SPE-175150-MS

Capillary Corrections to Buckley-Leverett Flow

Lichi Deng and Michael J. King, Texas A&M University

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Houston, Texas, USA, 28 30 September 2015.

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

Abstract
At the reservoir scale, multiphase fluid flow is well characterized by the Buckley-Leverett flow equations,
neglecting capillarity. However, as we extend our studies in higher resolution using multiscale calcula-
tions, or evaluate tighter or higher contrast heterogeneous or fractured reservoirs, capillarity becomes
increasingly important. To improve the understanding of these situations, we have extended the analytic
solution of the Buckley-Leverett equations to include capillarity. Specifically we have solved the
incompressible waterflood flow equations along a streamtube or streamline with arbitrary cross-section
for a heterogeneous porous media with variable injection water rate, including capillarity. The method-
ology is an application of a singular perturbation expansion with matched asymptotic solutions. The outer
solution is identical to the continuous portion of the Buckley-Leverett saturation profile while the inner
solution is the steady state solution first noted by Terwilliger experimentally. The two solutions match at
the Buckley-Leverett shock saturation and all solutions can be expressed in closed form. The result of this
analytical solution is tested against high resolution flow simulation to verify its validity. This analysis is
also applied to the calculation of capillary end effects in laboratory core floods, where the length scale of
the saturation correction can be predicted. We also demonstrate, as the shape of the saturation profile near
the front depends upon the capillary pressure function, that this analytical solution can used to interpret
experimental data and derive and calibrate the capillary pressure function.

Introduction
Immiscible displacement of oil by water or gas is a fundamental reservoir recovery mechanism which has
been discussed in numerous publications. Buckley and Leverett (1942) applied fractional flow theory to
immiscible displacement and developed the analytic solution for the saturation profile. The Buckley-
Leverett equation has been routinely used for the description of immiscible displacement at multiple
scales. In their paper they also discussed the function of capillary forces under different flow rates and
qualitative solutions were described.
In many expositions, especially at reservoir scale, capillary pressure is omitted from the immiscible
displacement theory as the viscosity/mobility ratio is the key factor that controls the efficiency and
stability of the displacement, especially after breakthrough of the injected fluid (Dake 1983, Chuoke et al.
1959, Welge 1952). The laboratory work of Terwilliger (1951) provided experimental verification of the
self-similar Buckley-Leverett saturation profile for gas displacing brine. In these experiments the steady
2 SPE-175150-MS

state front for the displacement has also been observed. This stable front solution follows Welges concave
envelope construction and has been studied in other papers. Jones-Parra and Calhoun (1953) calculated
the linear flood saturation profile using the stabilized zone method. McEwen (1959) and Fayers & Sheldon
(1959) presented numerical solutions to the displacement equation which included the effect of capillary
pressure, and exhibited this frontal steady state solution.
An analytical means of including capillarity in the analytic solution was presented by King and
Dunayevsky (1989) in their paper discussing waterflood stability. The concepts in that study form the
basis for this paper. We have resolved issues of closure of those equations through the use of a mass
balance relationship, and extended the solutions to variable rate and geometry. This has lead to the
introduction of a new dimensionless group which characterizes the strength of capillarity and which can
be used to distinguish viscous dominated flows with capillary corrections from capillary dominated flows.
We have applied the analytic solution to the calculation of laboratory coreflood capillary end effects, and
to the analysis of Terwilligers (1952) experimental saturation profiles, from which we can infer both
relative permeability and capillary pressure information.

Methodology

We will begin the discussion of the analytic solution with a review of incompressible fractional flow
theory. This will allow us to show its application to streamtubes, with variable flow geometry, and to
define notation for the analytic solutions.
Incompressible fractional flow for waterflood with capillary corrections
For incompressible waterflood in three dimensions,
(1)

(2)

The fractional flow of water may be defined if we express the water phase velocity in terms of the total
velocity, and normalize it to the total velocity.
(3)

The above equations may be expressed for flow within a streamtube, where x is the distance coordinate
along the tube.
(4)

(5)

(6)

The flux, qT(t), is related to the cross-sectional area A(x) and the velocity. Because we are restricting
our attention to incompressible flow, qT(t) does not depend upon position, but it may depend upon time.
(7)

For a streamtube, the flux and the cross-sectional area A(x) are inputs and the velocity is calculated
from this equation. When working with streamlines, the velocity and flux are inputs, and the cross-
sectional area may be inferred.
SPE-175150-MS 3

We are interested in solutions to these equations for capillary corrections, in which capillarity is, in
some sense, small. In the absence of capillarity, these fractional flow solutions will have a saturation
shock. Away from the shock, capillarity may be neglected in a consistent fashion. However, at the shock,
where and hence , the capillary pressure is not negligible. If we represent the capillary
pressure using a Leverett J-function, the spatial derivative of the capillary pressure can be expressed in
terms of the saturation gradient and the gradients in physical properties. Near the shock, the saturation
derivative dominates.
(8)

(9)

Here, Fw(S) is the fractional water mobility. It is also equal to the fractional flow in the absence of
gravity and capillarity. It is a function of the mobile saturation fraction, . We have

introduced a parameter (X) and a dimensionless capillary mobility function G(S). The parameter (X) is
small, in a sense to be determined as part of the solution of these equations.
(10)

(11)

We also define a time variable equal to the pore volume injected and a distance variable equal to the
mobile pore volume along the streamtube.
(12)

(13)

Combine the above derivation with Buckley-Leverett theory we can obtain the following capillary
corrections to Buckley-Leverett flow.
(14)

This equation is very general as it includes variations in flow rate, heterogeneity in porosity and
permeability, and the variation of cross-sectional area. As special cases it includes linear flow A
Constant and radial flow, A(x) 2xh or with a finite wellbore radius, A(x) 2(x rw)h.
Buckley-Leverett solution
Let us now consider the solution to Eq. (14) for waterflood in the absence of gravity and capillarity,
and with uniform initial saturation at the irreducible water saturation, S 0. This equation has a
continuous solution which describes a saturation profile for saturations above a shock saturation, S S*.
(15a,b)
4 SPE-175150-MS

The differential equation also supports discontinuous (weak) solutions. Consistency with the con-
tinuous solution determines the shock saturation, S*, and shock speed, c(S*).
(16a,b)

Here [Q] QLeft QRight signifies the discontinuity of a property, Q, across the shock. In addition to
this shock construction, there is an Entropy condition which selects among multiple possible shock
solutions (Osher 1984, Bell & Shubin 1985). In our context, both constructions are equivalent to obtaining
the shock saturation S*, which maximizes the shock speed, c(S*). This specification of the shock is also
applicable for tabular input to the fractional flow construction, where the local shock speed given by Fw
(S*) may not be well defined. This result may also be summarized as a concave envelope fractional flow
relationship.
(17)

This is Welges graphical construction and the self-similar Buckley-Leverett solution. The solution
satisfies the flux boundary condition at the water injector, fw 1, and the initial condition, S 0.
Still neglecting capillarity, if we consider the fractional flow relationship with gravity, we find that the
fractional flow need no longer be simply a function of saturation, as it has an additional term given by
. Even for a homogeneous medium, the dip angle, , and the cross-sectional

area, A(x), may depend upon location, and the flux may depend upon time. In this case, the Buckley-
Leverett equations may still be formulated, but there will not be a self-similar solution which depends
solely upon the dimensionless ratio X(x)/T(t). The exception is for simple geometries that may arise in
laboratory measurements or in mechanistic models, in which case the solution technique is identical to the
above but with a modified fractional flow. In the next section of this paper, we will neglect gravity and
focus on capillarity.

Matched asymptotic expansion

The fractional flow equation with a capillary correction, Eq. (14), is in the form where the highest order
spatial derivative has a small coefficient, and can be neglected except within the vicinity of a shock. This
kind of singularly perturbed differential equations have solutions that change rapidly in a narrow region
and can be analyzed using the method of matched asymptotic expansions (Hunter, 2004).
We will construct different asymptotic solutions inside and outside the region of rapid change and
match them together to determine a global solution. In our solutions, these are referenced as an outer
solution, SX, obtained away from the shock, an inner solution, Si, determined within the vicinity of the
shock, and a saturation at which they match, Sm. The global or composite solution, Sc, is expressed as,
(18)

We have introduced a new spatial variable that expands the length scale in the vicinity of the shock.
(19)

At the matched asymptotes, the inner limit of the outer solution ( 0, X L(T)) must match the
outer limit of the inner solution ( 0, ).
(20)
SPE-175150-MS 5

The composite solution is continuous and does not itself experience a shock. Away from the saturation
shock, the composite solution approaches the outer solution, Sc(X,T)SX(X,T) while near the shock it
approaches the inner solution Sc(X,T)Si (,T) and smoothly interpolates between the two.
Inner, outer and composite solutions The outer solution is obtained in the limit 0 and it is
essentially the continuous portion of the Buckley-Leverett profile.
(21)

(22a,b)

The inner solution is also obtained in the limit 0, but first the spatial scale must be expanded. The
inner differential equation may be derived using a change of independent variables from (X,t) to
and T(t). With this change of variable, and

. Here the speed is given by . As we are considering the limit of the inner solution,
is now a function of T: (X) (L(T)). The resulting differential equation for the inner solution is:
(23)

We can take the first integral and apply a saturation boundary condition at the foot of the inner profile,
to obtain the fractional flow relationship for the inner solution.
(24)

We see that the water flux increases linearly with saturation for the inner solution, just as in the concave
envelope of Welges graphical construction. We will also show that this linear increase will lead to the
steady state saturation profile first noted experimentally by Terwilliger.
We can integrate this first order differential equation:
(25)

We define an auxiliary function for the inner solution, H(S), in terms of this integral.
(26)

The integration of Eq. (25) is defined to within an arbitrary function of time, which can be specified
in terms of an unknown foot location, f(T), at which the composite solution vanishes, and at which the
inner solution takes on a small value, Sf(T).
(27)

This is the analytic inner solution expressed in the implicit form of X(Si,T). A mass balance relationship
of the composite solution will be used to determine the unknown functions of T in this solution. We will
show that it reduces to the steady state stabilized zone solution noted by Terwilliger. However, before
determining the mass balance we can complete the calculation of the match saturation.
From Eq. (22) we may take the inner limit of the outer solution ( 0, X L(T)) to obtain:
6 SPE-175150-MS

(28)

The outer limit of the inner solution ( 0, ) can only be achieved if the integrand in Eq.
(26) diverges.
(29)

These two equations are satisfied if the match saturation is the Buckley-Leverett shock saturation, Sm
S*. As a result, c is the shock speed and, L(T) cT is the shock location. Knowing the match saturation,
the water flux may also be expressed as a composite solution.
(30)

This recovers the concave envelope construction for the fractional flow. We may also approximate the
integral H(S), since the integrand has a double pole at S S*.
(31)

More generally, H(S) will be obtained by either analytic or numeric integration.

Mass balance closure The integral for the inner solution introduces unknown functions of T, which will
be determined by a mass balance relationship. To express the mass balance we need to introduce equations
for the saturations at the injector, X 0, and at the foot of the profile, X f(T).
(32)

(33)

The unknown position, f(T), and saturations Si and Sf may each be functions of T. From the outer
solution we can determine the position of the foot of the profile in terms of the other unknowns.
(34a,b)

We can then reference the inner solution to the foot saturation and position.
(35)

This expression describes the inner saturation profile: X(Si,T). We also obtain an equation for the
saturation correction at the injector in terms of Sf(T).
(36)

To leading order, Fw (SXf)c and we may take the limit Sf 0 to solve for Si. Because of the
double pole in this integrand, . The dimensionless group, , is the small

parameter which controls the validity of our perturbative expansion.

Notice that these equations will only have a solution so long as Fw (SXf) 0. This places an upper
limit on the magnitude of D, or equivalently, a lower limit on the water injection rate. Depending upon
the functional form of (L(T)), it may also place a lower limit on the value for T.
We need an additional equation from which to determine Sf(T). It is obtained from a mass balance of
the composite solution.
SPE-175150-MS 7

(37)

We substitute the outer solution and express the integrand as a function of saturation.
(38)

(39)

(40)

The integrand has a simple pole as S S* and so the integral diverges logarithmically. The function
of SXf approaches Sf S*Fw (S*) and vanishes as Sf 0. Hence . Again, the

dimensionless group D functions as a small parameter in these equations. We may approximate the
integral M(S), since the integrand has a single pole at S S*.
(41)

More generally, M(S) will be obtained by either analytic or numeric integration. Finally, we may
combine these equations.
(42)

These equations may first be solved for Si as a function of Sf, as shown in Fig. 1a for a specific value
of Sf. The intersection of the two terms implicitly determines the dimensionless group. Fig. 1b plots the
results as a function of the dimensionless group, and shows the range of inner saturations from inlet to
foot: (S * Si) Si Si. The outer solution also decreases from inlet to foot: 1 Sf (S* Sf).
The maximum value of the dimensionless group for which these equations close are given by the
saturation at the maximum value of Fw (SfX).

Figure 1Mass balance closure for the composite solution

8 SPE-175150-MS

This completes the derivation of all terms required for the composite solution.
Composite solutions Fig. 2 and Fig. 3 show the resulting composite saturation profile and fractional
flow relationships for homogeneous linear and radial models.

Figure 2Composite saturation profile and fractional flow plot for linear flow

Figure 3Composite saturation profile and fractional flow plot for radial flow

Dimensionless group D and small parameter

Notice that the composite saturation profile is now experiencing a smooth transition at the front
comparing to the shock front constructed from the traditional Buckley-Leverett method. The length scale
of the solution in the vicinity of the shock and its change with time are characterized by the small
parameter (X) or its dimensionless form D. For a linear geometry, (X) is independent of X while for
radial flow it increases linearly with X. Correspondingly, there will be a different scaling of the
dimensionless ratio with time in these cases. It will decrease with increasing T for the

linear geometry while it will be independent of T for radial flow.This implies that the dimensionless
group? D, scales inversely with injected volume for linear flow and thus the composite solution reduces
to the Buckley-Leverett solution at late time, irrespective of the strength of capillarity. For linear flow at
early time, the dimensionless group will increase beyond the range for which the capillary correction
solution is possible, indicating capillary dominance. In contrast, for radial flow, the dimensionless group
does not depend upon time and the impact of capillarity does not change with time. Fig. 4 shows the
relationship of the composite saturation profile vs. X/T for these two cases.
SPE-175150-MS 9

Figure 4 Composite saturation profile vs. X/T plots for linear flow and radial flow

The saturation profiles from Fig. 4 display the impact of the dimensionless group? D. For linear flow
the dimensionless group scales inversely with injected volume so the saturation front shrinks when T
increases. On the contrary, for radial flow, D does not depend on the volume injected thus the saturation
profile does not change with time.
Model Validation: Comparison against Simulation
In order to verify the validity of our analytical model, we compared the results of our model finite
difference flow simulation. The high resolution 1-D simulation was performed using Eclipse and the
following parameters were used.

The relative permeability is represented using a quadratic function and the capillary pressure relation-
ship is characterized by a logarithmic Leverett-J function. Both are functions of the normalized mobile
water saturation fraction. In the analytical model both the capillary pressure function and relative
permeability function are included within the derivation thus all derivatives and integrals can be
performed analytically. For the finite difference simulator, these functions are expressed in a tabulated
form. Capillary pressure and relative permeability curves are shown in Fig. 5.

Figure 5Relative permeability and capillary pressure curves (Sc1)

(43)
10 SPE-175150-MS

(44)

(45)

Both the analytical model and the flow simulation were tested under two different constant flow rates
qT 0.3 RB/day and qT 1 RB/day. Fig. 6 and Fig. 7 show the comparison results.

Figure 6 Analytical model and simulation results comparison for qT1 RB/day

Figure 7Analytical model and simulation results comparison for qT0.3 RB/day

From the above graphs its clear that the results from the analytical model are in close agreement with
the results from flow simulation. This verifies the validity of the analytical model. Another observation
that can be made from the above comparison is that at the foot of the saturation profile, the simulation
results have a slight incremental spread compared to the analytical results. This is due to the numerical
dispersion of the finite difference equation.
Model Application: Capillary End Effects
Another problem in which capillary corrections arise is at the outlet of a laboratory coreflood (Heaviside
& Black, 1983). The outlet boundary condition, pc (S) 0, follows from phase pressure continuity for
SPE-175150-MS 11

each of the two phases. If capillarity is neglected, then no additional boundary condition arises. However,
with the inclusion of capillarity in the description, the outlet saturation is fixed to a value S Sc for which
pc (Sc) 0. Instead of a moving shock, we now have a stationary boundary layer at the outlet, X L.
(46)

The development of the solution is similar to the above, although simplified since now c 0. The inner
limit of the outer solution is the Buckley-Leverett saturation at the outlet: L T. FW (Sm), so now the
match saturation is a function of T. The first integral of the inner solution must match the water phase
influx.
(47)

(48)

Unlike the moving boundary layer problem, there is no need for a mass balance constraint to close the
equations.
The results are shown in Fig. 8 for two different flow rates. For reference, the Buckley-Leverett
solution at the time of water breakthrough is shown. In both cases a water bank arises at the core outlet.
As the Buckley-Leverett outlet saturation increases, it will match and eventually exceed the outlet
saturation. At high saturations an oil bank is retained within the core. The relative magnitudes of these two
banks depend upon the wettability of the core.

Figure 8 Capillary end effects for two different flow rates

The integrand of Eq. (48) has a simple pole at S Sm which leads to a logarithmic saturation solution.
Inverted as a profile this indicates that the inner saturation correction decays exponentially with distance
away from the outlet, with a length scale inversely proportional to the flow rate. This is apparent when
contrasting the solutions in Fig. 8a and 8b. If we apply this analysis at the field scale near a producing
well, the flow rate is large over a small cross-sectional area, leading to a small dimensionless capillary
parameter. Although there may be a saturation bank due to the reservoir wettability, the length over
12 SPE-175150-MS

which this bank will form will be negligible. The mass balance integrand corresponding to Eq. (48) has
no pole, which indicates that the correction to the average saturation is , instead of the larger
correction of when the shock is still within the core.

We can evaluate the impact of the end effect on the laboratory determination of relative permeability
using the unsteady state JBN method (Johnson et.al. 1959). This analysis determines the relative
permeabilities using the fractional flow, the saturation, and the pressure gradient, all referenced to the core
outlet. Our analysis indicates that the end effect has no impact on the outlet fractional flow itself since
fwFw (Sm) for the inner solution. However, the outlet saturation is determined from the average
saturation, which does include the impact of capillarity. The mobility function is determined from the
outlet pressure gradient which itself is determined from the total pressure drop across the core, and is
impacted by capillarity. Based on the asymptotic analysis, we expect the error introduced into the inferred
outlet pressure gradient to be larger than that introduced into the outlet saturation, leading to an overall
suppression of the total mobility. Unsteady state laboratory procedures often involve a high speed bump
at the end of a laboratory test to reduce the length of the core which is impacted by capillarity, where an
increase in flow rate by a factor of 10 decreases the length scale of the capillary correction by a
corresponding amount.

Model Application Capillary Pressure Function Calibration

The shape of the composite saturation profile around the shock front depends upon the capillary pressure
function, so in principle we can calibrate the capillary pressure function if the saturation profile is
determined experimentally. We have already cited the early work by Terwilliger (1951) in which electrical
resistivity measurements were used to determine the saturation profile in a sand column for gas displacing
brine. More recent researchers have introduced other core flood saturation monitoring techniques, most
notably gamma attenuation measurements and X-Ray CT (Nicholls & Heaviside 1988, Wellington &
Vinegar 1987).
As a demonstration, we will analyze Terwilligers published experimental data to infer the capillary
pressure function and compare it with the conventionally measured capillary pressure curve reported by
Terwilliger. The flow direction in the experiment is strictly vertical and flowing downwards at a fixed
flow rate. In this case the outer solution will have a self-similar solution, based upon a fractional gas flow
which includes the effect of gravity.
(49)

The outer solution determines the main body of the saturation profile away from the saturation shock
and is controlled by gravity and viscous forces. The capillary force only impacts in the vicinity of the
saturation shock, and leads to a steady state solution. Thus a time lapse analysis method can first be used
to determine the parameters using the outer solution, while the shape of the profile can be analyzed using
the inner solution for the capillary pressure. In the time lapse analysis, we subtract the saturation profiles
at two different times and this data is used to calibrate the relative permeabilities, and any other required
parameters that are not otherwise specified. A non-linear optimization is needed to solve for the
reasonable parameters that would give the best match. These parameters include the two phase mobility
ratio, the mass balance time and the exponent n for the relative permeability function. For this calculation
we will utilize simple relative permeability and capillary pressure functions with few parameters.
(50a)
SPE-175150-MS 13

(50b)

(50c)

Two sets of experimental data from Terwilligers paper were used to demonstrate the application of this
analytical model. One has a liquid flow rate of 0.101 cm3/min and the other has a liquid flow rate of 1.538
cm3/min. At low flow rates, gravity dominates and this makes the time lapse method insensitive to the
relative permeability models characteristics. So the data with higher flow rate is used to infer the mobility
ratio factor and the exponent n for the relative permeability function.
Fig. 9 shows the result of the time lapse solution after the optimization for the higher flow rate case,
with a relative permeability exponent of n 2.9. The experimental data clearly shows the expected
fractional flow relationship with a steady state profile below the shock saturation and a self-similar
spreading solution above the shock. Fig. 10 shows the resulting fractional flow curve and its derivative
with respect to displacing phase saturation. Fig. 10a also includes the shock saturation construction and
Fig. 10b shows the saturation with maximum speed.

Figure 9 Time lapse model result and data after optimization

14 SPE-175150-MS

Figure 10 Fractional flow and fractional flow derivative curves with gravity

The two data sets are grouped into three pairs and the results of the outer solution time lapse match are
shown in Table 1. Only the relative times and injected volumes are provided and so a mass balance closure
has been used to infer the volumes of fluid injected, giving the calculated times shown in the table.

Table 1Time Elapsed Model Results for Different Groups of Data

Group 1 Group 2 Group 3

Liquid Flow Rate 1.538 cm3/min 0.101 cm3/min 0.101 cm3/min

Experiment Recorded Time 1.5 hr 52 hr 24 hr
5 hr 120 hr 52 hr
Calculated Effective Time 4.626 hr 81.6226 hr 59.0963 hr
7.9177 hr 140.2979 hr 81.9 hr

After determining the outer solution for each group, the inner solution including the effects of
capillarity is used to determine the capillary pressure function by matching the experimental saturation
profile with our composite solution. Here the process is essentially another optimization with the J_c
factor as the unklnown variable since all other parameters are given by the experimental settings. The
optimization objective function is the summation of the mismatch between the model result and the
experiment data from all three groups. After this second optimization, Jc 0.1019 is the converged
parameter value. Fig. 11 to 13 show the saturation profile comparisons between the composite saturation
solution and the experimental data for each group. Fig. 14 shows the resulting capillary pressure and
relative permeability matches.
SPE-175150-MS 15

Figure 11Composite saturation comparison with capillarity after optimization, Group 1

Figure 12Composite saturation comparison with capillarity after optimization, Group 2

16 SPE-175150-MS

Figure 13Composite saturation comparison with capillarity after optimization, Group 3

Figure 14 Capillary pressure and relative permeability comparisons between optimization result and experiment data

In the capillary pressure calculation, one assumption is that the threshold pressure is known. That is to
say the Leverett-J function should be expressed as . The reason for making this

assumption is that throughout the experiment no data was recorded after the gas breakthrough: the
capillary end effect was not present in the experiment. If further information is acquired, the threshold
pressure might also be determined.
The comparison result shows that the analytical model is capable of calibrating the capillary pressure
function from experimental saturation profiles. The remaining differences are probably due to the choice
of a logarithmic representation for the capillary pressure function, and the use of a single power for the
relative permabilities, which we will continue to investigate.
Discussion and Conclusions
We have provided an extension of the Buckley-Leverett solution for multiphase flow to include capillary
pressure corrections. The form of the solution is expressed in terms of pore volume coordinates which
captures linear, radial or more general streamtube geometries. The solution is expressed as a composite
of outer and inner solutions. The outer solution consists of the smooth portion of the Buckley-Leverett
profile while the inner solution consists of the stabilized zone steady state solution. The two solutions
SPE-175150-MS 17

match at the Buckley-Leverett shock saturation. The composite solution is a continuous solution which
smoothly transitions across the Buckley-Leverett shock. The solution has been validated using numerical
finite difference simulation.
As part of the analysis we have introduced a new dimensionless group that describes the relative
magnitude of capillary and viscous forces at the macroscopic scale. It differs from the capillary number
which describes the relative magnitude of these forces at the pore scale (Steigemeir 1977) and is more akin
to field scale dimensionless groups studied by other authors (Rashid et.al. 2012). It is interesting to note
that the current treatment which describes capillarity as a correction to viscous dominated flow cannot be
implemented if the capillary number is too large. For the examples studied herein, Fig.1b, the transition
from a capillary correction to capillary dominated flow occurs at a value close to 1. The range of solutions
for which capillarity can be treated as a correction require that Fw (SXf) 0, where SXf is the outer
solution at the foot of the saturation profile. In a reservoir context, this is usually the case, with the notable
exception of counter-current flow in fractured reservoirs, where this dimensionless group can be quite
large due to small total velocities.
The methodology has been applied to two laboratory scale examples. The first has been to the
prediction and analysis of the capillary end effect seen in core flood measurements. The second has been
to the use of the experimental saturation profile to infer both capillary pressure and relative permeability
information.

Acknowledgements
We gratefully acknowledge the support of the Foundation CMG, which has provided funding for the
RROM research chair at Texas A&M University. One of us (MJK) also thanks Dr. Dunayevsky for this
tutelage in the methods of analysis that have been applied in this study.
Nomenclature
S Mobile saturation fraction
Porosity
ui Phase velocity, m/s
uT Total velocity, m/s
A Cross-sectional area, m
qT Total flow rate, m3/s
i Phase mobility, m3sec/kg
T Total mobility, m3sec/kg
pc Capillary pressure, Pa or psi
p Density difference, kg/m3
g Acceleration due to gravity, m/sec2
Downwards unit vector
Interfacial tension, N/m
Fw Fractional water mobility
fw Fractional water flow
fT Foot location
c Buckley-Leverett frontal speed
SX Outer saturation
Si Inner saturation
Sc Composite saturation
S* Buckley-Leverett shock saturation
Sm Match Saturation
Si Inlet saturation correction
18 SPE-175150-MS

Sf Foot saturation correction

SXf Outer saturation at the foot
Sorw Residual oil saturation
Swirr Irreducible water saturation
x Distance, m
t Time, sec
X(x) Rescaled distance-mobile pore volume, m3
T(t) Rescaled time-volume injected, m3
G(s) Dimensionless capillary mobility function
H(s) Inner saturation function
G(s) Inner saturation mass function
X Small parameter, sec/m6
D Dimensionless capillary length scale group
Rescaled distance inner spatial variable
M Mobility ratio
n Exponent for relative permeability model
Jc Factor for Leverett-J function
Pth Capillary threshold pressure, Pa or psi

References
Bell, J. B., & Shubin, G. R. (1985). Higher-Order Godunov Methods for Reducing Numerical
Dispersion in Reservoir Simulation. Society of Petroleum Engineers. doi: 10.2118/13514-MS
Buckley, S. E., & Leverett, M. C. (1942). Mechanism of Fluid Displacement in Sands. Society of
Petroleum Engineers. doi: 10.2118/942107-G
Chuoke, R. L., van Meurs, P., & van der Poel, C. (1959). The Instability of Slow, Immiscible, Viscous
Liquid-Liquid Displacements in Permeable Media. Society of Petroleum Engineers. doi: 10.2118/
1141-G
Dake, L. P., Fundamentals of Reservoir Engineering, Elsevier, 1983
Fayers, F. J., & Sheldon, J. W. (1959). The Effect of Capillary Pressure and Gravity on Two-Phase
Fluid Flow in a Porous Medium. Society of Petroleum Engineers. doi: 10.2118/1089-G
Heaviside, J., & Black, C. J. J. (1983). Fundamentals of Relative Permeability: Experimental and
Theoretical Considerations. Society of Petroleum Engineers. doi:10.2118/12173-MS
Hunter, J. K. (2004). Asymptotic Analysis and Singular Perturbation Theory. Department of Math-
matics, University of California at Davis
Johnson, E. F., Bossler, D. P., & Naumann, V. O. (1959). Calculation of Relative Permeability from
Displacement Experiments. Society of Petroleum Engineers. doi: 10.2118/1023-G
Jones-Parra, J., & Calhoun, J. C. (1953). Computation of a Linear Flood by the Stabilized Zone
Method. Society of Petroleum Engineers. doi: 10.2118/953335-G
King, M. J., & Dunayevsky, V. A. (1989). Why Waterflood Works: A Linearized Stability Analysis.
Society of Petroleum Engineers. doi: 10.2118/19648-MS
McEwen, C. R. (1959). A Numerical Solution of the Linear Displacement Equation with Capillary
Pressure. Society of Petroleum Engineers. doi: 10.2118/1160-G
Nicholls, C. I., & Heaviside, J. (1988). Gamma-Ray-Absorption Techniques Improve Analysis of Core
Displacement Tests. Society of Petroleum Engineers. doi: 10.2118/14421-PA
Osher, S. (1984). Riemann Solvers, the Entropy Condition, and Difference Approximations. SIAM
Journal on Numerical Analysis, 21(2), 217235. doi: 10.1137/0721016
SPE-175150-MS 19

Rashid, B., Muggeridge, A., Bal, A. -L., & Williams, G. J. J. (2012). Quantifying the Impact of
Permeability Heterogeneity on Secondary-Recovery Performance. Society of Petroleum Engi-
neers. doi: 10.2118/135125-PA
Steigemeir, G.L. Mechanisms of Entrapment and Mobilization of Oil in Porous Media, in Improved
Oil Recovery by Surfactant and Polymer Flooding, Shah, D. D. and Schechter, R. S., eds.,
Academic Press, New York, 1977
Terwilliger, P. L., Wilsey, L. E., Hall, H. N., Bridges, P. M., & Morse, R. A. (1951). An Experimental
and Theoretical Investigation of Gravity Drainage Performance. Society of Petroleum Engineers.
doi: 10.2118/951285-G
Welge, H. J. (1952). A Simplified Method for Computing Oil Recovery by Gas or Water Drive.
Society of Petroleum Engineers. doi: 10.2118/124-G
Wellington, S. L., & Vinegar, H. J. (1987). X-Ray Computerized Tomography. Society of Petroleum
Engineers. doi: 10.2118/16983-PA