You are on page 1of 42

Multiphase flow mobility impact on oil reservoir recovery: an opensource simulation

A. F. Britto,1, a) C. S. Vivas,2, b) M. P. Almeida,3, c) I. C. da Cunha Lima,4, d) and A. T.


da Cunha Lima5, e)
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

1)
Cimatec, INCT. Salvador, 41650-010, Brazil
2)
Cimatec. Salvador, 41650-010, Brazil
3)
UFC, INCT. Fortaleza, 60455-760, Brazil
4)
Cimatec, INCT, Pursuelife. Salvador, 41650-010, Brazil
5)
UFBa, Cimatec, INCT. Salcador, 41650-010, Brazil

(Dated: 9 March 2020)

This work uses Computational Fluid Dynamics (CFD) to simulate the two-phase flow
(oil and water) through a reservoir represented by a sandbox model. We investigated
the influence in the flows of water having higher and lower mobilities than oil. To
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

accomplish this, we also developed a dedicated solver, with the appropriated equa-
tions and representative models implemented in the open-source CFD OpenFOAM
platform. In this solver, the black-oil model represented the oil. The results show
that the Buckley-Leverett water-flood equation is a good approach for the three-
dimensional flow. We observe that the water wall front is mixed to some extent with
the oil and evolves obeying an exponential law. Water with mobility lower than oil
is not common. However, in this case, the oil recovery is improved and the amount
of injected water reduced. The results comparing different mobilities show that a
careful economic assessment should be performed before the field development.
We have shown that the water low mobility can increase, as in this studied example,
the water front saturation from 0.57 to 0.73, giving a substantial improvement in the
oil recovery. The reservoir simulation can provide all process information needed to
perform an economical assessment in a oil field exploration.

a)
ffbritto@hotmail.com
b)
caduvivas@gmail.com
c)
murilopereiradealmeida@gmail.com
d)
ivandacunhalima@gmail.com
e)
atcl@ymail.com

1
I. INTRODUCTION

Because of the high investment required in the discovery of new reserves and the costs
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

involved in the exploration, prediction tools became very important for the oil industry.
The increasing advance in high-performance computing permitted the Computational Fluid
Dynamics (CFD) to obtain numerical solutions of the equations that model the different
systems in an acceptable time interval, highlighting computer simulations as one of these
tools. Several methodologies4 related to reservoir management applications have been de-
veloped. Most times, however, it is difficult to apply them to real reservoirs possessing many
unknowns and particular features. In these cases, the real reservoir is replaced by a reference
model with known properties and down-scaled to a laboratory level.
Simplified models, which are a reasonable approach to the complex reality of an oil
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

reservoir, permit the exploration of the fundamentals of the oil flow inside the rock matrix.
It also contributes to better understand the influence of the control parameters on the oil
recovering process. As proposed by Bear6 , we chose the reference model called sandbox
because we can easily recognize the concepts of geometric similarity, kinematic similarity,
and dynamic similarity.
Oil reservoir simulation has been cited in many works13 as a tool for reservoir evaluation
and management to minimize costs and to increase process efficiency. At present we can find
several commercial softwares dedicated to oil reservoirs simulations such as STARS of CMG,
ECLIPSE-100 of Schlumberger, REVEAL of Petroleum Experts and UTCHEM from Texas
University at Austin. We used the CFD software OpenFOAM14 based on the finite volume
method of discretization. The main variables here are the pressure and the saturation of the
wetting fluid, calculated by the implicit pressure and explicit saturation (IMPES) method8 ,
where water pressure is the implicit variable and water saturation is the explicit variable.
The water-flood equation from Buckley and Leverett2,6,8–10,19 was made for one-dimensional
flow, but it is a good approximation for a three-dimensional domain. Also, in a water-flood
with the water mobility less than the oil, we have a substantial reduction of the amount of
the injected water, considering the same quantity of recovered oil.
The models used for representing permeabilities and capillary pressure are of fundamental
importance for investigating the mobility’s influence in the immiscible displacement of the
oil. In this work, we selected for the permeabilities the model from Stone11 as described in

2
the following section. This model also applies to three-phase flow. For the capillary pressure,
we used the empirical correlations from Ref.8 .
Enhanced Oil Recovery aiming to remove the oil still in place by changing the mobility
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

ratio between water and oil has been pursued as a response to economic demands. Many of
these efforts were dedicated to how to improve the oil secondary recovery in what is called
Chemical Enhanced Oil Recovery (CEOR)3 . Since a few decades ago until recent years
we can find many works dedicated to developing surfactants and polymers as agents for
CEOR1,12,20 . Some others are proposing Microbial Enhanced Oil Recovery (MEOR)16 and
in more recent developments the usage of nanoparticles has been the subject of extensive
works3,17,18,21,24,25 most of then based on experiments results.
In Section II we give a brief review of the porous media for a two-phase model, and
we present the parameters used in the sandbox model. To end this section we present the
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

equations implemented by the solver in the OpenFOAM software. In Section III we present
the results of the numerical simulations. Initially, we consider one- and two-dimensional
models and calculate the time series of the saturation. A plot is shown as well as snapshots
for times before and after the water-flooding. The step-like appearance of the saturation
function of time is discussed in further detail later. Several results compare the flow behavior
for water having mobilities lower and higher than oil. Discussing the consequences of that
behavior, as well as a deeper view of the water-front pushing oil are presented in Section IV.

II. THEORETICAL REVIEW

A. The Porous Media Model

The mass continuity equation for a fluid of density ρ through an isotropic porous medium
is:
∂(φρ)
= −∇.(ρu) + q (1)
∂t
where φ stands for the porosity, q is the specific discharge, meaning the rate of flow per unit
area (Q/A) and u is the flowing fluid velocity.
The Darcy’s equation6 is a linear response transport relation in porous media,

−K
q= (∇p − ρg∇z) (2)
µ

3
which, in general, is valid as long as the Reynolds number based on average grain diameter
does not exceed some value between 1 and 10. In this equation, K is the permeability of the
medium, ∇p the pressure gradient, g is the gravity and ∇z the gradient of z coordinate6 .
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

Substituting this equation into the continuity equation, we obtain:

∂(ρφ) (ρK)
= ∇.(∇p − ρg∇z) + q (3)
∂t µ
Eq. (3) governs the single-phase fluid flow. In two-phase flow, the two fluids fill entirely
the porous volume. The saturation of the wetting fluid and the non-wetting fluid corresponds
to the volume percentage of each fluid in the total porous volume. Thus, if Sw is the
percentage of wetting fluid (water) and So is the volume percentage of non-wetting fluid(oil),
we have:
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

Sw + So = 1. (4)

From now on we use subscript w for water and o for oil. Both fluids have a critical saturation,
called irreducible, or residual saturation that is the minimum saturation value that could
be achieved for each one. So, we name subscript r as the endpoint of the corresponding
variable. Then, Srw and Sro are endpoint saturations of water and oil, respectively.
Relative permeabilities11 are defined according to the following equations:

 eow
o So − Sorw
krow = krow (5)
1 − Srw − Sorw
 eog
o 1 − Sg − SLrg
krog = krog (6)
1 − SLrg − Srg
 ew
o Sw − Srw
krw = krw (7)
1 − Srw − Sorw
 eg
o Sg − SLrg
krg = krg (8)
1 − Srg − SLrg
   
o krow krog
kro = krow o
+ krw o
+ krg − (krw + krg ) (9)
krow krow
o o o o
In these equations, krow , krog , krw , krg , and the exponents eow, eog, ew and eg are empir-
ical parameters. They correspond to end-point of two-phase oil relative permeability flowing

4
with water, end-point of two-phase oil relative permeability flowing with gas, two-phase oil
exponent when the other phase is water, two-phase oil exponent when the other phase is
gas, endpoint relative permeability of phase water and end-point relative permeability of
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

phase gas, respectively. The residuals saturations are Srw for water, Sorw for oil and SLrg
for the total liquid. The calculated parameters are the two-phase water relative permeabil-
ity, the two-phase oil relative permeability, the water relative permeability, the gas relative
permeability, and the oil relative permeability, respectively. In spite of these equations are
applied for three-phase permeabilities, in our case, as there is no gas flowing, Sg = Srg =
0. In Ref.11 the results from these equations are compared with the experimental data from
Corey, with the following assumptions:

• The water permeability during oil/water flow is the same as the oil permeability during
oil/gas flow;
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

• The oil permeability during oil/water flow is the same as the gas permeability during
oil/gas flow.

The two-phase relative permeability parameters such as residual saturations, endpoint rel-
ative permeabilities, and exponents are based on the two-phase data of Corey. References15
and22 propose correlations for capillary pressure and permeabilities based on experimental
data. In this work, we used the correlation from8 . The results were more consistent when
compared with those in reference7 . Below we show the equation used to obtain the capillary
pressure correlation. In this model, the equation depends only on water saturation.

 
Sw − Swr − 0, 01
pcow = B1 ln (10)
1 − Swr

pcowmax − pcowmin
B1 =   (11)
0,01
ln (1−S wr )

In Eq. 10, pcow is the capillary pressure between the oil and water and in Eq. 11,
pcowmax and pcowmin are the maximum and minimum limits for pcow . By replacing the
appropriate variables in Eq. 1, we may write the continuity equation for the wetting and
non-wetting fluids:

∂(ρw φSw ) ρw krw K


= ∇.(∇pw − ρw g, ∇z) + qw (12)
∂t µw

5
∂(ρo φSo ) ρo kro K
= ∇.(∇po − ρo g∇z) + qo (13)
∂t µo
After an algebraic manipulation the time derivative of So can be eliminated. Then, the
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

continuity equations divided by ρw for the wetting phase depends on variables pw and Sw :

∇.((λw + λn )K∇pw + ∇.(λn K∇pcow ) −


qw qn
−∇.((ρw λw + ρn λn )Kg∇z) = − − (14)
ρw ρn

∂(φSw ) qw
= λw K∇.(∇pw − ρw g∇z) + (15)
∂t ρw
where the mobilities are defined as the ratio between relative permeability and the viscosity:
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

krw kro
λw = , λo = (16)
µw µo

B. Model Scale

The dimensions of a real oil reservoir are very large, ranging in a hundred meters. To
simulate the flow within the reservoir, one would expect to consider real dimensions, rock
geophysical parameters, and fluid properties. Considering this scenario, the simulation run-
time can be too long. To overcome this problem, models can be built to reduce the runtime
to a few hours. In a scaled model, reservoir dimensions, fluid properties, and rock properties
are scaled for the laboratory model so that the ratio of various forces in the reservoir and
the physical model are the same5 . Physical models can be classified into two categories:
(a) scaled model, and (b) elemental model. In a scaled model, reservoir dimensions, fluid
properties, and rock properties are scaled for the laboratory model so that the ratio of var-
ious forces in the reservoir and the physical model are the same9 . Therefore, it is essential
to manipulate the order of magnitude of the variables involved to reduce the time required
for the simulations. A crucial point is to seek a non-dimensional number shared by the
model and the real reservoir. Similar to the Brooks and Corey7 approach, we arrived to two
dimensionless parameters that should be equal for both model and real reservoir, provided
that the variables T0 , ∆p0 , K0 , ρ0 , µ0 and L0 are given in proper scales for the time, pressure,
permeability, density, viscosity, and length:

6
T0 K0 ∆p0 T0 K0 ρ0 g
Nr1 = , Nr2 = (17)
µ0 L 0 L 0 µ0 L0
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

Table I shows the values of these parameters for a typical reservoir:

TABLE I. Real reservoir parameters


Parameter Value

Water density 1000 kg/m3

Oil density 800 kg/m3

Water Viscosity 0.0025 kg/ms


PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

Oil Viscdosity 0.004 kg/ms

Gravity 9.81 m/s2

Absoluty Permeabillity 1000 md

Porosity 0.2

Initial Water Saturation, Swi 0.2

Residual Water Saturation, Srw 0.15

Residual Oil Saturation, Sro 0.81

The present simulations were performed in a sandbox model in the shape of a cube with
dimensions of 30m x 30m x 30m. The injection and production wells were located at the
same level at 17m from the bottom. After some test runs, we concluded that the entire
domain should be split in an 80 x 80 x 50 cells, in a total of 320.000. This is necessary
to have a consistent material balance and a very small change in results with cells number
increasing. The time step was 0.001 s. Using the parameters in Table I, L0 = 30m and
considering the result of the time for breakthrough of T0 = 60s for ∆p = 300psi, Nr1 = 0, 34,
and Nr2 = 4. × 10−6 . At t = 0, Sw = 0.2, pw = 2.3 × 107 Pascal. The pressure bottom
hole (pbh) for the production well was set to 2.1 × 107 Pascal. For ∆p = 400, 350, 300
psi, the pbh for the injection wells were set to 2.376 × 107 , 2.342 × 107 , 2.307 × 107 Pascal,
respectively.

7
C. Solver for the biphasic immiscible flow

We developed a new solver for the system of coupled differential Eqs. 14 and 15 and we
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

used the CFD software OpenFOAM14 to programming the solver code. OpenFOAM CFD
uses classes in C ++ language that are developed specifically aimed at this type of application
where a system of coupled differential equations needs to be solved numerically using the
finite volume method23 . The mathematical solution is calculated by the IMPES method
[5]. In Eq. 14 the water pressure is the implicit variable and water saturation is the
explicit variable. We implemented a black oil model for representing oil behavior. The
calculation was deployed in such a way that, at each time step, all pressure-dependent
properties must be updated for the entire domain.
We can define the water fractional flow in the following manner:
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

qw
fw = (18)
qw + qo
Provided that the interfacial tensions between the fluids are small and can be neglected, and
also neglecting the gravity effects in Eq. 2, we obtain:

λw
fw = (19)
λw + λo

The displacement of oil from a porous medium by water plays an important role in
producing petroleum9 . In both natural water-driven and secondary water flooding, this
process is fundamental. In the case of the one-dimensional flow of incompressible immiscible
fluids, we can formulate a simple mathematical description of this process. As the water
flows, it creates a front water saturation that is higher than the initial reservoir saturation,
and it establishes a discontinuity or an abrupt change in water saturation.
The fractional flow derivative obeys the following relationship:

 
dfw (fw )Swf
= (20)
dSw Sw =Swf Swf − Swi

Si is the initial water saturation of the oil field. The fractional flow derivative related to the
water saturation will have its maximum at Sw < Swf . Considering a differential element of
a unidimensional porous media and applying the continuity equation we obtain:

8
x   
qt dfw
= (v)Sw = (21)
t Sw φA dSw Sw

In this equation, qt is the total water injected, A is the cross-section and x is the distance.
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

This equation tells us how fast a given saturation Sw travels inside the domain.
In spite we have introduced the front displacement concept, depending upon the fluid
mobilities, the flow interface may not be sharp. As shown by Dake10 , the most favorable
condition is when we have a piston-like flow. For this to be true, we need the flow ahead of
the interface to be oil flowing at the initial water saturation or connate water, Swi . At this
point, we have the highest oil mobility, and we can define:

0 .
kro = kro(Sw =Swi ) (22)

In this case of low water mobility, behind the interface, water flows at oil saturation So = Sro .
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

Thus,

0
krw = krw(Sw =1−Sro ) (23)

For the flow to be a piston-like,

0
krw µn
0
= Mep ≤ 1 (24)
kro µw
Where Mep is called the endpoint mobility ratio. If Mep ≤ 1, it means that, under an
imposed pressure unbalance, the oil can travel in a velocity equal to, or greater than, that of
the water. Since the water is pushing the oil, there is no tendency for the oil to be by-passed
which results in the sharp interface between the fluids. However, Mep ≥ 1 is more common.
Here water can travel faster than the oil, and then the latter will be by-passed, leading to
an unfavorable water saturation profile.

III. RESULTS

We have performed simple procedures to validate the model before proceeding to a 3-D
simulation. First, we built a domain with only one row and 30 columns representing, as
close as possible, a 1D-domain. We assumed the water to have a lower mobility than the
oil.

9
Fig. 1 shows the time series of the water saturation at the production well, with an
applied ∆P of 400 psi. We can see an abrupt change in the saturation after 62-time units,
and then, from this moment on, only water is produced. Dake10 shows that we can calculate
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

the movable oil volume (MOV) according to the equation:

M OV = φV (1 − Sro − Srw ) (25)


PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

FIG. 1. Time series for water saturation at the production well

The displacement efficiency Ed is defined by Eq. 26:

S¯w − Swi
Ed = (26)
1 − Swi
All the contained oil in the reservoir was recovered in that 1D-model by the injection
of 1 MOV, considering the displacement efficiency equal to one. In the simulated case,
the MOV was 7.9 cubic meters. The total injected water was 10.3 cubic meters with a
displacement efficiency of 0.73 and so, in good agreement with the theoretical MOV. Fig.
?? shows domains for the one-dimensional and two-dimensional cases and snapshots of the
flow simulations before and after the water-flooding.
The plots of fw versus Sw for one-, two-, and three-dimensional cases, are coincident
showing that they only depend on the saturation. Fig. 3 shows the well-known S shape
fractional flow, the waterfront saturation having the same behavior in all dimensions. The
trivial solution of the Buckley-Leverett equation gives us two possible saturations for each
different location. Naturally, this is not possible. To overcome this problem, Pinder19 states
that such anomaly arises because of the dependency of dx/dt on saturation that allows a fast-
moving saturation to overtake a slow-moving-saturation. To eliminate this contradiction, we

10
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

(a)

(b)
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

FIG. 2. (a)1-D domain and snapshots before and after breakthrough. (b)2-D domain and snapshots
before and after breakthrough.

FIG. 3. General shape of the fractional flow versus saturation

require that there must be a shock front at which there is a discontinuity in the Sw function,
this way, creating the concept of a shock-front wave.
We performed the three-dimensional simulations with three values of the pressure unbal-
ance ∆P ( 300, 350 and 400 psi) and for water mobility higher and lower than oil. In every
iteration, we calculated the variables pw , Sw , fw , dfw /dSw and Ed . S¯w is the reservoir average
saturation and Swi is the initial water saturation. Fig. 4 show plots of the time series Ed for
each ∆P , in the cases of high and low mobilities. For high mobility, the plot was performed
until finishing the possible recoverable oil. Here, for the same time unit, less oil is recovered

11
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

(a) (b)

FIG. 4. Time series evolution of the efficiencies for water mobilities higher (a) and lower (b) than
that of oil.
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

that in the case with low mobility, and the simulation lasts much longer than the former
case. We show the accumulated and instantaneous oil production in Fig. 5.
We can calculate the waterfront saturation from Eq. 20. Fig. 6 shows the comparison of
the fractional flow for high and low mobility. The results are Swf = 0.73 and 0.57 for lower
and higher water mobilities, respectively. The former shows a better efficiency displacement
because in this case, the water is pushing the oil and not by-passing it, as shown in Fig. 6.
The S shape curves for the fractional water flow fw as a function of the water saturation
Sw do not depend on the pressure unbalance. However, as Fig. 6(a) shows, they are
different for water mobilities lower and higher than that of the oil. The same happens
with the derivative dfw /dSw . Fig. 7(a) and 7(b) demonstrates the advantage of the oil
recovery on low water mobility. In (a) we see the plot of the oil and water permeabilities.
The permeabilities curves are coincident for low and high water mobility because they only
depend on the saturation11 . However, mobilities also depend on the viscosities, as can
be shown in Figs. 7(a) and 7(b). If we reduce the water mobility, we have the oil flowing
ahead of the waterfront saturation with mobility higher than that of the water, which is
benefiting for the recovering oil efficiency.
According to the one-dimensional flow model developed by Buckley and Leverett, a water
wall front is created at the injection well, pushing the oil ahead. We have shown that the
model developed by Buckley and Leverett gives good results even in a three-dimensional
domain. A question arises about how important are capillary and gravity effects. Including

12
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

(b and d).
(c)

(a)
(a)

13
(b)
(d)
(b)

FIG. 6. Comparison of fractional flows (a) and derivatives (b).


FIG. 5. Oil productio (accumulated and instantaneous) for water high (a and c) and low mobility
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

(a) (b)

FIG. 7. Relative permeabilities (a) and mobilities (b).

these terms requires a great effort calculating the fractional flow derivative. We make a
considerable simplification if we consider these contributions negligible.
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

∇pc − ∆ρ∇z ≈ 0 (27)

IV. DISCUSSIONS

In most cases, water mobility is higher than that of the oil in the reservoir. This brings
undesirable effects on the flow because, as the water moves faster, it should by-pass the
oil, reaching more easily the producing well. However, in the opposite case, if we consider
the one-dimensional flow, for instance, we should expect that the water pushing the oil
ahead would create a sharp water-front with a water saturation of Swf = 1 − Sro . This was
confirmed, at first glance, by Fig. 1. However, expanding the time scale to better observe
how the water-front evolves, what we obtain is shown in Fig. 8. By fitting this plot into an
exponential curve, we obtain:

Sw = −0.03 + 0.24 exp[(t − 61.)/1.2] (28)

These results suggest that there is a region where we should have oil and water mixing to
some extent. This is possible if tiny droplets of water and oil coexist in the wall front.
Our simulations show that the water creates a spherical front that is deformed as shown
by the snapshots of Figs. 9(a) and 9(d). As the front approaches the producing well,
they form something like a water-spout. When the water-spout reaches the producing well,

14
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

FIG. 8. Time series for water saturation at the production well


PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

it establishes the breakthrough and reduces the oil recovery rate. In the very beginning,
the water spherical front is deformed by gravity. As it moves on, a spout is formed in
the gradient pressure direction. With water having higher mobility, the waterfront travels
faster and creates an interface with higher pressure, as shown in Fig. 9(a). In the case of
low water mobility, the waterfront travels slower and also at a slower pressure allowing for
a more efficient sweeping effect like in Fig. 9(d). In both figures, the contour filter from
P araview in the post-processing was applied to the Swf level. We can see in Fig. 9(d) that
the oil volume displaced by the water is greater than in Fig. 9(a). Both figures show the
moment where the waterfront saturation Swf touches the producing well.
In Figs. 9, we demonstrate how much efficient is the water-flooding at water low mobility.
In comparison, Fig. 9(e) exhibits a higher water saturation and a greater swept volume than
the Fig. 9(b), both at the time of the water breakthrough.
Beyond the breakthrough, we see that, in case of high water mobility, the simulation shows
that after the water creating a preferential path, oil no longer flows into the producing well,
and it is almost useless keeping the water injection, as seen in Figs. 5(a) and (c). Conversely,
it would be expected the oil flowing lasts much longer in the case of low water mobility, by
noting that in Figs. 5 (b) and (d) oil was still flowing through the producing well at the
time that simulation was interrupted. We made another profile comparison at the endpoints
of Fig. 5(a) to (d). Fig. 9(b) to (f) reinforce what was noted in Fig. 5(a) to (d).
In a water-flooding with high water mobility, the ratio oil − produced/water − injected

15
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

(a) (b) (c)


PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

(d) (e) (f)

FIG. 9. Snapshots of the water flood obtained with Paraview at times close to the water break-
through. Water high mobility (a) to (c) and water low mobility (d) to (f)

(a) (b)

FIG. 10. Produced oil/Injected water at the production well. (a) Water high mobility. (b) Water
low mobility.

16
decreases quickly after the water breakthrough, approaching to zero after a certain time.
This ratio decrease is faster as higher is the pressure difference applied. In the water-flooding
with low water mobility the recovered oil is greater because, after the water breakthrough, oil
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

still keeps flowing for a long time. In the Figs. 10(a) and (b) the ratio oil−produced/water−
injected is plotted for instantaneous and accumulated oil production.

V. CONCLUSIONS

Even though displaying a greater oil recovery in the water-flooding with low water mo-
bility, we should perform a very detailed economic assessment before the field development.
Depending upon the commercial commitment with oil production, we should consider the
lowest possible pressure differential. A cost-benefit analysis is anticipated as a tool for pro-
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

viding a suitable configuration for the field. The reservoir simulation can provide all process
information needed to perform this analysis. Most of simulators are commercial packages. In
this work we are using an open-source simulator. We introduced a tool in such way that we
can add any model for the black-oil relative permeability and capillary pressure. This way
anyone can freely configurates and customizes his/her model for fluid and rock properties.
OpenFOAM CFD comprises a very large users’ community that is always committed to how
to develop and deploy solutions for specific problems. So, by delivering this work we are
contributing with a tool that may be further improved by aggregating new functionalities.
As an open-source tool, the proposed solver can be adjusted and parameters customized, as
well as the capillary pressure and relative permeabilities models can be fine-tuned providing
a better representation of the studied phenomena.

VI. ACKNOWLEDMENTS

A.F. Britto acknowledges INCT-GP for a fellowship DTI-C from CNPq-Brazil.

REFERENCES

1
AZ Abidin, T Puspasari, and WA Nugroho. Polymers for enhanced oil recovery technology.
Procedia Chemistry, 4:11–16, 2012.

17
2
T. Ahmed. Reservoir Engineering Handbook. Gulf Professional Publishing, Burlington,
2010, 2010.
3
Jagar A Ali, Kamal Kolo, Abbas Khaksar Manshad, and Amir H Mohammadi. Recent
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

advances in application of nanotechnology in chemical enhanced oil recovery: Effects of


nanoparticles on wettability alteration, interfacial tension reduction, and flooding. Egyp-
tian journal of petroleum, 27(4):1371–1383, 2018.
4
Guilherme Daniel Avansi and Denis José Schiozer. Unisim-i: synthetic model for reser-
voir development and management applications. International Journal of Modeling and
Simulation for the Petroleum Industry, 9(1), 2015.
5
K. Aziz and A. Settari. Petroleum reservoir simulation. Applied Science Publ LTD,
London, 1979, 1979.
6
J. Bear. Dynamic of fluids in porous media. Dover Publications, New York, 1972.
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

7
R. Brooks and A. Corey. Hydrology properties of porous media. Hydrology Papers, 2(3):
10–23, 1964.
8
Zhangxin Chen, Guanren Huan, and Yuanle Ma. Computational methods for multiphase
flows in porous media, volume 2. Siam, 2006.
9
R. E. Collins. Flow of fluids through porous materials. Petroleum Publishing Co., Tulsa,
1976, 1976.
10
L. P. Dake. Fundamentals of Reservoir Engineering. Elsevier Publishing Company, Ams-
terdan, 1998, 1998.
11
M. Delshad and G. Pope. Comparison of three-phase oil relative permeabilities models.
Transport in Porous Media, (4), 1989.
12
WB Gogarty et al. Mobility control with polymer solutions. Society of Petroleum Engineers
Journal, 7(02):161–173, 1967.
13
Ali Goudarzi, Mojdeh Delshad, Kamy Sepehrnoori, et al. A critical assessment of several
reservoir simulators for modeling chemical enhanced oil recovery processes. 2013.
14
C. Greenshields. OpenFOAM Userguide. OpenFOAM Fundation LTD, 2015.
15
David D Huang, Matt M Honarpour, and Rafi Al-Hussainy. An improved model for relative
permeability and capillary pressure incorporating wettability. 9718:7–10, 1997.
16
R Khademolhosseini, A Jafari, and MH Shabani. Micro scale investigation of enhanced oil
recovery using nano/bio materials. Procedia Materials Science, 11:171–175, 2015.

18
17
Caetano Rodrigues Miranda, Lucas Stori de Lara, Bruno Costa Tonetto, et al. Stabil-
ity and mobility of functionalized silica nanoparticles for enhanced oil recovery applica-
tions. In SPE International Oilfield Nanotechnology Conference and Exhibition. Society of
This is the author’s peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

Petroleum Engineers, 2012.


18
Chegenizadeh Negin, Saeedi Ali, and Quan Xie. Application of nanotechnology for en-
hancing oil recovery–a review. Petroleum, 2(4):324–333, 2016.
19
George F Pinder and William G Gray. Essentials of multiphase flow and transport in
porous media. John Wiley & Sons, 2008.
20
David J Pye et al. Improved secondary recovery by control of water mobility. Journal of
Petroleum technology, 16(08):911–916, 1964.
21
Abbas Roustaei, Sadegh Saffarzadeh, and Milad Mohammadi. An evaluation of modi-
fied silica nanoparticles efficiency in enhancing oil recovery of light and intermediate oil
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002719

reservoirs. Egyptian Journal of Petroleum, 22(3):427–433, 2013.


22
SM Skjaeveland, LM Siqveland, A Kjosavik, WL Hammervold, GA Virnovsky, et al. Cap-
illary pressure correlation for mixed-wet reservoirs. 1998.
23
H. K. Versteeg and W. Malalasekera. An introduction to computational fluid dynamics:
the finite volume method. Pearson Education, London, 2007, 2007.
24
H Yousefvand and A Jafari. Enhanced oil recovery using polymer/nanosilica. Procedia
Materials Science, 11:565–570, 2015.
25
Magda I Youssif, Rehab M El-Maghraby, Sayed M Saleh, and Ahmed Elgibaly. Silica
nanofluid flooding for enhanced oil recovery in sandstone rocks. Egyptian Journal of
Petroleum, 27(1):105–110, 2018.

19

You might also like