GORs produce about 40% more from gas injection than primary production over a 3 year period.
3. The quarter-fracture model shows that gas coming out of solution during the production cycles is
part of the reason more oil is produced when gas is dissolved into the oil phase. Since the model
cannot resolve this mechanism from the oil-swelling mechanism, the two were combined for the
purpose of this analysis.
4. Of the five mechanisms evaluated in this study, vaporization is the most important for high-GOR
reservoirs (6000+ scf/STB); oil swelling/secondary solution gas drive is the most important
for low-GOR reservoirs (500–2000 scf/STB). Pressure support played a minor role (~10%) for
many cases, and viscosity reduction had a very small impact for the lowest GOR reservoirs
(500–2000 scf/STB) and was insignificant for the other cases.
5. Other mechanisms (injection-induced fracturing, interfacial tension reduction, rock/fluid interactions,
wettability changes, etc.) may be important for recovery as well and should be further assessed,
particularly for other types of EOR in unconventional reservoirs.
6. Measures of pattern efficiency such as the cumulative gas utilization factor are important metrics
for understanding the effectiveness of the huff-n-puff gas injection process, and standards and
rules of thumb should be developed for unconventional reservoirs.
7. The largest impact on incremental production occurs when (1) the injection rate is increased and
(2) the overall time of the project is increased. In both situations, more gas is accessing the same
volume of reservoir. Other parameters such as increased matrix permeability and fracture surface
area increase overall production from gas injection; however, because the primary recovery also
increases in these cases, the incremental production due to gas injection is not as much.
8. More mature gas injection projects appear to be influenced more by vaporization and less by
the other mechanisms. More mature patterns are ones where there is more free gas present,
which often occurs later in the life of the project. The following are some reasons for a pattern
becoming more mature:
• Doing more cycles of injection/injecting longer;

• Injecting gas at a higher rate;
• Less fracture surface area.
Author Contributions: Both authors contributed equally to the model development, results, and write-up.
All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.
Energies 2020, 13, 4944 23 of 24
References
1. EIA. U.S. Field Production of Crude Oil. Available online: http://www.eia.gov/dnav/pet/hist/LeafHandler.
ashx?n=PET&s=MCRFPUS1&f=M (accessed on 4 August 2019).
2. Flannery, J.; Kraus, J. Integrated analysis of the Bakken petroleum system, U.S. Williston Basin.
American Association of Petrloem Geologists Search and Discovery Article 10105. 2006. Available online:
http://www.searchanddiscovery.com/documents/2006/06035flannery/ (accessed on 18 September 2020).
3. Downey, M.W.; Garvin, J.; Lagomarsino, R.C.; Nicklin, D.F. Nicklin Quick Look Determination of Oil-in-Place
in Oil Shale Resource Plays, Eagle Ford Shale. American Association of Petroleum Geologists (AAPG) Search
and Discovery Article No. 40764. In Proceedings of the AAPG Annual Convention and Exhibition, Houston,
TX, USA, 10–13 April 2011.
4. Gherabati, S.A.; Hammes, U.; Male, F.; Browning, J. Assessment of Hydrocarbon in Place and Recovery
Factors in the Eagle Ford Shale Play. SPE Reserv. Eval. Eng. 2018, 21, 291–306. [CrossRef]
5. Clark, A.J. Determination of Recovery Factor in the Bakken Formation, Mountrail County, ND. In Proceedings
of the SPE Annual Technical Conference and Exhibition, New Orleans, LA, USA, 1 January 2009; p. 10.
6. King, G.E. Maximizing Recovery Factors: Improving Recovery Factors in Liquids-Rich Resource Plays
Requires New Approaches. The American Oil and Gas Reporter, 1 January 2014.
7. Shoaib, S.; Hoffman, B.T. CO2 Flooding the Elm Coulee Field. In Proceedings of the SPE Rocky Mountain
Petroleum Technology Conference, Denver, CO, USA, 1 January 2009; p. 11.
8. Wang, D.; Butler, R.; Liu, H.; Ahmed, S. Flow Rate Behavior in Shale Rock. In Proceedings of the SPE Eastern
Regional Meeting, Morgantown, WV, USA, 1 January 2010; p. 10.
9. Alfarge, D.; Wei, M.; Bai, B. IOR Methods in Unconventional Reservoirs of North America:
Comprehensive Review. In Proceedings of the SPE Western Regional Meeting, Bakersfield, CA, USA,
23 April 2017; p. 24.
10. Kanfar, M.S.; Ghaderi, S.M.; Clarkson, C.R.; Reynolds, M.M.; Hetherington, C. A Modeling Study of EOR
Potential for CO2 Huff-n-Puff in Tight Oil Reservoirs-Example from the Bakken Formation. In Proceedings
of the SPE Unconventional Resources Conference, Calgary, AB, Canada, 15 February 2017; p. 12.
11. Jin, L.; Hawthorne, S.; Sorensen, J.; Pekot, L.; Kurz, B.; Smith, S.; Heebink, L.; Bosshart, N.; Torres, J.;
Dalkhaa, C.; et al. Extraction of Oil From the Bakken Shales With Supercritical CO2. In Proceedings of the
SPE/AAPG/SEG Unconventional Resources Technology Conference, Austin, TX, USA, 24 July 2017; p. 17.
12. Tovar, F.D.; Barrufet, M.A.; Schechter, D.S. Gas Injection for EOR in Organic Rich Shale. Part I:
Operational Philosophy. In Proceedings of the SPE Improved Oil Recovery Conference, Tulsa, OK, USA,
14 April 2018; p. 25.
13. Li, L.; Sheng, J.J.; Sheng, J. Optimization of Huff-n-Puff Gas Injection to Enhance Oil Recovery in Shale
Reservoirs. In Proceedings of the SPE Low Perm Symposium, Denver, CO, USA, 5 May 2016; p. 18.
14. Gala, D.; Sharma, M. Compositional and Geomechanical Effects in Huff-n-Puff Gas Injection IOR in Tight
Oil Reservoirs. In Proceedings of the SPE Annual Technical Conference and Exhibition, Dallas, TX, USA,
24 September 2018; p. 24.
15. Hoffman, B.T.; Evans, J.G. Improved Oil Recovery IOR Pilot Projects in the Bakken Formation. In Proceedings
of the SPE Low Perm Symposium, Denver, CO, USA, 5 May 2016; p. 22.
16. Hoffman, B.T. Huff-N-Puff Gas Injection Pilot Projects in the Eagle Ford. In Proceedings of the SPE Canada
Unconventional Resources Conference, Calgary, AB, Canada, 13 March 2018; p. 18.
17. Thomas, W.R.; Helms, L.W.; Driggers Timothy, K.; Trice, D.W.; Thomas, G.L. EOG Resources (EOG) Earnings
Call. 2016, Volume 6. Available online: https://investors.eogresources.com/events-and-presentations/event-
calendar/default.aspx (accessed on 23 May 2016).
18. Tovar, F.D.; Barrufet, M.A.; Schechter, D.S. Gas Injection for EOR in Organic Rich Shales. Part II: Mechanisms
of Recovery. In Proceedings of the SPE/AAPG/SEG Unconventional Resources Technology Conference,
Houston, TX, USA, 9 August 2018; p. 21.
19. Hamdi, H.; Clarkson, C.R.; Ghanizadeh, A.; Ghaderi, S.M.; Vahedian, A.; Riazi, N.; Esmail, A. Huff-N-Puff
Gas Injection Performance in Shale Reservoirs: A Case Study From Duvernay Shale in Alberta, Canada.
In Proceedings of the SPE/AAPG/SEG Unconventional Resources Technology Conference, Houston, TX,
USA, 9 August 2018; p. 25.
Energies 2020, 13, 4944 24 of 24
20. Hoffman, B.T.; Rutledge, J.M. Mechanisms for Huff-n-Puff Cyclic Gas Injection into Unconventional Reservoirs.
In Proceedings of the SPE Oklahoma City Oil and Gas Symposium, Oklahoma City, OK, USA, 8 April 2019; p. 13.
21. Wu, C.H. A Critical Review of Steamflood Mechanisms. In Proceedings of the SPE California Regional
Meeting, Bakersfield, CA, USA, 1 January 1977; p. 12.
22. Burger, J.; Sourieau, P.; Combarnous, M. Thermal Methods of Oil Recovery; Editions Technip: Paris, France, 1985.
23. Hoffman, B.; Kovscek, A. Efficiency and oil recovery mechanisms of steam injection into low permeability,
hydraulically fractured reservoirs. Petrol. Sci. Technol. 2004, 22, 537–564. [CrossRef]
24. Coleman, M. EIA updates Eagle Ford maps to provide greater geologic detail. In Today in Energy;
Energy Information Agency: Washington, DC, USA, 2015.
25. Raterman, K.T.; Farrell, H.E.; Mora, O.S.; Janssen, A.L.; Gomez, G.A.; Busetti, S.; McEwen, J.; Friehauf, K.;
Rutherford, J.; Reid, R.; et al. Sampling a Stimulated Rock Volume: An Eagle Ford Example. SPE Reserv.
Eval. Eng. 2018, 21, 927–941. [CrossRef]
26. Hammes, U.; Eastwood, R.; McDaid, G.; Vankov, E.; Gherabati, S.A.; Smye, K.; Shultz, J.; Potter, E.;
Ikonnikova, S.; Tinker, S. Regional assessment of the Eagle Ford Group of South Texas, USA: Insights from
lithology, pore volume, water saturation, organic richness, and productivity correlations. Interpretation 2016,
4, SC125–SC150. [CrossRef]
27. Cho, Y.; Apaydin, O.G.; Ozkan, E. Pressure-Dependent Natural-Fracture Permeability in Shale and its Effect
on Shale-Gas Well Production. In Proceedings of the SPE Annual Technical Conference and Exhibition,
San Antonio, TX, USA, 1 January 2012; p. 18.
28. Todd, B.J.; Reichhardt, D.K.; Heath, L.A. An Evaluation of EOR Potential in the Elm Coulee Bakken Formation,
Richland County, Montana. In Proceedings of the SPE Unconventional Resources Conference, Calgary, AB,
Canada, 15 February 2017; p. 16.
29. Jia, B.; Tsau, J.-S.; Barati, R. Investigation of Shale-Gas-Production Behavior: Evaluation of the Effects of
Multiple Physics on the Matrix. SPE Reserv. Eval. Eng. 2020, 23, 68–80. [CrossRef]
30. Whitson, C.H.; Sunjerga, S. PVT in Liquid-Rich Shale Reservoirs. In Proceedings of the SPE Annual Technical
Conference and Exhibition, San Antonio, TX, USA, 1 January 2012; p. 25.
31. PVTi. Simulation Software Manuals, Eclipse Pre-Processing Suite; Schlumberger: Houston, TX, USA, 2018.
32. Whitson, C.H.; Torp, S.B. Evaluating Constant-Volume Depletion Data. J. Pet. Technol. 1983, 35, 610–620. [CrossRef]
33. Kazemi, H. A Reservoir Simulator for Studying Productivity Variation and Transient Behavior of a Well in a
Reservoir Undergoing Gas Evolution. J. Pet. Technol. 1975, 27, 1401–1412. [CrossRef]
34. Whitson, C.H.; Brule, M.R. Phase Behavior; Society of Petroleum Engineers: Richardson, TX, USA, 2000.
35. Montanez. H-13 Certification: Enhanced Oil Recovery-Baker Deforest Unit; Texas Railroad Commission: Austin,
TX, USA, 2018.
36. Jia, B.; Tsau, J.-S.; Barati, R. A review of the current progress of CO2 injection EOR and carbon storage in
shale oil reservoirs. Fuel 2019, 236, 404–427. [CrossRef]
© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).

EWnergy

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

EWnergy

Uploaded by

Copyright:

Available Formats

energies

Keywords: EOR; cyclic gas injection; unconventional; recovery mechanisms; vaporization;

Energies 2020, 13, 4944; doi:10.3390/en13184944 www.mdpi.com/journal/energies

Energies 2020, 13, x FOR PEER REVIEW 4 of 24

Pressure Dependent Permeability

Figure 5. As the pressure

2.2. Fluid Model

Black Oil Volatile Oil Gas Condensate

2.3. History-Matching Single-Well Model

2.4. Quarter-Fracture Model

3. Methodology to Isolate Unique Recovery Mechanisms

Energies 2020, 13, x FOR PEER REVIEW 12 of 24

Figure 11. Figure 11. Cumulative

4.1. Reservoir Fluid

Figure 18. Total

4.2.2. Fracture Surface Area

4.2.3. Gas Injection Parameters

4.2.3. Gas Injection Parameters

4.2.3. Gas Injection Parameters

Figure 22. Gas

Figure 25. Total oil production when

• Doing more cycles of injection/injecting longer;

You might also like