A p p l . sci. Res.

Vol. A I

CALCULATIONS ON THE FLOW OF

H E T E R O G E N E O U S MIXTURES THROUGH
POROUS MEDIA *)
by

H.

C. BRINKMAN

Laboratory N.V. De Bataafsche Petroleum Maatschappij

Summary
M u s k a t a n d M e r e s 1) h a v e f o r m u l a t e d b a s i c d i f f e r e n t i a l e q u a t i o n s
g o v e r n i n g t h e m o t i o n of h e t e r o g e n e o u s fluid m i x t u r e s t h r o u g h p o r o u s
m e d i a . T h e y o b t a i n e d a s o l u t i o n of t h e s e e q u a t i o n s b y n u m e r i c a l i n t e g r a t i o n
for t h e p r o b l e m of a c o l u m n i n i t i a l l y filled w i t h l i q u i d s a t u r a t e d w i t h gas,
w h i c h is closed a t one e n d a n d k e p t a t a c o n s t a n t low p r e s s u r e a t t h e
o t h e r 1).
Buckley
and Leverett2)
o b t a i n e d a n a n a l y t i c a l s o l u t i o n for t h e
p r o b l e m of a c o l u m n i n i t i a l l y filled w i t h l i q u i d w h i c h is f l o o d e d w i t h a
s e c o n d i m m i s c i b l e liquid. T h e y f o u n d a s o l u t i o n in w h i c h t h e s a t u r a t i o n
is a t h r e e - v a l u e d f u n c t i o n of t h e c o o r d i n a t e a l o n g t h e c o l u m n .
I n o u r p a p e r in t h e first place a d i s c u s s i o n is g i v e n of B u c k l e y a n d L e v e rett's solution. It appeared that the true solution which contains a discontin u i t y m a y b e d e r i v e d f r o m t h e t h r e e - v a l u e d s o l u t i o n b y a d i s c u s s i o n of t h e
integral relation which represents the total liquid recovery from the column.
T h i s d i s c u s s i o n b e a r s a f o r m a l r e s e m b l a n c e to t h a t o c c u r r i n g in t h e t h e o r y
of v a n d e r W a a l ' s e q u a t i o n of s t a t e .
T h e s e c o n d p r o b l e m t r e a t e d in o u r p a p e r is t h a t of a v e r t i c a l c o l u m n
i n i t i a l l y filled w i t h l i q u i d s a t u r a t e d w i t h gas u n d e r a h i g h p r e s s u r e , w h i c h
is o p e n e d a t its l o w e r end. F o r t h i s p r o b l e m we s u c c e e d e d in f i n d i n g a n
a n a l y t i c a l s o l u t i o n for low v a l u e s of t h e p r e s s u r e g r a d i e n t . H e r e a g a i n t h e
p a r a d o x of a t h r e e - v a l u e d s o l u t i o n o c c u r r e d a n d led t o a d i s c o n t i n u i t y in
the saturation.
A g e n e r a l a r g u m e n t is g i v e n to t h e effect t h a t t h r e e - v a l u e d s o l u t i o n s are
u n a v o i d a b l e for a t h e o r e t i c a l t r e a t m e n t b a s e d o n M u s k a t ' s e q u a t i o n s of
p r o b l e m s w h i c h lead t o d i s c o n t i n u i t i e s in t h e s a t u r a t i o n .

w 1. Introduction. The equations o/ [low. The flow of a homogeneous liquid or gas through a porous material may be described by
Darcy's law:
k
v = - - - grad p
(1)
*) Paper presented at the VIIth Congres of Applied Mechanics, London 1948.
Appl. sei. Res. A i

2l*
--

3 3 3 -

334

H.C.

BRINKMAN

where: v is the mean rate of flow, ~ the viscosity, grad p the pressure
gradient, and k the permeability of the porous material.
Relation (1) is a semi-empirical law which m a y be regarded as an
extension of Poiseuille's law b y the introduction of the empirical
constant k.
M u s k a t and M e r e s i) have further generalized (1) in order
to make it applicable to the flow of heterogeneous liquid-gas mixtures. T h e y introduced two equations analogous to (1) :
kL
v t = - - - - grad p
~t
v~ --

kG

grad p

(2a)
(2b)

where the indices l and g refer to the liquid and the gas phase
respectively, while L and G, called the relative permeabilities, are
two factors which were chosen so as to obtain a fit between the
experimental results and the equations (2a) and (2b).
It should be emphasized t h a t this is an entirely formal procedure;
(2a) and (2b) m a y be regarded as definitions of L and G. The experimental work of W y c k o f f
and B o t s e t 3 )
on the flow of
heterogeneous mixtures, indicated t h a t (2a) and (2b) m a y be very
useful for calculations, because of the fact t h a t L and G appeared to
have very simple properties, In a good approximation L and G
proved to be functions of the saturation only. The saturation S is
defined as the volumefraction of the pores filled with liquid.
At first sight this property of L and G seems to be rather surprising. L and G might be expected to be functions of m a n y of the properties of the fluids and of the porous.material, e.g. grad p, ~j/~, k
and the interfacial tension. We m a y be sure t h a t W y c k o f f and
B o t s e t's result only holds for a limited range of these variables.
However, as a detailed knowledge of the properties of L and G is
lacking, we based our calculations on W y c k o f f and 13 o t s e t's
experiments. Their curves for L and G are reproduced in fig. I. This
figure shows t h a t the L and G curves are very dissimilar. The G curve
shows an inflection point while the L curve does not. This is
probably caused by the different wetting properties of the liquid
and the gas. The liquid will wet the walls of the pores.
M u s k a t and M e r e s J) have applied (2a) and (2b) in order to

C A L C U L A T I O N S ON T H E F L O W O ~ H E T E R O G E N E O U S M I X T U R E S

335

treat the following problem : A column consisting of porous material

is initially filled with liquid, saturated with gas under a high pressure
The pressure at one end of the column is then lowered and kept at a
constant low value. A mixture of liquid and gas will leave the column
at this end and the course of pressure and saturation is calculated by
numerical methods. A reaHnsight in the consequences of the assumption of (2a) and (2b) can only be reached by obtaining analytical
solutions for some problems and by studying their properties. It was
therefore thought worth while to investigate some problems which
allow of an analytical solution. The fundamental problem how to
derive (2a) and (2b) from a detailed study of the flow of two phases

,I
permea3i//g/e.s
re/a b ~
l

---',5

F i g . 1.

through a pore system, is not discussed here.

The basic equations for our calculations are obtained by writing
down the equations of continuity for the liquid and the gas phase.
The equation for the gas is complicated by the fact that the gas may
dissolve into the liquid. As a result we have to distinguish between
the pure liquid and a liquid phase which is a mixture of pure liquid
and dissolved gas.
We introduce: [ - - the porosity of the medium; S - - the saturation; M - - the mass of the gas dissolved in unit volume of the liquid
phase; N - - the mass of the pure liquid in unit volume of the liquid
phase; Q~ the density of the gas phase. The quantities M, N and 0~
are functions of the pressure p.
The equations of continuity read:
div (Nvl) = - - 1

O(NS)
Ot

(3a)

336

H.c. BRINKMAN

div (~.v~) + div (My,) =

--la(MS)at

laEq,(lot--S)] (3b)

Substitution of vt and v, from ( 2 a ) a n d (2b) leads to two partial

differential equations for the saturation S and the pressure p. These
equations are non-linear. Nevertheless analytical solutions exist
for nontrivial cases.

w 2. The /low o/ two immiscible liquids. A very simple case is

obtained for the problem of two immiscible, incompressible liquids,
indicated as a and b. In this case the mass of dissolved material
M = 0, while the liquid densities of both phases are constant. We
discuss the one-dimensional case of a column initially filled with
liquid a into which liquid b is introduced at one end. The saturation
S is defined as the fraction of the pore volume filled with liquid a.
In the equations of flow (2a) and (2b) relative permeabilities A and
B instead of L and G have to be introduced. A and B are functions of
S like L and G. However, depending on the wetting properties of a
and b, the exact course of A and B m a y differ considerably from t h a t
of L and G.
Equations (3a) and (3b) then become, after substitution of (2a)
and (2b):
0 (kA op)

aS

apl

as

(4a)

(4b)

A solution of these equations was given b y B u c k 1 e y and L ev e r e t t 2). We will extend the discussion of the general solution of
the equations, as this will give an introduction into the more complicated problem of the next section.
Adding (4a) and (4b) gives

a /'A

~ ) ap [ _ o

(s)

or

op

ax

A/n,, + Bin,,

(6)

C A L C U L A T I O N S ON T H E F L O W OF H E T E R O G E N E O U S M I X T U R E S 3 3 7

where ck is t o t a l rate of flow of b o t h phases together.

S u b s t i t u t i o n of (6) into (4a) gives

I n t r o d u c t i o n of a function of the s a t u r a t i o n called F and defined

by
F(S) =

A/~la
A/n~ + B/no

yields

dF OS
/ OS
dS Ox -- kc Ot

(7)

The a p p r o x i m a t e course of the functions F and dF/dS has been

sketched in fig. 2.

l
0

--.,_s I

--'.S

F i g . 2.

The general solution of (7) is:

s =

x +-[-

ds t

(8)

where v/is an a r b i t r a r y function, to be d e t e r m i n e d from the initial

conditions.
L e t the column e x t e n d from x = 0 to x = X, and assume t h a t
initially S = [ for 0 < x < X , while at x = X liquid b is introduced into the column at the rate ck. Hence at t = 0, S is a function
of x as indicated b y the solid lines in fig. 3. According to (8) the argu 2
m ent of ~p will v a r y with t but at a different rate for different
Appl. sci. Res, A !

22

338

H. C. BRINKMAN

values of S, in consequence of the presence of the factor dF/dS."

As a result according to equation (8) after a certain time t the course
of S will have the shape as indicated in fig. 3 by the dotted curve.
This course of S is paradoxical, as it is a three-valued function for a
certain region of values of x. In this region the physical meaning of
the solution is not yet cleaI'.
It should be kept in mind, however, that initially S, considered as
a function of x, shows a discontinuity at x ---- X. This discontinuity
will move through the column at a certain rate. It will now be proved
that the true solution (indicated by crosses in fig. 3) has a discontinuity at x = Xl, where the value of x I is a function of the time which
has to be determined.

'I
/
/

.......
..-.,1

.-.... g.,,/

j C_
i

x~

'%,

~ x

Fig. 3.
The velocity of the discontinuity may be determined by considering two imaginary cross-sections through the column at both sides
of the plane of discontinuity x I. The change in liquid content of the
region bounded by these cross-sections can be determined in two
ways, on the one hand by applying (2a) and (2b) to a calculation of
the rate of flow through the cross-sections, on the other hand by
calculating the change in liquid content due to the movement of the
boundary x 1.
Equating the two results yields
/ (1 - - S) dxl
kB
C
d--t = - - ~b A/~h + B/rl--b"

(9)

This is an equation determining the rate of propagation of x 1.

However, a much easier way of determining x I is presented by considering the total quantity R of liquid b in the column:

R = / ) ( I - - s ) dx.
0

(I0)

CALCULATIONS

ON T H E F L O W O F H E T E R O G E N E O U S

MIXTURES

339

In fig. 4 R/] is indicated b y the shaded areas a for the three-valued

solution and b for the true solution. These shaded areas should be
equal. Therefore, the construction of the discontinuity at x I may be
performed as indicated in fig. 4c where the two shaded regions should
be of equal area.

~s

c?
---* .lr

,5

,s

]
C
"---'*X

Fig. 4.
It is easily proved that this construction is equivalent to a solution
of equation (9). In fact (9) is a differential equation derived from the
equations of flow, while (10) is an integral relation into which the
solution for S of the equations of flow has to be substituted. The
equivalence of (9) and (10) may be proved by straightforward calculation b y differentiating (10) with respect to t.
The paradox discussed in this paragraph bears a formal resemblance to that occurring in the theory of van der Waals' equation of
state. There also the discussion of an integral relation (f Vdp) leads
to an analogous geometrical construction.
An analogous paradox occurs in the theory of chromatography.

w3. A vertical column containing liquid and gas. In this section a

more complicated problem is considered, viz. a vertical porous

340

H . C . BRINKMAN

column initially filled with liquid s a t u r a t e d with gas, which column

is opened at its lower end. For this problem the equations of flow (2a)
and (2b) have to be extended so as to include gravitational forces.
The only way in which this can be done simply and consistently is:

kL

vl ~ - - - vg --

*h

kG

--

p + Q:g)

(l 1a)

(grad p + egg)

(l 1b)

(grad

where 0~ is the liquid density and g the acceleration of gravity.

In how far (1 l a) and (1 I b) are a good description of this t y p e of
flow remains to be determined by experiment..The precise meaning
of p, for instance, is a very difficult problem which involves a discussion of the interfacial tensions; p as well as S should be considered
to be mean values over a small volume element which still contains
m a n y pores.
The following assumptions about various physical properties of
liquid and gas are made:
H e n r y ' s law: M = cp;
Boyle's law: eg = cp;
and further: N = 0~ = constant.
The constants of B o y l e ' s l a w and H e n r y ' s law are taken to be the
same. This is often the case in actual gas-liquid mixtures.
W i t h these simplifying assumptions the introduction of (1 l a) and
(llb) into the equations of continuity (3a) and (3b) yields, for our
one-dimensional case:

O I L ( Op
)I ~h] OS
Sx
~x + Ng -- k St
Sx
- - p L ~ + Ng'

(op +

+ ~

\, Sx

(12a)

rh] op

(12b/

k St

The following b o u n d a r y conditions, determining the solution of

these equations, are chosen. At its upper end (x = X) the column is
assumed to be shut off. Therefore v~ ---- 0 and vg = 0 at x = X. This
condition is met b y taking S = 0 and sp/ax = --~g. Physically
this means t h a t at the upper end of the column an infinitesimal
region, where only gas is present, is formed; all liquid proceeding
downwards.

CALCULATIONS ON THE FLOW OF HETEROGENEOUS MIXTURES 341

At the lower, end (x = 0), where the column is open, a condition

determining the rate of flow of one of the phases might be chosen as
a boundary condition. Another possibility is to give the value of
@lax at x = 0. However, a different condition will prove to be more
expedient, as we shall see at a later stage in the discussion.
At first sight one might expect the search for an analytical solution of equations (12a) and (12b) to be a hopeless task. It is a surprising fact, however, that such a solution can be found if one further
assumption is made, viz. that the pressure gradient @lax is everywhere small compared to p / x . The assumption of a small value of
9p/gx means that at any moment the pressure p and therefore the
solubility M and the density of the gas Qgcan be taken to be constant
throughout the column. The value of ap/ox may be kept small in
practice by regulating the rate of flow at the lower end of the column.
The calculation is further simplified if we assume that the pressure
in the column is given as a function of t:

p = po ~

(13)

This means that the rate of pressure decrease ir~ the column has
been fixed. This rate, determines the rate at which the column is
depleted. It is related to the rate of saturation decline and, therefore,
to the rate of flow at the lower end of the column. Relation (13) may,
therefore, be accepted as a substitute for the missing boundary
condition at the lower end of the column.
With these assumptions we have introducr such simplification as
to-make the solution of the equations a relatively easy affair.
Equation (12b) can be integrated with respect to x from an arbitrary value of x to x = X . As the left hand member of (12b), is a
derivative with respect to x, while the right hand member does not
depend on x, according to our assumptions, the integration yields an
equation from which we obtain the following expression for ap/ax

~p _ (~,/a/k ) ( X - - x) - - L N g - - G (~/~g) cpg

Ox
L + (~7,1~7~)C

(14)

This expression for @lax is now introduced into equation (12a).

A differential equation for S is obtained:
k

st -

ax

(bp - - a)

(15)

342

H.C. BRINKMAN

where H is a function of S" H = 1 + ,hG/,igL, while b and d are

constants" b = ~/t cg/~ and d = ~t Ng/~g. The general solution of
this equation is:

~(X--x)
G. b
H
+(d--bp) ~- + p exp {--fHdS}.fG exp {fHdS}dS
----- ~v [p exp

{--fndS}l

(16)

where ~0 [ ] is an a r b i t r a r y function and d = ~l[a/k.

In order to determine ~0, the b o u n d a r y conditions have to be introduced into the solution. These b o u n d a r y conditions are"
fort=O
:P:Po;
f o r t = O a n d x : / : X : S = I;
fort > Oandx=X:S=O.

s"

-"* X

P
Fig. 5.
The i n t r o d u c t i o n of these b o u n d a r y conditions in (16) is not v e r y
easy. In order to clarify our procedure an S-x-p space has been
represented in fig. 5. T h e s a t u r a t i o n S is a function of x and t, but
instead of t the pressure p is i n t r o d u c e d as an i n d e p e n d e n t variable
b y means of (13). In the figure the b o u n d a r y conditions are drawn as
solid lines t h r o u g h which an integral surface should be laid. It is
i m m e d i a t e l y clear t h a t no simple surface can be laid t h r o u g h the
three solid lines which intersect at angles of 90 degrees.
An integral surface satisfying conditions a and b is easily found.
It is given b y
S

P = Po exp {fgdS}.
1

(17)

C A L C U L A T I O N S ON T H E F L O W O F H E T E R O G E N E O U S

MIXTURES

343

In this solution S is a function of p only. The integral surface,

therefore, consists of straight lines parallel to the x-axis. It is indicated in fig. 5.
The construction of a solution also satisfying condition c was found
to be possible by laying an integral surface through the vertical line
x = X, p = P0 in fig. 5. Substitution of these values of x and p in
the solution (16) leads to a definition of ~v:
G

(d - - bPo) -~ + bPo exp {-- f H d S } . f G exp {f HdS} dS =

= ~v {P0 exp { - - f H d S }

(18)

or introducing:
S

Y = P0 exp {-- f HdS}

(20)

%v(y) = (d - - bPo) ~ + b y f G exp {fHdS} dS..

1

(21)

F o r a n y value of S a value of y and a value of ~p m a y be deduced

from these two equations. Thus ~v is known as a function of y. Substitution of this function in (16) gives a definite relation between S,
p and x.
In principle, therefore, we have now solved our problem. When
making numerical applications, however, it was f o u n d t h a t the solution in this form is not of m u c h use. B o t h ~vand y appeared to attain
enormous values, which cancelled again after substitution in (16).
This made impossible satisfactory numerical evaluation of the relation between S, p and x. It was found, however, t h a t these difficulties could be avoided by introducing a new function instead of ~v.
A function S 1 of p and S was defined b y the following relation:
S

P = P0 exp {f HdS}.

(22)

St

Introduction of (22) into (20) and

(21) leads to

an expression for ~v:

St

~v[p exp { - - f HdS}] -~ %v[po exp { - - f ndS}] ---1

GI

s,

s,

= (d - - bPo) -H-~+ bp~ exp. {--/lidS}, .fG, exp {fHdS} dS

where G t = G(S1) and n t ---- H(St).

344

H. C. B R I N K M A N

The solution is now obtained in the form :

,~(X - - x)

G,

+(d--bp)-~--(d--bpo ) ~ +
S~

+ bPoexp {-- f HdS} .f G e x p {f HdS} dS =

I

Sj

0.

(23)

For numerical applications a value of S and S 1 should be chosen

and p and x be'calculated from (22) and (23).
In this way the integral surface may be constructed. A crosssection through it for a certain value of p (and, therefore, because
of (13), for a certain value of t) is drawn in fig.-6. In this problem
I

,s

1
X

,x

F i g . 6.

S again proves to be a many-valued function of x. Another remarkable property of the curve in fig. 6 is that it cannot be extended
"beyond a certain S-value, at which S1, as defined b y (22), reaches
the value 1. For larger Sl-vahies H is not defined and (23) cannot be
applied numerically. Now for S l = 1 relation (22) reads:
S

P = P0 exp {f HdS}
1

(24)

which is equal to (17).

This means that the solutions (17) and (23) intersect at the endpoint of the S--x curve of fig. 6.
In fig. 7 both solutions have been drawn. It appears that the
solutions (17) and (23) together form a complete solution of our
equations meeting all boundary conditions. The integral surface
consists of the surface already drawn in fig. 5 together with the
surface (23) which meet along a certain curve. Surface (23) contains
a fold. A cross-section through this complicated integral surface for
a certain value of p is given in fig. 7.

CALCULATIONS ON THE FLOW OF HETEROGENEOUS MIXTURES 3 4 5

In fig. 7 S is again a three-valued function of x. This difficulty is

overcome in the same way' as discussed in the preceding paragraph.
A discontinuity of the S - x curves occurs, which is indicated in fig.
7. Its x-value is determined by equating the shaded areas in fig. 7.
11
J

~x

Fig. 7.

The physical meaning of this discontinuity is that in the upper

part of the column a regio.n of low saturation (a gascap) is formed,
separated by a sharp boundary from the rest of the column. In this
lower region the saturation appears to be constant throughout the
column at any time.

w 4. The inevitability o/ three-valued solutions. In this section it

will be shown that the equations of flow (2a) and (2b) will inevitably
lead to three-valued S - x functions.

,.s

(,X
Fig. 8.

In fig. 8 we have represented by a solid curve a situation where S

is a monotonous function of x. Now according to (2a) and (2b) for
S - ~ 1, v~-+0 and for S ~ 0 , v~-+0. This means that for these
S-values one of the phases is immobilized. Accordingly the saturation remains constant.

346

C A L C U L A T I O N S ON T H E F L o w OF H E T E R O G E N E O U S M I X T U R E S

Therefore, when e.g. a liquid is introduced into the porous material at a high x-value, an increase of S will only occur in the middle
region of the S-interval. This is sketched in fig. 8. The area under
the dotted S-curves increases more and more as more liquid is introduced into the column, leading to three-valued S-functions as
indicated in fig. 8.
A discontinuous S - x curve which corresponds to the actual situation in the porous material cannot be obtained as a solution of the
differential equation describing the rate of change of S.
We conclude that three-valued S - x functions are inevitable with
this way of treatment.
Received 16th September, 1948.

REFERENCES
1) M u s k a t, M . , a n d M. W. M e r e s , P h y s i c s 7 (1936) 346.
2) B u c k l e y ,
S . E . , a n d M.C. L e v e r e t
t, Trans. A.I.M.M.E. 146(1942) 107.
3) W y c k o f I ,
R . D . , and H. G. B o t s e t ,
Physics 7 (1936) 325.