You are on page 1of 42

ANNUAL

REVIEWS Further
Quick links to online content
Copyright 1976. All rights reserved

MULTIPHASE FLUID FLOW 8089

THROUGH POROUS MEDIA


Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

R. A. Waoding
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Applied Mathematics Division, Department of Scientific and Industrial Research,


Wellington, New Zealand

H. J. Morel-Seytoux
Department of Civil Engineering, Colorado State University,
Fort Collins, Colorado 80523

INTRODUCTION

As is frequently the case in topics related to fluid mechanics, the field of muItiphase
flow in porous media embraces highly diverse phenomena, ranging from the motion
of immiscible fluids, through interaction with the medium via the exchange of heat
and/or soluble mass and the exchange between phases, to fluid-solid-phase flows
accompanied by colmatage (clogging) and suffusion (leaching). The latter area, which
includes the whole field of filtration (Davies 1973, Polubarinova-Kochina 1969), is a
topic in itself and cannot be considered in the present survey, which is limited to
the discussion of immiscible-fluid phases, usually in two-phase flow. Again, while
heat or solute may be regarded as an additional phase and may significantly affect
mass flow and phase exchange, the transport of these quantities is generally outside
the scope of the present discussion.
Superficially, flow through porous media appears to offer the possibility of two
alternative analytical approaches, based upon either the scale of pore dimensions or
the scale of the macroscopic system. The former approach would be expected to lead
to detailed boundary-value problems, would incorporate physicochemical properties,
and should, in principle, lead to exact solutions from which numerical coefficients
can be evaluated. Unfortunately, such a procedure is not fully practicable owing
to the geometric complexity of porous media, although valuable qualitative models
have been obtained. Usually, authors abandon the capillary scale and take Darcy's
law, involving an undetermined constant-the permeability-as the basis of an
axiomatic, macroscale approach (Philip 1970, 1973a, 1974).
However, in the field of mUltiphase flow many important questions remain
unanswered. For example, what factors affect the flow regime and hence the pore
distributions of immiscible fluids, with effects of pressure gradient, etc? In
immiscible displacement, what proportion is recovered of a fluid initially in the

233
234 WOODING & MOREL-SEYTOUX

medium? How is the process of wetting of a porous material affected by the


surface constitution of the solid, the composition of the liquid, and the rate of
uptake? Studies on surface-interface reactions should provide valuable insights into
such mechanisms and may be helpful in the formulation of macroscopic relation
ships. The process of wetting, for example, involves scales so small that details of
pore geometry may (temporarily) be neglected. Part of this review, therefore,
discusses the role of interactions within capillaries.
Unless it is otherwise stated, fluids are assumed to be incompressible and
Newtonian, while the porous media are taken to be incompressible and un
deformable.
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

MECHANICS OF IMMISCIBLE DISPLACEMENT IN PORES


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Immiscible displacement of one fluid phase by another in a porous medium implies


movement of a liquid-fluid interface through the pore spaces of the solid. Whether
the interface meets the pore wall, forming a moving three-phase contact line so that
the displacing fluid wets the solid, or whether the interface remains within the
fluid, leaving a film of the displaced fluid on the walls, in a draining process,
depends particularly upon relationships between interfacial tension and surface
tensions and the hydrodynamic forces generated by the bulk flow.
Suppose that an interface between two immiscible fluids, with interfacial tension
1', advances with speed U along a capillary of equivalent diameter d. It is assumed
that d greatly exceeds the molecular mean free path. The magnitudes of viscous,
inertial, and gravitational forces, relative to 1'. depend upon the dimensionless
groups

Jl.U/Y. (1)
(a)

where JI. is the viscosity of the appropriate fluid phase, p is the corresponding
density, and !J.p is the magnitude of the density difference between the two fluids.
These combinations have been discussed by Bretherton (1961) for the liquid-gas case.
Wallis ( 1969, p. 285) discusses similar groups, taking buoyancy as the reference.
The ratio of (lb) to (la) is the Reynolds number Re = pUd/JI.; a Reynolds number
can be defined for each fluid component. If Re is small, as in Darcy (Stokes) flow,
the equations of motion are linear. The combination (lb) is the Weber number,
whereas the ratio of (lb) to (lc) gives the Froude number. Because d is usually
very small in applications to porous media, groups involving d are often not
important. This may not be the case, however, in some engineering applications
such as packed towers and trickle filters.

Displacement Flow in Capillary Tubes


Particular significance can be attached to the combination (la). Fairbrother &
Stubbs (1935) observe that the rate of movement of a gas bubble in a horizontal
capillary tube, relative to the mean flow velocity, is a one-valued function of
/lU/y. taking /l as the liquid viscosity. Taylor (1961) describes experiments in which
MULTIPHASE FLUID FLOW mROUGH POROUS MEDIA 235

a viscous liquid is expelled by gas from a circular capillary tube, and observes that
the proportion of liquid retained as a film on the tube walls increases towards an
asymptotic limit-about O.56-as jlU/"1 becomes "large" (2, or greater). Here Re
is kept small. This work has been applied by Saffman (1963) to immiscible-fluid
displacement in a porous-medium model consisting of randomly oriented capillary
tubes, in particular to take into account the fact that a proportion of the original
fluid is left behind.
In many situations involving immiscible fluids in porous media, we have
jlU/y 1, implying that, although the capillary flow is viscous dominated (Re 1),
the dynamics of the interface are affected by viscosity only in the neighborhood of
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

the solid walls, where a wetting fluid may be present as a thin film. Within that
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

region the lubrication hypothesis appears to be valid (Bretherton 1961), whereas


outside the wall region, provided that inertia and gravity as measured by (lb,c) are
negligible, the profile of the interface is determined by an equilibrium between "I and
the pressure difference across the interface (cf Orr, Scriven & Rivas 1975). For
decreasing values of the velocity U, the thickness of the viscous wall region appears
at first to decrease as a fractional power of J1.U/"1' However, there is a discrepancy
between Bretherton's theoretical value of j for the power-law exponent and the
empirical value of t obtained by Fairbrother and Stubbs and extended in range
by Taylor. Further careful experiments (Bretherton 1961) show marked departures
from the power law as jlU/y is reduced to very small values 10- S), with some
indication that the thickness of the retained film becomes constant.
As the dimensionless group ( lc) does not involve U, this group is likely to be
important in static as well as dynamic situations where the shape of the interface
is affected by gravity as well as by interfacial tension. Bretherton (1961) demon
strates that a long air bubble in a vertical capillary tube sealed at one end cannot
rise steadily under gravity (under conditions in which the lubrication theory is valid)
unless 6.pgd2/y exceeds a critical value-about 3.37 for a circular tube. Only when
this criterion is satisfied can the profile of the bubble cap be matched to the profile
of the thin-film wall region. Since the criterion is not usually satisfied in a porous
medium, trapped bubbles are a likely occurrence.

MULTIPHASE FLOW REGIMES IN CAPILLARY MEDIA

Flow at Low Reynolds Number


If flow boundary conditions are held constant and sufficient time is allowed to
elapse, a "steady" muItiphase flow regime may be established in a porous medium.
For a given fluid phase, the boundaries will consist of solid pore walls broken by
"islands" where the fluid forms interfaces with other fluids. Such interfaces are
bounded by solid-liquid-fluid contact lines and form appropriate contact angles
[see, for example, Orr, Scriven & Rivas (1975)J.
Interface properties are likely to exert a dominant effect upon the flow patterns
that occur. If there is significant contact-angle hysteresis (discussed in greater
detail in the section on wetting), zones of stagnant liquid may occur, interspersed
with bubbles of other phases [the "Jamin effect" (Scheidegger 1960, Schwartz,
236 WOODING & MOREL-SEYTOUX

Rader & Huey 1964)]. Phase entrapment may also be induced by medium
geometry (Morel-Seytoux 1969), and pressure history may be important.
The interface shape is expected to be independent of velocity, provided that
jlUfy is not of order 1. With increasing jlU/y, the "islands" formed by fluid-fluid
interfaces could become elongated and extend through the pores as one fluid forms
a core within a film of the other, as suggested by the results of Taylor, Bretherton,
and others as already discussed.
While the nonslip boundary condition may be applied at solid-fluid interfaces, the
condition at fluid-fluid interfaces will depend upon physical factors that affect
interface rigidity; a "clean" surface is nonrigid and is capable of transferring
momentum between phases, whereas a nonslip boundary condition may be more
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

appropriate at a "dirty" interface, which has high rigidity. The possible influence of
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

different surface conditions has been pointed out by Yuster (1951). As an extreme
example, in the displacement of water by viscous oils in certain hydrophilic media,
a film of water appears to be retained on the solid. This may act as a lubricant
that greatly enhances the conductivity to oil (Evgen'ev 1965).
On the other hand, Philip (1972a) argues that the importance of the fluid-fluid
interface condition may be relatively minor, since the effect of the nonslip condition
that holds on the surrounding solid-fluid boundaries is likely to be dominant.
Philip solves several cases involving idealized boundary conditions leading to uni
directional Stokes flows-e.g. creeping flows between parallel plates and in tubes,
with longitudinal no-shear slots. He also gives integral properties of such flows
(Philip 1972b), which could assist in estimating the effect of the gas-liquid boundary
conditions upon momentum transfer between phases.

Flow at High Reynolds Number


For single-phase flow in a porous medium, four different regimes have been
distinguished experimentally as the Reynolds number is increased (Wright 1968),
extending from the laminar (Darcy) range, though laminar-inertial (steady nonlinear)
flow, to transitional (unsteady) flow, and finally to fully developed turbulent flow.
This is an example of the sequence typical of flow of viscous fluid past obstacles.
A similar sequence would be expected to apply for each phase when the flow is
multiphase, but is likely to be obscured by interactions with other phases. Such
interactions would be functionally dependent upon dimensionless groups (la,b,c)
evaluated for each phase, or ratios of these. No systematic study involving all
parameters, over a full range of Reynolds numbers, is known to the authors.
In partly turbulent, co-current gas-liquid flow through a porous bed composed
of I-mm-diameter spheres, Eisenklam & Ford (1962) have found no dependence
of the flow parameters upon either the Weber number or the Froude number,
indicating that surface tension and gravity may be neglected in comparison with
inertia effects under those experimental conditions.
However, these studies have demonstrated a transition where the flow regime
apparently changes in a fundamental way. Typically, this is jnduced when the gas
Reynolds number Ref is steadily increased through the experimental range while
the liquid Reynolds number Re, is held constant. Prior to transition, the flowing
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 237

gas and liquid phases largely occupy separate pores. On the other hand, the regime
above transition is apparently annular flow, where each pore contains a gas core
with a mobile liquid film attached to the walls. The latter regime could be related
to the "long-bubble" flows of Taylor and Bretherton. It involves lower flow rates,
for a given pressure gradient, than the former regime, possibly because of greaer
tortuosity and greater resistance in the smaller flow cross sections.
For the experimental ranges covered-5.42 < ReI < 140 and 20 < Ref < 914-
correlations of the data indicate that the liquid-flow regime is almost fully turbulent,
whereas in the gas phase the flow regime is at least partially laminar.
Packed beds and towers used in chemical-engineering applications generally
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

employ very coarse media, and the Reynolds numbers are large (Scheidegger 1960).
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Gas-liquid co-current downflows through three different beds of spheres are


described by Weekman & Myers ( 1964), whose semiquantitative diagram of flow
regimes is given in Figure 1a. At low liquid-flow rates, the liquid trickles under
gravity as a laminar flow over the packing, and continuous-phase gas flow is
possible up to high rates. At higher liquid-flow rates, but low gas flows, there is a
transition region in which the liquid flow appears to be turbulent or rippling. With
higher gas-flow rates, dense liquid-dominated pulses or slugs appear, having sharp
leading edges and long trailing wakes. The Reynolds number appears to have little
effect on the location of regime boundaries. Weekman and Myers base their
correlations of data upon the parameterization of Lockhart & Martinelli (1949),
which was devised for two-phase flow in ducts or pipes (Wallis 1969).
Using packed beds of somewhat larger particles, Turpin & Huntingdon ( 1967)
also observe three regimes, which [following convention (Wallis 1969)] they desig
nate as bubble flow, slug flow, and spray flow (Figure 1b). At very low gas-flow rates,
bubbles of gas are transported in liquid. The slug-flow regime is actually a
pulsating flow. When the flow is predominantly gaseous, spray flow develops with
the liquid transported as a dense mist.
Elementary theories applicable to the results of Turpin and Huntingdon have
been proposed by Hutton & Leung (1974), based upon experience gained with
countercurrent-flow studies (Buchanan 1967, Hutton, Leung, Brooks & Nicklin
1974). It seems reasonable to assume that the degree of liquid saturation or
"holdup" h (Scheidegger 1960) is a function of liquid volume-flow rate QI and
longitudinal pressure gradient dp/dz, while the latter, in tum, is directly dependent
upon h and the gas flow rate Qf. Explicit expressions in these forms have been
obtained for a linear annular-flow model at low-to-medium liquid Reynolds
number.
At very high liquid Reynolds numbers, however, which are generally typical of
co-current operating conditions in packed towers, the main resistance to flow arises
from form drag (Schlichting 1960) due to the coarse medium. It follows by
dimensional analysis that
(dp/dz + gpr)d ex: !PI(Qr/W, (2)
which is valid as ReI ..... 00, where d is the pore scale of the bed, g is gravity, and
the constant of proportionality will depend upon the geometry of the porous bed.
t-)
""
00
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

I
9!'

jo

hlfIl 8 SPRAY SLUG flOW

i
FlOW

,
,

,t\

.
Lt 0'5 .
t
I

PUlSG flOW trl


O

.....
"",-'----

,..l;---
" "
,,'

o
BUBBLE FLOW
GAS ,

ICONTINUOUS --
TRANSITION'---
__

FLOW
I '\ "

o 10 20 10 102 103

LIQUID FLOW (KG1M:SEC.) MASS FLOW RATIO (UQUID/GAS)

(a) (b)
Figure 1 Regimes of co-current, high-Reynolds-number, water-air !lows in packed beds; (a) O-0.65-cm-diameter alumina spheres, .-
0.475-cm glass spheres, o-0.378-cm TCC beads (after Weekman & Myers 1964); (b) beds of tabular alumina particles of diameters
0.76 em ahd 0.82 em (in different experiments) (redrawn from Turpin & Huntingdon 1967).
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 239

Empirical formulas relating dp/fiz and Qr are discussed by Hutton and co-workers.
Hutton & Leung (1974) find that the theoretical results from their annular-flow
model are in reasonably good agreement with data of Turpin and Huntingdon for
the experimental ranges 80 < Re, < 1700, 200 < Rer < 700, approximately.
Historically, the use by chemical engineers of countercurrent flow in packed beds
predates the use of co-current flow, and the literature on the former case is
correspondingly more extensive. Early workers on countercurrent flows such as
Sherwood (Scheidegger 1960) have identified two breaks on the logarithmic plot of
dp/dz versus Q, as, first, the load point where the slope of the curve first deviates
significantly from a value of about 2, and second, the jiood point wht"re the slope
becomes very large. (Note that flooding is not observed with co-current flows.)
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Lerner & Grove (1951) have suggested that both loading and flooding arise from
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

the same mechanism-interfacial wave-type instability that appears at a critical gas


velocity, the critical point for interstices of minimum cross-sectional area occurring

10-1
N

0
-

::t.

c:tlci:
(/)1'"
"'_ c
:J 0>

10-2

10

Figure 2 Comparison of theoretical criteria for the onset of flooding; - - - - "gravity


inertia" regime, - - - - - "gravity-viscosity" regime, with observations of flooding onset: 1.
Morton, King & Atkinson (1964), 2. grid packing, 3. dumped packing, 4. stacked rings,
plotted using the correlation of Lobo, Friend, Hashmall & Zenz (1945) (after Hutton,
Leung, Brooks & Nicklin 1974).
240 WOODING & MOREL-SEYfOUX

at a lower mean velocity (i.e. the load point) than that which applies for the porous
medium as a whole (i.e. the flood point).
A seemingly analogous flooding phenomenon occurs for annular countercurrent
flow in a straight tube where, for gas velocities higher than a critical value
calculable from small-perturbation theory, interfacial waves form and may grow
large enough to bridge the tube. For horizontal and vertical tubes, respectively,
Thorpe (1969) and Cetinbudaklar & Jamieson (1969) have calculated critical
conditions that are in good agreement with observation. Experimentally, large
amplitude almost-stationary waves develop.
In contrast with the foregoing, the criterion used by Wallis (1969) is based upon
a kinematic-wave model of the flow and states that flooding occurs when the
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

derivative oQ//oh vanishes. An illuminating analogy is afforded by a kinematic-wave


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

model of road traffic (Lighthill & Whitham 1955); the rate of traffic flow ceases to
rise at a critical density and a bottleneck results. Hutton, Leung, Brooks &
Nicklin (1974) adopt the Wallis criterion and calculate critical flow rates, based
upon empirical formulas, for the case of fairly low R, (their "gravity-viscosity
regime") and for very large R/ ("gravity-inertia regime"), which obeys a relation of
the type (2) with the sign of dp/dz reversed. Note that the sign change is all-important:
if (2) holds, which is the case for co-current flow, kinematic shocks do not occur.
The comparison by Hutton and co-workers of their results with experimental data
is reproduced in Figure 2.
Two quite different mechanisms of the flooding phenomenon have now been
considered-(a) the wave-instability type, which is not specifically dependent upon
pressure gradient, and (b) the kinematic-wave "bottleneck," based upon the
relationship between pressure gradient and saturation. While mechanism a is typical
of wetted-wall columns for which the effective voidage (proportion of gas space) is
high, it is likely thai mechanism b is typical of packed columns where the voidage
is low and R, is correspondingly low.

MECHANISMS OF THE WETTING OF SOLIDS

Undoubtedly the wetting properties of the solid surfaces within a porous medium
are of major importance to phenomena of capillary uptake. However, while the
physical chemistry of interfacial phenomena has been the subject of many detailed
studies, these have been concerned mainly with static conditions; the hydro
dynamics of motion near a moving interface are not well understood. For a "wetting"
fluid, and mainly for low Reynolds numbers, Huh & Scriven (1971) have given an
historical review of qualitative ideas and observations, commencing with the rolling
motion or "fountain effect" described by West (1911) and Rose (1961). This effect
has been made visible experimentally by Schwartz, Rader & Huey (1964) by
injecting a little dye into a short length of transparent liquid in a capillary tube;
this shows also that the nonslip condition is apparently obeyed at the solid-fluid
boundary.
Generally, the simple rolling flow appears on just one side or the other of a
fluid-fluid interface, and depends upon the direction of flow at the interface itself.
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 241

A careful study of the kinematics of the flow near the three-phase contact line,
formed when the fluid-fluid interface meets the solid boundary at a finite angle, has
been made by Dussan V. & Davis (1974). Their qualitative experiments demonstrate
the simple roll and also a "double roll" in the second fluid (Figure 3a-d); relative
to the contact line, fluid particles approach along the fluid-fluid interface (if) and
along the solid-fluid interface (sf), and leave along a ray that divides the angle
between the interfaces. The reversibility of the flow has been demonstrated. It is
stressed that the flow near the contact line does not resemble stagnation-point
flow. Dussan V. and Davis show that, while the moving contact line and the no-slip
boundary condition are kinematically compatible, for a contact line on a rigid
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

boundary the velocity field is multivalued and forces become unbounded as the
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

contact line is approached. This is a general result, independent of constitutive


assumptions.
This singUlarity has been encountered in the related problem of plane viscous
flow at a wedge-shaped corner under inhomogeneous boundary conditions, which
has been treated by Taylor (1960) and by Moffatt (1964) using the fact that the
Stokes approximation is valid in a neighborhood of the corner. A similar approach
for the three-phase contact line has been attempted by Huh & Scriven (1971), on the
assumption that the important length scales are so small that curvature of the
interfaces may be neglected. Because of this planar assumption, continuity of
normal stress across the fluid-fluid interface cannot be satisfied. However, their
results are in good qualitative agreement with experiments (Figure 3e) and provide
a functional relationship between the contact angle, the viscosity ratio of the two
fluids, and the consequent motion along the interface. Huh and Scriven also
discuss the singularity at the contact line and suggest that physicochemical effects,
such as long-range intermolecular forces, may dominate there.
A critical review of hydrodynamic theories of viscous fluid-fluid displacement in
the wetting problem has been given by Dussan V. and Davis.

(a) (b)
f

4- :'ei
(:jJ !y4
I
-
s s

(c) (d)

s


s-- =
s

Figure 3 Qualitative descriptions of flow ahead of a moving contact line; (a-d) positions of
dye spots at successive times [redrawn as a single figure from Figures 14 and 15 of Dussan
V. & Davis (1974)]. A deformable solid is shown. (e) Streamlines calculated for c/J 300 and
=

viscosity ratio IlJIIlI 0.01 (after Huh & Scriven 1971).


=
242 WOODING & MOREL-SEYTOUX

Contact Angle
Many studies of interactions at the three-phase contact line have been concerned
'
with the thermodynamics of the equilibrium relationship represented by the
Young-Dupre equation (Zisman 1964, McLaughlin & de Bruyn 1969, Morrow
1970a, etc)
'YI! cos . = 'Ys!- 'Ysi. (3)
where . is the equilibrium contact angle, I' is an interfacial tension, and the
subscripts s. I. and f refer to solid, fluid (liquid), and fluid, respectively (Figure 3).
Here 'YI! corresponds to 'Y in the notation of (1).
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

The contact angle is a useful measure of the "wettability," i.e. the affinity 01 a
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

given liquid for a given solid as measured under suitably determined conditions.
Fox & Zisman (1952) and Shafrin & Zisman (1960) consider a critical surface
tension 'Yc corresponding to the value of 'YI! for which cos . == 1 in equation (3);
liquids for which 'YI! > 'Yc make finite contact angles with the solid surface, but
liquids for which 'YI! < 'Yc will spread indefinitely. Whether or not a finite contact
angle exists, the wettability can be defined precisely as the loss of Gibbs free energy
per unit surface area during wetting (Briant & Cuiec 1971). These authors also give
a valuable review of methods of measuring wettability.
Whether equation (3) should include a further term representing the disjoining
pressure (Deryaguin 1940) and possibly other effects is an open question. The issue
is complex since numerous physical phenomena may be involved (padday 1970).
Equation (3) has been tested experimentally for several solid-liquid-vapor systems by
Bailey, Price & Kay (1970), who have measured y,! for saturated vapor and Ysl
from the work done during slow cleavage of a mica sheet,in the appropriate
environment-an elegant technique. Here the solid surfaces involved are covered
witli adsorbed vapor films. Values of 'Ysl are found to be rather higher than expected;
this is attributed to hydrophobic effects of the surfaces used. Otherwise, (3) appears
to conform quite well to the data provided that the latter are obtained under
constant (nonchanging) conditions.
These authors observe considerable hysteresis in the contact angle for water but
not for nonpolar liquids. Hysteresis is attributed to a reorientation of polar molecules
in the surface layers so that interfacial tensions may adjust to maintain equilibrium
as the contact angle Changes; when the limit of available "reorientable" molecules
is reached, the contact line must begin to move (see also Schwartz, Rader & Huey
1964).
Relatively few studies have been reported of the dynamic contact angle . i.e. for
the moving contact line. In a review of these, Huh & Scriven (1971) note the series
of different regimes of velocity dependence that occur as the rate of wetting (or
draining) is increased. For very slow motion 0.1 cm min -1), is velocity
independent and remains at the recently advanced or retarded static values
(Elliott & Riddiford 1967), which differ from . as a consequence of contact-angle
hysteresis. At higher speeds, is velocity dependent up to about 1 cm min -1 (for
air-water-siliconed-glass), after which another velocity-independent regime occurs.
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 243

A further transition occurs at very high speeds of wetting; a film of the displaced
fluid remains on the solid surface and the contact line disappears (Deryaguin &
Levi 1964). Since the displacing fluid no longer wets the solid boundary, the flow
is then more typical of draining.

Liquid-Gas Displacement
If the fluid f is a gas, how does displacement by a liquid at the solid surface take
place, and what is the physical interpretation of 'Is! in (3)? The original suggestion
(Hardy & Doubleday 1922) that a thin film of adsorbed liquid molecules separates
the solid surface from the gas is well established (Frenkel 1946, Bascom, Cottington
& Singleterry 1964, Padday 1970,.etc). In Hardy'S mechanism of wetting, adsorption
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

of liquid molecules from the vapor phase, or surface diffusion ahead of the contact
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

line, builds up a film over which the bulk liquid then spreads. Equations attempting
to describe such a mechanism have been formulated by Ruckenstein (1969). While
the existence of a spreading film is well established experimentally, Bascom,
Cottington & Singleterry (1964) indicate that surface-tension gradients induced by
volatile contaminants may exert a dominant effect. In some cases a secondary film
may be formed, leading to enhanced spreading rates, while in other cases spreading
may be inhibited or the liquid may recede.
Churayev, Sobolev & Zorin (1970) have observed that in very fine capillary tubes
with dry walls the contact angle is nonzero and exhibits velocity dependence at
very low rates of flow. It is demonstrated fairly convincingly that the latter effect
is due primarily to the finite rate of formation of the adsorbed film.
Miller & Ruckenstein (1974) d iscuss a theoretical model for the long-range
intermolecular forces close to the contact line. On this scale the curvatures of the
solid surface and of the fluid-fluid interface are assumed to be negligible; also, for
simplicity, the authors neglect gas interactions. For a liquid molecule at point Q in
the interface (Figure 4), F is the resultant of the force F. due to molecules of the
solid and of the force F, due to molecules of the liquid. The component F T of F
tangential to the interface will cause flow toward or away from the contact line at

GAS
LIQUID

o
SOLID
Figure 4 Forces at a point Q near the contact line 0 due to long-range molecular
interactions (after Miller & Ruckenstein 1974).
244 WOODING & MOREL-SEYTOUX

0, depending upon its sign, and will vanish when 4> = 4> For different liquids on
.

a single solid the predicted variation of 4>. is consistent with experimental data of
Fox & Zisman (1952), and the predicted dynamic behavior is qualitatively correct.
At this stage the model is suggestive of a promising development-the incorporation
of long-range (in particular London-Van Der Waals) intermolecular forces in the
hydrodynamic equations-appropriate near the contact line.
Considerably uncertainty surrounds the influence of contamination and surface
roughness upon the wetting process. Bascom, Cottington and Singleterry have
observed that surface microscratches of the order of O.l-Jilll width will attract fluid
and enhance spreading. On the other hand, experiments by Schwartz & Tejada
(1972) using roughened solid surfaces indicate that the velocity-con tact-angle
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

relationships are unaltered provided that the physicochemical properties have not
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

been altered by roughening. In practice, the moving contact line may exhibit
"stick-and-slip" irregularities, small lengths of line appearing to break loose
suddenly and move rapidly over short distances (Huh & Scriven 1971). Contact
angle hysteresis is a likely contributory cause. The nearby flow is clearly three-
dimensional and unsteady. Many observations of wetting phenomena may in fact
represent averages of such motions.

Liquid-Liquid Displacement
Surprisingly few studies are available on the solid-liquid-liquid contact line. El-Shimi
& Goddard (1974) list six references describing measurements of the equilibrium

-02

-04
"'&- Q
CJ)
0-06
0

-08

10-4 10-2 10-1


V CM SEC-1
Figure 5 Comparison of theory, , with experiments on benzene-water contact angle
4> versus displacement velocity V in two cylindrical capillary tubes; Q benzene -+ water,
-

.-water benzene (tube A), O-benzene -+ water (tube B) (after Blake & Haynes 1969).
-+
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 245

contact angle between low-energy solid surfaces and water-to-nonpolar-liquid inter


faces. Some results may be deduced from results obtained with liquid-gas interfaces,
since the equilibrium contact angle follows from these by repeated application of
the Young-Dupre equation (3).
A review of several theoretical and experimental studies on dynamic contact
angles (Schwartz & Tejada 1972) mentions one specifically concerned with liquid
liquid displacement. Blake & Haynes (1969) consider the influence of the adsorptive
properties of the solid surface on the movement of the contact line. Using Eyring's
theory of absolute reaction rates (Glasstone, Laidler & Eyring 1941), they obtain
expressions relating velocity to contact angle in terms of the rate constant of the
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

molecular-displacement process at the solid surface. It is suggested that the contact


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

line will be extremely thin-perhaps spanning a thickness of only a few adsorption


sites-but that it will continually fluctuate about a mean position as molecules of
one fluid interchange with molecules of the other. The approach is considered to
be complementary to hydrodynamic studies of flow near the contact line, and in
fact the no-slip condition at the fluid-solid interface is preserved. Comparison of
the theory with experiments using benzene and water on siliconed glass show good
qualitative agreement over a considerable velocity r;mge (Figure 5). A good fit is also
obtained with experimental results on gas-liquid interfaces (Schwartz & Tejada
1972) but the constants obtained empirically tend to differ from theory, and different
values apply in different ranges.

CAPILLARY UPTAKE-INFILTRATION

Early workers (Green & Ampt 1911, Washburn 1921) have considered the analogy
between uptake of water into a straight capillary tube and the wetting of a porous
solid. In spite of the apparent crudity of such a model, which can be equivalent
only to the advance of a liquid profile in the form of a step function, there is
surprisingly good agreement with observed rates of infiltration into laboratory
media and natural-soil profiles (Swartzendruber & Huberty 1958, Childs 1969, p.
277).
For a capillary tube of radius r, inclined at angle IX below the horizontal, the
well-known differential equation and its solution for the distance x of liquid uptake
as a function of time t can be written as
d(X2)/dT= aX, T= X-log(1 X), (4)
(a) (b)

where X = pgx i sin lXi/P, T = (pgr)2t/(4pAP), p and Jl are the density and viscosity
of the liquid, g is gravity, and P is the driving force at the liquid-gas interface.
Typically, we have P = (2/r)YII cos </>, where </> is the dynamic contact angle; the
gas pressure is assumed to be constant. In (4) the upper signs are taken for
IX > O.
Expansions of (4b) are of particular interest to hydrologists and soil physicists.
Capillarity is dominant when X is small, and X = Tl/2tT+. In terms of
dimensioned variables the first term involves capillarity only and the second term
24 6 WOODING & MOREL-SEYTOUX

involves a gravity effect only. This has the form of the Philip infiltration equation
[reviewed by Childs (1969)]. For large X. possible only for (X > 0, we have
X - t T + O(log T) as T -> 00, i.e. the rate of infiltration is asymptotically uniform
under gravity. If (X < 0, we have X- I as T -> 00, corresponding to the equilibrium
level.
The applicability of (4) with the simple form for IlP assumed above is probably
dependent upon the nature of the adsorbed film ahead of the liquid-gas interface.
In considering the anomalously high rate of uptake of certain liquids by certain
porous bodies, Good (1973) concludes that the simple expression for IlP is more
likely to be applicable provided that the solid surface ahead of the interface is
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

already in equilibrium with the saturated vapor. If the solid surface is initially dry,
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

the reduction in free energy of the surface on becoming covered by the liquid film
may provide a further driving force, suggesting that the rate of uptake may be
increased. It is difficult, however, to interpret the experimental result already noted
of Churayev, Sobolev & Zorin (1970) in terms of this theory; possibly the surface
energy of the medium and the nature of the liquid are decisive factors.
At very short times the simple theory breaks down because the Hagen
Poiseuille equation is not valid near the inlet of the capillary tube and also does
not hold at the initiation of flow. Szekely, Neumann & Chuang (1971) include
inertial effects and formulate the energy equation for the starting flow. The most
important energy dissipation is assumed to be associated with the formation of a
vena contracta at the entrance to the capillary. From a comparison of numerical
results, their rigorously derived solution differs appreciably from (4) for times of
the order of 1000r 2 sec, where r is the capillary-tube radius in centimeters-e.g.
0.1 sec for r = 0.01 cm. Since experimental measurements usually begin at much
longer times after flow has been initiated, the asymptotic solution (4) is generally
valid over the range of interest. However, it is possible that inertial effects are present
in the uptake of ink in crtain printing operations.

PHYSICAL EFFECTS ARISING FROM PORE GEOMETRY

In the foregoing, the general properties of solid-fluid-fluid interactions have been


examined with only minor reference to details of the pore structure of the solid
medium. This approach has been a matter of convenience in discussing a complex
system; for the aspects considered in this section, the irregular geometry of the
medium is of major importance.

Capillary Hysteresis
A discussion of the principles of equilibrium of a wetting liquid in contact with a
gas in a porous medium is given by Childs (1969, Chapter 8). The equilibrium
pressure drop flP across a "clean" liquid-fluid interface is given by the well-known
Laplace relation
(5)

where r1 and r2 are the principal radii of curvature of the interface. Since r1 and
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 247

'2 are strongly constrained by local pore dimensions, by the inclinations of the
solid boundaries (of the pore), and by the contact angIe (which may be influenced
by contact-angle hysteresis), it follows that M is a rapidly varying, in fact quasi
oscillatory, function of interface position.
Conversely, suppose that I!lP can be fixed externally, as in a pressure-plate
apparatus (Childs 1969, p. 99). For a given I!lP that establishes partial saturation
of the medium, an indefinitely Iiuge class of different interface configurations is
po$sible; the configuration actually realized will depend upon the previous history
of wetting and draining. The degree of saturation or, as is usually appropriate, water
content depends upon interface configuration as well as upon M, so that the
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

system will exhibit rate-independent or "permanent" hysteresis (Morrow 1970a).


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Each inter'ace configuration is stable in the Gibbs sense, as discussed by Haynes


(1930). A slow change of I!lP forces part of the configuration into a position of
unstable equilibrium, from which it "flips over" into a new stable configuration in
a Haynes jump (Miller & Miller 1956, Philip 1970). On the basis of this physical
picture, the system of capillaries might be subdivided into "pores," each pore
consisting of a cavity connected to other pores through relatively narrow necks. The
essence of the independent-domain theory (Preisach 1935, Neel 1942), as applied to
this problem by Everett and co-workers and by Enderby (Childs 1969, pp. 125-34),
is that each pore has a unique filling pressure I1Pwand a unique draining pressure
I1Pd The difference I1Pw-I1Pd is the main contribution to permanent hysteresis. On
this fundamental hypothesis, it is appropriate to define a distribution function
f(I!lP I!lPd) that is proportional to the total volume of the aggregate of pores that
W'

wet and drain at the pressures I1Pwand I!lPd, respectively. (If the domains are not
independent, f will be a function of other parameters as well.)
Experimentally, the amount of hysteresis observed in the relationship between
I1P and the degree of saturation is considerable and provides a sensitive test of the
independent-domain model. Many such experiments have now been performed.
Usually, a series of rewetting and redrying cycles is carried out between various
upper and lower saturation limits. One series, e.g. the rewetting curves, can be
used to "calibrate" the model, from which a numerical prediction of the other series
can be made. Two variants of the experimental method have been used by different
workers-(a) "steady-flow" conditions, where the flow rate is changed in finite small
steps, and (b) the "unsteady-flow" situation, in which the flow rate is changed slowly
and continuously.
The experimental results so far obtained appear to depend partly upon the
method of measurement (Davidson, Nielsen & Biggar 1966, Topp, Klute & Peters
1967, Topp 1971a). Experiments based on a indicate that the independent-domain
model provides a moderately good fit to data for glass beads and sands
(Poulovassilis 1962, 1970), although Talsma (1970) has found discrepancies at high
saturaion levels. Poorest agreement is found in comparison with experiments using
b for glass-bead media (Topp & Miller 1966, Bomba 1968) and for various loam
soils (Topp 1969, 1971a).
It should be noted that these discrepancies are not considered in relation to the
first experimental cycle, which, whether wetting or draining, differs from later cycles
248 WOODING & MOREL-SEYTOUX

and is normally disregarded. The difference arises from phase-entrapment effects,


already discussed, which also account for the "irreducible" liquid content of - a
drained porous solid [considered by Morrow (1970b) for water-air interactions and
by Slattery (1974) for oil-water interactions] and for the presence of gas bubbles in
the medium in the rewetted state after IlP has been returned to zero.
Discrepancies of the magnitudes observed in the above experiments have led to
very careful studies of wetting and draining at different rates in both horizontal and
vertical columns (Smiles, Vachaud & Vauclin 1971, Vachaud, Vauclin & Wakil
1972), using vented systems to avoid pressure buildup in the gas phase. Surprisingly,
in both configurations it is found that liquid uptake is uniquely related to !J.P,
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

irrespective of rate, but a multiplicity of relationships is obtained for different rates


of draining. Since there exists a clear asymmetry between the hydrodynamics of
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

I
I
I
I
-2 ft -2
1\
II
1\
0 1\ q
: \ I

\
Cl. \ Cl.
"l "l
\
\
\

-1 -1

b 1 I I o ----------
0-2 0-3 0-4 02 0-3 04
n8 n8
(a) (b)

Figure 6 The main hysteresis loop obtained by Topp (1969) for Rubicon sandy loam,
with predicted scanning curves denoted by - - --- for Neel-Everett model and for
Philip-Mualem model, and experimental results (points) for (a) redrying and (b) rewetting ;
capillary pressure!:.P vs saturation referred to total volume nO (after Mualem 1974).
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 249

liquid displacing gas. (wetting) and gas displacing liquid (draining), it may be
suspected that a dependence upon the parameter pUly as discussed earlier is
particularly important during draining, that is, the amount of liquid left behind
does depend upon the draining rate.

Extensions of Independent-Domain Theory


A somewhat artificial but potentially useful variant of independent-domain analysis
has been initiated by Philip (1964) and further developed by Mualem (1973, 1974).
If the distribution function f(IlPw.llPd) is assumed to be separable in the form
G'(IlP ) H'(IlPd). where the primes signify differentiation with respect to llP. then,
w '
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

after a series of rewettings and redryings, the saturation can be expressed


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

analytically as a function of G. H. and boundary values of the saturation


evaluated on the main hysteresis loop. The problem is simplified to a series of
numerical quadratures. While drawbacks of the independent-domain model in
relation to "unsteady" experiments are still evident (Figure 6), the advantages of
analytical simplicity are considerable.
In a development of earlier work by Everett (1967), Topp (1971b) discusses
pore-interaction effects, i.e. nonindependent domains. This nonindependence is
particularly significant near saturation, where pore "blockage" against air entry (i.e.
loss of gas-phase continuity) impedes initial redrying, and in a relatively dry
medium, where liquid reentry is obstructed during the early stages of reweUing.
While such elaborations of the theory lead to better agreement with observation,
the labor involved in the analysis is considerable.

Hysteresis Effects in Hydraulic Conductivity or Relative Permeability


For multiphase flow, the (hydraulic) conductivity to a given fluid phase can be
defined macroscopically as the volume flux of that phase per unit total potential
(arising from capillary and gravity effects, for example). It is related linearly to
relative permeability (Morel-Seytoux 1969).
In contrast to the significant hysteresis effects observed in the relations involving
tlP already discussed, the saturation-conductivity relationship shows very little
hysteresis. This important property has been known for ma.
experiments of Leverett (1939)]. It has been further demonstrated in the recent
studies of hysteresis detailed above, in particular for glass beads by Topp & Miller
(1966), for sands by Talsma (1970) and Vachaud & Thony (1971), and, in spite of
greater experimental scatter, for loam soils by Nielsen & Biggar (1961) and Topp
(1969, 1971a).
Poulovassilis (1969, 1970) observes a very narrow hysteresis loop for sands, in
which the hydraulic conductivity in the drying state is higher than that in the
wetting state. Youngs (1964), using a medium composed of slate dust, obtains the
opposite result. Since the large pores should tend to be the first to fill on wetting
and the first to empty on drying, it could be suggested that the conductivity should
be higher in the wetting state, in accord with Youngs's result (Poulovassilis 1969).
However, other unspecified liquid-distribution effects may also be involved.
250 WOODING & MOREL-SEYTOUX

THE MACROSCOPIC FLOW EQUATIONS

In the petroleum industry, reservoir engineers have been faced with problems of
multi phase flow since production from oil fields began. The theoretical approach
adopted, and also followed by other disciplines such as hydrology and soil science,
is to formulate the equations in the macroscopic, or Darcy, sense [reviewed by
Philip (1970)], involving scales much larger than the pore scale. While the form of
Darcy's law is conditioned by the assumption of a steady, Stokes-flow regime and
uniformity in some statistical sense, it is partially empirical and has been verified
in the laboratory for simple flow geometries.
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Muskat (1949) [cf also Douglas, Peaceman & Rachford (1959)] generalizes
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Darcy's law for simultaneously flowing immiscible phases in vector form as


(6)

for phase i. writing Aj = kkr;/Jli for the mobility of that phase (Morel-Seytoux
1973a). Here v is the volume flow rate, or velocity in the Darcy sense, k is the
intrinsic permeability of the medium, and kr is the relative permeability to that
phase (Morel-Seytoux 1969), p and Jl are the phase density and dynamic viscosity,
respectively, P is the pressure within the phase, and g is the gravity vector. In
gravity-dominated systems it is often convenient to define the hydraulic conductivity
as Ki = gpiAi
For mass conservation of phase i. we write
(7)

in which n is the porosity of the medium and IJ denotes saturation with respect to
phase i. i.e. the volume of phase i per unit volume of voids. (An alternative
definition, frequently used, refers the phase volume to the volume of medium plus
voids and is equal to nlJ in the present notation. In oil-field literature, saturation
is usually denoted by S. equivalent to IJ.) Since the totality of phase fills the void
space in the medium, we have I:.i Oi = 1.
Because of the near absence of hysteresis, both the relative permeability and
hydraulic conductivity are taken to be uniquely related to saturation o.
Elimination of Vj between (6) and (7) gives an equation for each phase involving
only scalar independent variables. However, additional information is needed in
order to specify the problem fully. Across the interface between two phases i and
j. the pressure difference is Pi-Pi liPjj; the static value orMji may be known
=

as a function of IJj and OJ. Note, however, that the relationship is subject to
hysteresis. Equations of state relating pressures and saturations may also be known.
At present, most studies have been concerned with the simplest important class of
problems-that of two-phase flow-which is more easily specified fully. In this case
we can write IJ == IJl = 1-02 and P == Pl = P2-liP. where liP may be treated as
a single-valued function of 0 during either wetting or draining (although the
functions are not the same in the two cases).
As an example of a steady two-phase flow, consider a horizontal porous layer
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 251

containing a liquid, with gas released at a steady rate at the lower boundary. If
there is to be zero liquid flow (i.e. the gas does not expel the liquid), (6) gives
simply PI = gpl z + constant (with the previously defined suffices), where
. z is the
downward vertical coordinate. For the gas, (6) gives
V, =
-A/{g(PI - PI) + o (dP)/oz} , (8)
i.e. the gas flows upward through the stationary liquid in the pores under the
influence of buoyancy and the capillarity gradient. As the gas flow VI is specified,
it is necessary that AI have a value that will satisfy (8); the saturation 0 will be
such that kr/(8) has the appropriate value. A mechanism of this type has been
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

developed by McNabb, Grant & Robinson ( 1975) as a basic solution for two-phase
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

flow of steam and/or carbon dioxide in the groundwater of certain geothermal


areas. The thickness of such two-phase zones may be of the order of 1 km.
Evidently the greatest mathematical difficulties associated with solving macro
scopic equations (6) and (7) arise from the nonlinear physical relationships,
especially those involving the relative permeability, the capillarity effects, and degree
of saturation. It is natural that an extensive literature on numerical methods of
treating this system should have appeared. A very comprehensive review has been
given recently by Settari & Aziz (1975). These authors consider particularly the
inherent nonlinearities and the associated stability problems and truncation errors
arising in the numerical treatments.
A treatment of the numerical liteFature is regarded as outside the scope of the
present review. Here we shall attempt to trace the development of some of the
analytical or quasi-analytical ideas associated with problems of multi phase flow.

Buckley-Leverett Theory
In the practice of water flooding of an oil field (Smith 1966), water is injected at
high pressure to dIsplace oil. Provided the field pressure has not dropped significantly
since the beginning of production, no significant mobile-gas phase will have evolved
from solution (Craft & Hawkins 1964, Morel-Seytoux 1973a) and only two phases
move, i.e. water and oil. If, in addition, flow is mainly one-dimensional and
horizontal (in the x-direction), the volume flow rate of the combined fluids,
v = Vw + Vo in an obvious notation, is independent of x, and the gravity terms do not

enter. Buckley & Leverett (1942) have introduced the fractional flow function
fw vw/v. With the further assumption that, under usual injection and reservoir
==

pressures, the water-oil capillary pressure is insignificant, it becomes possible to


eliminate pressure from (6), and a single equation in a single unknown, 0, is
obtained from (7), namely,
n oO/ot + v(t) ofw(O)/ox = 0, (9)

where fw(O) (1 + Aoj)'w)- l. Typical curves of relative permeability and of f for


=

different viscosity ratios as a function of 0 are given in many articles and books
(e.g. Pirson 1958, Scheidegger 1960, Smith 1966, Morel-Seytoux 1969, 1973a, Bear
1972, Houpeurt 1974, Institut Franais du Petrole 1974). (See Figure 7.) Equation
(9) can be rewritten in characteristic form and integrated to obtain
252 WOODING & MOREL-SEYTOUX

-:.;;-_:;:OO'"
1 .0
I
- - - -- - - - - - - - - - - - - - --- - -

1
I
0.8 ,
I
I
,
0.6 I
..! ,
I
I
0.4 I
1
I, / //
/ //
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

0. 2 /
9wr I / /!!/
/ ' 9 0r
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

cl //
,
1 7---'---
,,. , , , ,
o 0. 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
S a tu r a t i o n

Figure 7 Effect of viscosity ratio on the fractional flow function f (after Morel-Seytoux
1973a, p. 141).

(J = constant, x,((J) = xo((J) + n- !f'(6) t vCr) dr = Xo + n - ! WI', (10)


(a) (b)
where the prime signifies differentiation with respect to 8, Xo is the initial abscissa
of the given water saturation 8, and W is the cumulative injected volume per unit
c!'03s-sectional area ( = length in this one-dimensional formulation). The derivation
can be explained from elementary mathematical notions (Morel-Seytoux 1969).
The complication with the Buckley-Leverett approach is that (10) generally leads
to a profile that is multiple-valued in () as a function of x, requiring the introduction
of a discontinuity or "front" (Figure 8). Buckley & Leverett (1942, p. 111) fully
understood the reason why their model broke down and the need for correction
by a front :
The new computed curve is S-shaped and is triple-valued over a portion of its length,
obviously a physical impossibility. ! The correct interpretation is that a portion of the
curve is imaginary and that the real saturation-distance curve is discontinuous. The
imaginary part of the curve is dotted in Figure 8 and the real distribution curve is
shown by the solid line labelled Q\> discontinuous at U2. The position of the plane UI
is determined by a material balance, the shaded area between the original and the new
saturation curve being equal to QdA. [QI is cumulative injection volume and A is
injection area.]
In any actual displacement of oil from sand by gas or water, no such saturation
discontinuity as that indicated in Figure 8 can exist in a uniform sand .
forces arising from the interfacial tension between oil and the displacing fluid and the
curvature of the interfaces in the sand tend in all cases to maintain uniform saturation

I Italics are present authors', and the figure number has been changed to match the

present paper.
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 253

throughout any continuous homogeneous sand body. The degree of equalization


obtained depends upon the combined effects of the capillary pressure gradient,
gravitational pressure gradient, and the impressed pressure gradient. Although in the
regions of gradual transition of saturation with distance the capillary pressure gradient
may be small in comparison with the other forces, at any saturation discontinuity the
capillary pressure gradient would become exceedingly large, with the result that the
plane of saturation discontinuity would be converted into a zone of more gradual
transition in saturation, the width of the zone depending for a given system primarily
upon the rate of displacement.

Many articles have been written on this problem (e.g. Welge 1952, Cardwell
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

1 959, Sheldon, Zondek & Cardwell 1959). In particular, they establish the analogy
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

between the Buckley-Leverett front and the shock wave of supersonic aerodynamics
(Courant & Friedrichs 1948).
Petroleum engineers realized that a more general fractional-flow function could
be derived. If one-dimensional flow is assumed but gravity and capillarity terms
are retained, this takes the form (cf Morel-Seytoux 1973a)
1Fw = fw{1 +Ao v - 1 [(pw - Po)g sin lX + a(p)/ax]}
(11)

for flow with a downslope 0( . For a water-air system, the quantities with subscript
o would be assigned air values. The functions of water saturation Gw (dimensions

1 00 r-------T---,-----oro___-__r--
et-
c:

o
..-
80 ,..--'--+-----t---+---t
o
.....
::J
-0 60 r----+-----t---t
en

5 10 15 20 25 30
U, D i s t a n c e : A r b i t r a r y U n i t s
Figure 8 Calculation of saturation history during water flood (after Buckley & Leverett
1942, p. 1 1 1).
254 WOODING & MOREL-SEYTOUX

of a velocity) and Ew (dimensions of a diffusivity) are obtained by identification


of terms in ( 1 1 ) and represent effects of gravity and capillarity, respectively.
Formally, the Buckley-Leverett approach would lead to an equation for the
propagation of a given water saturation (J as dx/dt.= vIF:'. the prime indicating
differentiation with respect to (J at a fixed 'time. If the term involving capillarity Eo
is small, this is integrable, leading to

x, = xo + n - 1!'(J) W(t) + n - 1 G'(J)t. (12)

Again the problem of multiple values arises and is resolved by the introduction of
fronts (Martin 1960, Morel-Seytoux 1967). The "image" of a front in the fractional
flow domain is a strajght line ; that is, the introduction of a front in the saturation
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

domain (Figure 9) is equivalent to replacing a portion of the IF-curve between two


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

points by a chord or secant (Figure 10).


Valuable insight has been gained by the use of graphical methods for the solution
of such problems (e.g. Welge 1952, Morel-Seytoux 1969) including some cases of
three-phase flows (Morel-Seytoux 1 973a). The success of the Welg construction
(Morel-Seytoux & Khanji 1975), which invol yes replacing part of the curve uf fw
(or F w ) by a tangent meeting the toe of the profile and preserves material balance,
appears to be due to the fact that it corresponds to the time-asymptotic limiting

0. 9 ...

--- B u c k l e y - L e v e r e t t
0.8
3 - S econts Approx i m a t ion

.....

j
0.7 ..........
...........
'
0.6 ' o t"'.,, , " - -" __
'
-- -
-
Inter face 2 -- - .
-- -
c: 0.5 -- - - -
0

0

::>
0 0.4
If)

0.3 I n t e r face 3

0.2

0.1

I I I I I I I I I I
0. 1 0.2 0.3 0.4 0. 5 0.6 0.7 0.8 0.9 1 .0 1. 1

Reduced D i s t a n c e

Figure 9 Saturation profiles.


MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 255

form of the unknown function IF. The realization that for large times the IF-curve
will consist largely of a straight line, i.e. that the saturation profile will evolve
almost entirely by translation, led Terwilliger et al (1951) (cf also Morel-Seytoux
1969) to the determination of the shape of the stabilized profile. However, their
method encounters difficulty in matching the translating profile to the "stretching
tail" profile ; it is able to predict only the profile shape in the front region but not
its actual position in space.
In order to include effects of capillarity, Le Fur (1962) has applied the boundary
layer approach to the equations for immiscible fluid flow. As viscous terms are
neglected in the Navier-Stokes equations to obtain the boundary-layer equations
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

1.1
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

1 .0

0.9

0.8

0.7

0.6

..!
0.5

0.4

0.3

0.2

0.1

o I I I I I I I
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Ow
Figure 10 Fractional flow curve showing "three-secants" approximation.
256 WOODING & MOREL-SEYTOUX

(Schlichting 1960), similarly some capillarity terms can be neglected in the


saturation equation. Again, following classical boundary-layer technique, the
simplified saturation equation is matched to the Buckley-Leverett solution adjacent
to the wetting front (Le Fur 1962, Morel-Seytoux 1969, 1973a). Although Le Fur
assumes that the total velocity is a given constant, the formal derivation using
Carrier's (1953) strained-coordinate method is not affected if v varies with time
(Noblanc 1970). In almost all petroleum applications, it can be assumed that v is
controlled by the injection pressure of the displacing fluid.

THE HYDROLOGIC PROBLEM


Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Buckley-Leverett theory has not been utilized to any great extent by hydrologists,
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

who are concerned particularly with the relatively thin moisture boundary layer
close to the ground surface. External pressure gradients are usually negligible. The
depth of ponding of water over the soil surface may be negligible or null. Forces
of capillarity and gravity, which are often comparable in magnitude, will determine
the velocity, which is unknown a priori.
In the traditional approach, the effect of displaced air upon infiltration is
neglected entirely. In (6) there is no contribution to VP from air flow, and the only
pressure effect upon the water phase is that due to capillarity-LV>. From (6) and
(7), dropping the suffix, we obtain
(13)
This is Richards's equation (Richards 1931, Philip 1970) in one of its forms. Both

A and K can be taken as one-valued functions of e. However, Il.P exhibits hysteresis,
as already discussed, and it is customary to restrict the use of (13) to monotonic
changes in e-either wetting or draining. Either Il.P may be selected as the dependent
variable (Rubin 1968) or, more traditionally, fJ may be selected (Klute 1 952). In the
latter case, the right-hand side of (13) is often written as V ' [D(8)Ve], where
D == A(Il.P)' is the diffusivity (Philip 1969). The diffusivity is sometimes generalized
to include the effects of other transport phenomen such as vapor diffusion. Thus
its physical significance may be complex, and it may behave either as a smoothly
continuous or as a piecewise continuous function of e. A further complication is
that D is singular at natural saturation (Morel-Seytoux, Khanji & Vachaud 1973),
which makes it both very difficult to measure directly in that region and impossible
to extrapolate objectively in that crucial range of water content.
An extensive literature exists on solutions of Richards's equation (Philip 1969,
1970), particularly for one-dimensional absorption or imbibition (capillary uptake
of water in the absence of gravity) and for vertical infiltration, in which gravity
becomes progressively more important with time. An early numerical solution has
been used by Philip (1957) to evaluate D(e) in detail from the experimental data
of Moore (1939). This has served as a useful test case for many of the quasi
analytical procedures subsequently developed.
Usually, the theoretical initial and boundary conditions assumed are either a
saturation step function-8 80 for t 0, z > 0 (z-axis directed downwards), 8 81
= = ==
MULTlPHASE FLUID FLOW THROUGH POROUS MEDIA 257

t
for 0, z O-or a correspondingly prescribed step function in the boundary
=

flux.
For one-dimensional absorption, Richards's equation is a nonlinear diffusion
equation, and the similarity variable X/til2
eliminates t for the conditions imposed
above. Vertical infiltration is more difficult ; physically, a diffusive boundary layer
first forms adjacent to z = 0 and then moves into the medium under the influence
of gravity. Expansions in tI12
are useful only at small times. An iterative scheme
proposed for this problem by Parlange (1971a,b) is, in principle, uniformly valid
z
in time. If the appropriate form of (13) is inverted so that becomes the dependent
variable, integration with respect to gives8
f' (oz/ot) dO,
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

[K - D(8)/(oz/oO)]:'
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

= Q- V = n (14)

(a) (b)

where we have defined a volume flux V, with V(81) = Q. This follows the notation
of Knight & Philip ( 1974) ; (14a) is a particular form of Darcy's law, and ( 14b) is
an integrated form of the continuity equation. As a zeroth-order approximation,
ParJange takes V = Q, assuming that the integral in ( 14b) is small. This is valid as
saturation is approached and, as the expansion is effectively about 0 = 01, accuracy
diminishes as 0 -+ 80, Parlange then proposes to iterate, using (14a) and ( 14b) in
tum. However, Knight & Philip ( 1974) point out that this particular iterative
process is divergent and that continuity, in the integral sense, is not maintained
after the first approximation. Philip & Knight (1974) then propose an alternative
iterative scheme that establishes integral continuity at each step. Independently,
Cisler (1974) has suggested an "improvement" to Parlange's second approximation,
which is really a further iteration by a different path, using the inverted differential
equation, which appears to converge. For the test case of Yolo light clay, the
numerical error of about 1 % or less falls between the second and third approxi
mations of Philip and Knight.
Generally, iterative methods generate expansions that are asymptotic in character
and therefore nonunique and inherently divergent (Van Dyke 1964, pp. 30ff, 34ft).
In some cases it may be a matter of trial to determine which variants of the
scheme give useful approximations.
In contrast to the close attention given in the literature to the above special
problem, apparently few analytical studies exist for infiltration described by
Richards's equation with an arbitrary, time-dependent boundary condition on the
saturation or the flux. Wooding (1975) treats such a case in which it is assumed
that the porous material is fairly permeable, e.g. a sandy soil, and the infiltration
zone depth L is large enough for the parameter )"(!l.P),/(A'gpL) to be small relative
to unity. This parameter is the coefficient of the capillary term after ( 1 3) has been
rendered nondimensional. Solution of the problem is carried out by matched
asymptotic expansions (Van Dyke 1964, Chapter 5). Away from the steep gradients
of saturation 0, the capillary effect is small and an outer expansion of Buckley
Leverett type is obtained. In the neighborhood of saturation fronts, however,
capillary and gravity effects are of comparable magnitude ; each front is described by
258 WOODING & MOREL-SEYfOUX

means of an inner expansion. Positions of fronts are determined by deriving jump


conditions based on mass conservation.
It is interesting to note that, since the first term of the outer expansion does
not involve capillarity, hysteresis does not enter at that stage. The effects of
capillary hysteresis upon higher terms can therefore be calculated directly, i.e.
noniteratively, on the assumption that the hysteresis properties of the medium are
known. This is of some importance in the present instance because sandy materials,
in particular, exhibit considerable hysteresis (Figure 6).

Two-Phase Infiltration
Most texts that discuss Richards's equation fail to mention the basic assumption
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org


that air pressure remains uniform with depth and constant with time-and the
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

experimental verification of the validity of the approach has invariably been reported
for cases where the columns of soil are carefully vented (Morel-Seytoux 1973b,
Noblanc & Morel-Seytoux 1974). On the other hand, a careful review of the
experimental evidence (in the second paper) shows that air effects are far from
negligible. In spite of the widely held belief that air can easily escape when displaced
by infiltrating water, recent field investigations (Linden & Dixon 1975) show clearly
that this is not the case. Naturally, when measurements are taken with an air effect
present, the effect may be included, albeit unknowingly, in the measured quantities.
If, instead of neglecting air effects, one formulates the problem as a two-phase
flow problem (which it is), two partial differential equations must be solved rather
than one (Phuc & Morel-Seytoux 1972). Under not very restrictive conditions,
however, the system can be reduced to a single differential equation in 8(z, and t)
1.5 I
I
I
I
I
I
I
1.0 - - - ----- +- - - - - - - - - - - - -
I
I
I
I
I
I
0.5 I
I
I
I
-'

l
I
I
I I I I I I I
o 0.1 0.2 0.3 Q4 0.5 0.6 0.7 0.8 0.9 1.0
Waler Saturation

Figure 1 1 Effect of total velocity on the fractional flow function F for a viscosity ratio
w

of 100 (after Morel-Seytoux 1973a, p. 1 51).


MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 259

the total velocity v(t), and one algebraic equation (Morel-Seytoux 1973a, Morel
Seytoux & Khanji 1974a).
While relatively few hydrologists and soil scientists have yet adopted the two-phase
approach, facets of the theory are entering the literature in indirect ways. The
"apparently new viewpoint of the flux-concentration relation" is introduced rather
cautiously (Philip 1973b), in contrast to the direct application of two-phase
concepts to infiltration by Brustkem & Morel-Seytoux (1970, 1975), Morel-Seytoux
& Noblanc (1970), Noblanc & Morel-Seytoux (1972), Morel-Seytoux & Khanji
(1974a, 1975), Morel-Seytoux (1974a), and to drainage by Morel-Seytoux (1974b).
In the notation of (14), Philip's flux-concentration relation is defined as fF =
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

(V K)/(Q - K) for infiltration, and the hydraulic conductivity K is set equal to 0


-
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

for absorption. There is it superficial resemblance to the fractional flow function


fw = vw/v. However, since the influence of the air phase is neglected in Richards's
equation, it does not enter the continuity relationship and the concept of the total
velocity v becomes irrelevant ; g- loses some of the physical significance and
usefulness attached to fw. A further limitation appears to be the requirement that
fF is a nonincreasing function of the space coordinates (Philip 1973b). In fact, many
significant infiltration problems display nonmonotonic curves [cf Figure 11 and
Morel-Seytoux (1967, 1973a, 1974a), Brustkem & Morel-Seytoux (1975)].
In two-phase infiltration, determination of the total velocity v is the central
problem. Brustkem & Morel-Seytoux (1970) combine a generalized Buckley-Leverett
approach with a method for the determination of the total velocity, which accounts
for the effect of capillarity as a driving mechanism but neglects some of the capillary
effects in the determination of the total resistance to flow. The validity of the
approach has been established by comparison with a numerical solution that
solves the complete system of equations (Morel-Seytoux 1973a, p. 155). The
effective capillary drive is obtained exactly without the need for prior determination
of the saturation profile (Brustkern & Morel-Seytoux 1970, p. 2539), and the
significant effect of air compression and counterflow on the infiltration rates is
clearly demonstrated (Figure 12). This figure shows that the standard formulas of
infiltration (e.g. Horton, Kostiakov, Philip, Holtan) that yield consistently monotoni
cally decreasing rates lack physical significance, although they probably are
adequate for most practical situations.
Next (Morel-Seytoux & Noblanc 1970, Noblanc & Morel-Seytoux 1972), the
strained-coordinate and boundary-layer methods are combined with the integral
equation for the determination of total velocity. Very realistic moisture profiles are
obtained. It is shown also (Morel-Seytoux 1972) that solutions of the unsaturated
flow (Richards) equation can be obtained by this approach, although the physically
more realistic two-phase formulation can be solved just as easily.
Recentstudies comparing the "empirical" version of the Green and Ampt formula
[equation (4)] with other "physically based" infiltration equations have tended to
neglect air effects (Swartzendruber 1974, Swartzendruber & Youngs 1974). These
authors note that the two-term Philip infiltration equation (Philip 1969, p. 284)
gives higher reults than does the Green and Ampt equation but never by more
than about 15%, which "for practical purposes . . . may frequently be ignored."
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

tv
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

g;

I
z
Cl



i3

j2

Jr v
) i 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 30 3.2 3.4 3.6 3.8

Tim.lhr.)

Figure 12 Comparison of infiltration rates when the depth of water table (D) varies (after Morel-Seytoux 1973a, p. 161).
MULTIPHASE FLUID FLOW THROUGH POROUS MEDiA 261

Morel-Seytoux & Khanji ( 1974a) also observe that, experimentally, the Green and
Ampt formula yields the better predictions. They derive the correct expression for
the effective capillary drive in the Green and Ampt equation and determine values
of the viscous correction factor for several soils. Subsequently, Morel-Seytoux
(1974a) and Morel-Seytoux & Khanji ( 1974b) take compression and counterflow
effects into account, leading to further modifications of the Green and Ampt
formula.
Once it is realized that a drainage problem is one of air infiltration. it seems
natural that the two-phase theory should lead, quite symmetrically, to a drainage
equation that has the same functional form as the Green and Ampt equation.
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Youngs's proven formula (Whisler & Bouwer 1 970) appears as a special case of
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

this more comprehensive drainage equation (Morel-Seytoux 1974b, p. 187).

MACROSCOPIC INSTABILITY IN DISPLACEMENT FLOWS

Experience of oil production in the presence of a natural or artificial water drive,


or gas drive, has been responsible for an early awareness that simple piston models
of displacement flows are not necessarily adequate to explain "break-through"
phenomena. In particular, water may appear at producing wells after only a small
proportion of the estimated oil volume has been extracted. In the laboratory model
demonstration of Figure 14a, this proportion amounts to about 1 3 %. Since on
Taylor'S (1961) results more than 40% should be extracted from individual pores, it
becomes evident that the water is tending to follow "preferred channels." As
Figure 14a also demonstrates, uneconomically large volumes of water must pass
through the system to expel further oil in significant quantities (van Meurs & van
der Poel 1958).
Dietz (1953) and Kidder (1956) have investigated some problems connected with
the stability of an interface in a porous medium. This early work has been
extended to include an impressed velocity and variable viscosity (though neglecting
interfacial tension) in a classic paper by Saffman & Taylor (1958), which includes
experiments based upon the Hele-Shaw analogy of two-dimensional flow in a
porous medium. For an immiscible interface in a Hele-Shaw cell, the effect of
interfacial tension upon the linearized instability can be calculated readily within
the limits of the Hele-Shaw approximation (Chuoke, van Meurs & van der Poel
1959).
In the problem most intensively studied, a horizontal interface separates two
immiscible fluids in a vertical cell-fluid 1 below and fluid 2 above. The motion
of each fluid satisfies Darcy's law (6) with i = 1,2. Although this problem would
appear to involve the interface between two fluids in single-phase flow, the use of
relative permeability kr is usually necessary since a film of the displaced fluid
remains on the solid walls. The continuity condition is usually taken as V ' Vi = 0,
implying that the residual fluid film is immobile. This is probably satisfactory for
the thin films typical of low velocities of displacement. For uniform film thickness
(another approximation), the functions PI + gPi Z <Ili (i = 1 ,2) satisfy Laplace's
=

equation in the plane of the cell, with the z-axis upwards. A uniform upward
262 WOODING & MOREL-SEYTOUX

velocity U may be applied to the fluid in the system, giving QrJ>i/oz .... - U as
Izl .... 00 . The balance of normal forces at the interface and the kinematic condition
at the interface give rise to nonlinear boundary conditions (Nayfeh 1972).
A solution of the linearized stability problem is of the form
" = e exp (O'ot) cos mx,
where " is the displacement of the interface from the mean and m is the wave number.
If the cell is closed by vertical side-walls that limit the horizontal (x-) dimension,
then m is an eigenvalue. The characteristic equation determining the linear growth
rate 0'0. given by Chuoke, van Meurs, and van der Poel, and verified experimentally
(cf Figure 1 3a), is
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

o
0.22
m.

(a) (b)
Figure 13 Displacement of oil by water in two Hele-Shaw cells. (a) Development of
asymmetrical fingers at oil-water viscosity ratio 2.5 (drawn from photographs by Chuoke,
van Meurs & van der Poe! 1959) and (b) transition of mUltiple fingers to a single finger at
oil-glycerine and water viscosity ratio 17 (after Gupta, Varnon & Greenkorn 1973).
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 263

0"0(A,1 1 + A,i l)/m = (A,i l - A,1 1) U + (P2 - P I)g - m 2y, (15)

where y is the interfacial tension. The right-hand side involves contributi6ns to the
linear growth rate due to mobility contrast, gravity contrast, and interfacial tension.
Equation (15) gives a maximum growth rate at a finite value of wave number m,
and there is a finite cut-off wave number. On the other hand, if interfacial
tension is disregarded, we have Go oc m and there is no maximum growth rate and
no cut-off wave number.
Effects of weak nonlinearity at small 'but finite amplitudes have been shown by
Nayfeh (1972) to render the stability conditions amplitude dependent. The principal
effect is destabilizing, due to a factor (1 + IV'112)- 1/2 in the coefficient of y, i.e. due
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

to a reduction in the effect of the surface-tension force as the amplitude increases.


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Deviations from linearity of Darcy's law at high flow rates produce an effective
body force opposite in direction to that of the flow, and so are stabilizing or
destabilizing depending upon whether the flow is from the denser to the less dense
fluid, or vice versa (Yih 1965, Saville 1969, Nayfeh 1972).
Equation ( 1 5) is also of interest to infiltration studies based on the Green and
Ampt model [equation (4)] , which assumes the existence of a step function in
saturation. If the upper fluid 2 is water, while the lower fluid 1 (air) is neglected,
(15) reduces to O"o/m = U + K and a capillary term that is stabilizing but is
negligible for small m, From its definition, the hydraulic conductivity K is greater
than 0, but U < 0 for downflow. The system is stable (0"0 0), provided that
- U K. This should hold for a homogeneous medium as the depth of infiltration
increases, since - U --> K. the limit being approached from above as infiltration is
assisted by capillary drive. Ponding on the surface also increases infiltration rate,
contributing to stability. On the other hand, instability may be induced by a
reduction of lui, e.g. by compression of air ahead of the front (Peck 1965a, b), or by
the presence of a low-permeability layer behind the front, as clearly demonstrated
by Hill & Parlange (1972, Figure 14b). It is possible that some recent studies on
the application of Green-Ampt theory to infiltration through a surface crust by
Ahuja (1974) as developed from Hillel & Gardner (1970) contain an inherent
instability. A gradual (i.e. nonabrupt) change of parameters such as soil
permeability with depth may also affect the stability characteristics of the wetting
front. Some of the qualitative effects of non uniformity have been considered by
Raats (1973), although precise stability criteria are not treated. For typical soil
parameters, instability effects may not appear in laboratory experiments using
samples in packed columns of limited horizontal dimensions. In the field, also,
systems already water wet may exhibit an enhanced degree of stability (Richardson
1961, Gupta & Greenkorn 1974), possibly due to the increased capiIIary diffusivity.
In a study related to the formation of water-oil emulsions during oil production,
Raghaven & Marsden (1971) consider the linearized stability analysis of a piston
model consisting of two fluids of different properties separated by a homogeneously
emulsified fluid zone-effectively a third fluid-so that two interfaces exist [see also
Nayfeh (1972)]. For this "generalized Saffman-Taylor problem," the interfaces
behave independently at perturbation wavelengths small in comparison with the
264 WOODING & MOREL-SEYTOUX

n ,. 0 .. t. 3,.0 R = 1 3,. ; 0 ,. l. '10


Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

R " 0 .. 6. 0,-. R lO,"; 0 " ., '"

R " 34.,.; 0 c 1 80,"

R .. 0 l Z ,. R .. 5/',.; 0 .. b50,.
(a)
Figure 14 Illustrations of the growth of fingers in porous media. (a) Linear displacement
of oil by water at oil-water viscosity ratio 80 in a three-dimensional porous medium
arranged as a Christiansen filter (cf Deutsch 1960), showing immobility of water protrusions
\arrow) ; Q-total volume expelled, R-"recovery" volume, expressed as percentages of total
pore volume (after van Meurs 1957, van Meurs & van der Poel 1958). (b) Unstable
MULTIPHASE FLUID FWW THROUGH POROUS MEDIA 265
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

(b)

(c )

infiltration of water displacing air in coarse sand (500-- 1 000 Jlm grain size) overlain by a
"retarding" layer of fine sand (5(}-100 Jlm grain size) in a transparent chamber 0.75 m long.
0.30 m high. and 0.025 m deep (after Hill & Parlange 1972). (e) A gravitationally unstable
diffusing interface between miscible fluids of equal viscosity in a Hele-Shaw cell 0.25 m
wide (Wooding 1969a b). .
266 WOODING & MOREL-SEYTOUX

layer thickness, but coupling increases with wavelength until finally the mixed wne
behaves as a single interface.
An alternative and seemingly more realistic approach to this question is due to
Hagoort (1971), who treats the stability of the one-dimensional Buckley-Leverett
displacement flow (Figure 8) with the front modified by capillarity to the shape of
a "diffuse" profile. The thickness of this frontal zone depends upon the capillarity
viscosity interaction and is measured by the length scale J..(fl.P),/v. where v is total
velocity, or, for infiltration under gravity described by ( 1 3), by the dimensionless
capillarity coefficient (Wooding- 1975). Saffman-Taylor theory indicates that the
frontal wne is unstable when the effective mobility behind the front exceeds that
ahead of the front, but the finite thickness of the frontal zone ensures the existence
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

of a cut-off wave number and a (smaller) wave number of maximum linear growth
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

rate. For flow along a sufficiently narrow channel, the lowest possible wave number
can equal or exceed the cut-off value, so that the flow is stable.

Growth of Fingers
In spite of the apparent significance of fingering in mixing processes at unstable
interfaces in porous media, the phenomenon seems not to be well understood.
During the inherently unstable process of water drive to expel oil, do fingers
remain small and tend to "interdiffuse," or do a relatively small number of "giant"
fingers develop? Are laboratory studies such as those of van Meurs (1957),
Richardson (1961), and later workers, using dimensional analysis and scaling, the
most satisfactory approach?
Studies of fingering in Hele-Shaw cells have received attention, but their
relevance to porous media, where the flow is three-dimensional and "dispersion"
of the interface can occur by Taylor'S (1961) mechanism, is not clear.
Saffman & Taylor (1958) have described a class of "exact" solutions for the
profile of a single two-dimensional finger advancing steadily along a Hele-Shaw
channel, but find experimentally that just one particular solution is selected-that
for which the finger width is one half of the channel width-provided that jJ.U/y
is not small. As far as is known, two-dimensional studies have not yielded any
criterion to account for this result. It has been conjectured that, since a film of
displaced fluid remains on the cell walls, the displacement process should be
treated as a three-dimensional problem. In a further theoretical development,
Saffman (1959) has obtained exact solutions for the growth, from an initially nearly
flat surface, of a periodic array of fingers. The profile is antisymmetric about the
line of the original interface, which is probably true for fluids of equal viscosity
(Figure 14c). However, where the viscosity of the displaced fluid considerably
exceeds that of the displacing fluid, a noticeable departure from precise anti
symmetry appears at finite amplitudes (Figure 13a). This could be partly a
three-dimensional effect ; highly viscous fluid is left behind on the cell walls, some
of it being removed from the lower parts of the finger so that the lower parts of
the profile appear narrow relative to the leading parts of the profile. Such a
mechanism may contribute to merging of the bases of the fingers-a phenomenon
noted by Gupta and co-workers.
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 267
Strong coalescence effects between fingers in a long Hele-Shaw cell at high
viscosity ratios have been reported by Gupta, Varnon & Greenkom (1973, Figure
13b) and Gupta & Greenkom (1974). In these experiments, the final state is that
of the single central finger studied by Saffman and Taylor, irrespective of the initial
number of fingers, and it is not certain that the sidewalls are not influencing the
result. The explanation of Gupta and coauthors relates to the pressure gradient
and the viscosity contrast ; and advancing low-viscosity finger reduces the longi
tudinal pressure gradient in its neighborhood, so that a nearby smaller finger
experiences a reduced driving-pressure gradient ; its amplitude growth rate may
actually become negative. An effect of this type has also been observed in
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

nuerical modeling ofgeothermal convection, involving plumes of hot, low-viscosity


by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

groundwater (E. Bradford, unpublished).


In an experimental and theoretical study of fingers at an unstable interface
between misdble fluids, Wooding (1969a,b) also observes coalescence. Since there
is no viscosity contrast in this case, the mechanism suggested by Gupta and
coauthor would exert a weaker effect, and an explanation is suggested in terms of
diffusive entrainment, which the data appear to support. Although for immiscible
fluids in a porous medium the capillarity term exerts a pseudodiffusive effect, the
equations lack a term contributing to lateral entrainment.

MULTIPHASE FLOW AND HEAT FLOW UNDER


COLD CONDITIONS
Multiphase Flow in Snow
In spite of its importance in snow hydrology, the theory of fluid flow in porous
media has not been applied to water percolation through this peculiar granular
medium until very recently (Colbeck 1972). It can be expected that heat exchange
will take place concurrently with mass flow and that, for the general case, coupled
equations must be solved. Initially, however, Colbeck considers the isothermal case
to avoid the complication of the blocking effect of refreezing. There is also
considerable glaciological interest because isothermal flow does occur "in temperate
glacial fim and seasonal snowpacks during the ablation season" (Colbeck 1972, p.
370). Starting with the two-phase approach and making simplifying assumptions
because "water saturation is generally small in snow undergoing simple gravity
flow," Colbeck obtains an equation for water velocity that could have been
obtained from unsaturated-flow theory. However, there is an advantage in starting
with the two-phase approach since it provides insight into the conditions under
which the unsaturated-flow approach is adequate.
In addition, Colbeck's method of solution proceeds from a creative understanding
of the generalized Buckley-Leverett theory of wave and shock propagation. Like
Brustkem & Morel-Seytoux (1970) and Morel-Seytoux & Noblanc (1970), Colbeck
drops capillary terms from the Richards equation, obtains a hyperbolic partial
differential equation, and looks for a local solution by studying the velocity of
propagation in terms of water flux-the standard kinematic-wave approach. Here
there is no great difficulty in predicting the effect of an arbitrary time-varying flux
268 WOODING & MOREL-SEYTOUX

boundary condition, and good agreement with observation is reported. In a later


paper, Colbeck (1974) reintroduces the capillarity, viewed as a perturbation on a
gravity-dominated equation except in front regions. The conclusions based on the
calculations are most significant : the diffused appearance of wave fronts observed
in homogeneous snow cannot be attributed to capillary effects, and the shock-front
approximation is sufficient for most purposes.

Multiphase Flow in Soil


'
Experience and experiments have shown that, as a soil cools below the freezing
point, not only does the "local" water in the freezing zone tum into ice but so also
does "foreign" water that moves into the frozen zone from the surrounding
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

unfrozen soil and then freezes. When the amount of migrating water is important
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

relative to the in situ water, heaving may take place as a result of ice-lens formation
(Jumikis 1956). This is a matter of engineering concern, and many experiments have
been carried out to ascertain the effects of soil porosity (Jumikis 1969), soil type
(Aguirre-Puentes & Le Fur 1967), freezing rate (Penner 1972), and depth to water
table (McGaw 1972) on heaving and the rate of heave. A first conclusion is :
associated with the water movement is a soil deformation. Whereas this deformation
may not significantly affect the soil-temperature distribution, it will affect the
moisture distribution appreciably. The resultant moisture distribution in the soil at
the time of snowmelt may considerably affect the soil capacity to infiltrate water
(Peck 1973). ' It appears imperative to include the phenomenon of heave in the
description of the system.
The mechanism causing attraction of water into the frozen zone remained a
puzzle for a long time (Jumikis 1954, 1960, 1962, Aguirre-Puentes, Khastou &
Chalhoub 1971). The explanation was thought to be in terms of a critical pore
radius. This argument is still used (Miller 1973, Aguirre-Puentes, Vignes & Viaud
1972, Aguirre-Puentes & Azouni 1973) as an aid to qualitative reasoning. Gradually,
however, there has developed the concept of a capillary pressure between ice and
water in a soil, just as there is between air and water (Miller 1965), and that these
soii characteristics might possibly be deduced from one another (Koopmans &
Miller 1966). It is not altogether surprising that the idea comes from a soil scientist
(Miller), rather than from the engineers, who are probably aware of the capillary
rise of water in a fine tube but not familiar with the concept of a "moisture-retention
curve" (Baver, Gardner & Gardner 1972). A clear relationship between capillary
pressure and temperature has been introduced by Everett (1961) (cf Miller 1965,
p. 193).
In spite of these accomplishments in understanding the phenomenon, it is still
not possible to deduce the ice-water capillary-pressure curve from the air-water
curve except for a soil-soil contact of particles (Koopmans & Miller 1966, p. 680)
or a pure SLS soil (i.e. a soil with a liquid film between particles). In addition,
the relative permeability-saturation relationship and other hydraulic characteristics
of the water in the frozen zone are not quantitatively known with any accuracy.
Based upon actual measurements of the ice-water capillary-pressure curve
(Koopmans & Miller 1966, p. 683), a temperature drop of only O.2C is apparently
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 269

sufficient to leave in place just a small residual unfrozen water content, probably
of very low mobility (see Figure 15). The small temperature drop may occur over
a relatively thin zone (say a few centimeters), and it seems possible that this zone
of drastic reduction of water content-from a high value to a very low one-may
be treated as a front to a reasonably good approximation.
Relatively few attempts at mathematical description and solution of this class of
problem have appeared in the literature, and those only within the past few years
(Kennedy & Lielmezs 1973, Harlan 1973, Guymon & Luthin 1974, Kay &
Groenevelt 1974, Groenevelt & Kay 1974). The lack of satisfactory measurements
of soil properties under these conditions, needed for model descriptions, may be a
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

partial explanation. Also, the problem involves the combined areas of thermo
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

dynamics, heat transfer, multiphase flow in porous media, and soil mechanics.
Some of the assumptions made by the investigators of water movement under
freezing conditions are quite drastic (cf Kennedy & Lielmezs 1973, p. 395). These
authors formulate but do not solve their equations. So also do Kay and Groenevelt
in their elegant thermodynamic study. Harlan tests a mathematical model against
experimental data, involving a sequence of freezing, thawing, and infiltration, while
Guymon and Luthin obtain numerical solutions to some apparently realistic

A I R PRESSU RE , bars
o 2 3 4 5

. 40 o Free z i ng
8 "V D r ying
"-
5, 30
t-A

Z
20
z
o
<.>

0::
w
10
!;t
3:
0 - - 1

CL _I J
_0. 1
1 I.
-0.2
o 0.5 1.0 1 .5 2.0
TEM PERATURE , o c a ICE PRESSURE , bars

Figure 15 Soil-freezing characteristic data for second freezing and soil-water characteristic
data for drying for a solid-solid soil, 4-8 JIm fraction (after Koopmans & Miller 1966,
p. 683).
270 WOODING & MOREL-SEYTOUX

problems. However, these authors still assume that the porous medium is rigid ; the
soil-mechanics aspect of the problem is not attempted.

CONCLUSIONS

Some aspects of the theory of flow through porous media-notably those involving
the capillary geometry-may never be solved deterministically. However, physical
concepts sometimes lead to simple models that are very useful in these almost
insoluble problems. Besides Darcy's law, examples are the Green-AmptjWashburn
theory of capillary uptake and the Buckley-Leverett theory of immiscible displace
ment and its developments.
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Unfortunately, while some topics under the main heading have been almost
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

beaten to death, others suffer from neglect insofar as flow through porous media
is concerned. Within its scope, hopefully, this article may draw attention to the
importance of the phenomenon of wetting and its relationship to immiscible flow
in capillaries, the use of multiphase ideas in hydrology, unstable immiscible dis
placement flows in porous media, and liquid percolation under freezing conditions.

ACKNOWLEDGMENTS

The contribution of Professor M9rel-Seytoux was made possible by research


contracts from the Office of Water Research and Technology and from the
National Oceanic and Atmospheric Administration. The support of OWRT and
NOAA is gratefully acknowledged.

Literature Cited

Aguirre-Puentes, J., Azouni, M. 1973 . . IIF Bomba, S. J. 1968. Hysteresis and time-scale
IIR-Commission B1 Zurich, 1 973-1974, invariance in a glass-bead medium. PhD
101-3 thesis. Univ. Wis., Madison, Wis. 1 16 pp.
Aguirre-Puentes, J., Khastou, B., Chalhoub, Bretherton, F. P. 1961. J. Fluid Mech. 1 0 :
M. 1971. Proc. Int. Congr. Refrig., 13th, 1 66-88
Washington DC 1 : 759-64 Briant, J., Cuiec, L. 1971. Rev. lnst. Fr. Pet.
Aguirre-Puentes, J., Le Fur, B. 1967. Lab. Ann. Combust. Liq. 26 : 591-616
Aerotherm. CNRS Rep. 67-9. 12 pp. Brustkern, R. L., Morel-Seytoux, H. 1. 1970.
Aguirre-Puentes, J., Vignes, M., Viaud, P. Proc. Am. Soc. Civ. Eng. 96 : HYI2, 2535-
1972. Lab. Aerotherm. CNRS Rep. 72-3. 48
54 pp. Brustkern, R. L., Morel-Seytoux, H. J. 1975.
Ahuja, L. R. 1 974. Soil Sci. 1 18 : 283-'88 J. Hydrol. 24 : 21-35
Bailey, A. I., Price, A. G., Kay, S. M. 1970. Buchanan, J. E. 1967. Ind. Eng. Chern.
Spec. Discuss. Faraday Soc. No. 1 : 1 1 8-27 Fundam. 6 : 400-7
Bascom, W. D., Cottington, R. L., Single Buckley, S. E., Leverett, M. C. 1 942. Trans.
terry, C. R. 1964. Adv. Chern. Ser. 43 : Am. lnst. Min. Metall. Pet. Eng. 146 :
355-79 107-16
Baver, L. D., Gardner, W. H., Gardner, Cardwell, W. T. Jr. 1959. Pet. Trans. Am.
W. R. 1972. Soil Physics. New York : lnst. Min. Metall. Pet. Eng. 2 1 6 : 271-76
Wiley. 498 pp. Carrier, G. F. 1953. Adv. Appl. Mech. 3 :
Bear, J. 1972. Dynamics of Fluids in Porous 1-18
Media. New York : Am. Elsevier. 764 pp. Cetinbudaklar, A. G., Jamieson, G. J. 1969.
Blake,T. D., Haynes, J. M. 1969. J. Colloid Chern. Eng. Sci. 24 : 1669-80
Interface Sci. 30 : 421-23 Childs, E. C. 1969. The Physical Basis of
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 271

Soil Water Phenomena. London : Wiley. Sci. 4 : 1-24


493 pp. Groenevelt, P. H., Kay, B. D. 1974. Soil Sci.
Chuoke, R. L., van Meurs, P., van der Poel, Soc. Am. Proc. 38 : 400-4
C. 1959. J. Petrol. Tech. 1 1 : 64-71 Gupta, S. P., Greenkorn, R. A. 1974. Water
Churayev, N. V., Sobolev, V. D., Zorin, Resour. Res. 1 0 : 371-74
Z. M. 1970. Spec. Discuss. Faraday Soc. Gupta, S. P., Varnon, J. E., Greenkorn, R. A.
No. 1 : 213-20 1973. Water Resour. Res. 9 : 1039-46
Cisler, J. 1974. Soil Sci. 1 17 : 70-73 Guymon, G. L., Luthin, J. N. 1974. Water
Colbeck, S. C. 1972. J. Glaciol. 1 1 : 369-85 Resour. Res. 10 : 995-1001
Colbeck, S. C. 1974. J. Glaciol. 13 : 85-97 Hagoort, 1. 1971. The Stability of Water-Oil
Courant, R., Friedrichs, K. O. 1948. Super- Displacement in Connate- Water-Bearing
sonic Flow and Shock Waves. New York : Porous Media. Presented at IUTAM/
Interscience. 464 pp. IUGG Symp., Calgary, Alberta, Can.
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Craft, B. c., Hawkins, M. F. 1964. Applied Hardy, W. B., Doubleday, I. 1922. Proc; R.
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Petroleum Reservoir Engineering. Engle Soc. London Ser. A 1 00 : 550-74


wood Cliffs, NJ : Prentice-Hall Harlan, R. L. 1973. Water Resour. Res. 9 :
Davidson, J. M., Nielsen, D. R., Biggar, 1314-23
J. W. 1966. Soil Sci. Soc. Am. Proc. 30 : Haynes, W. B. 1930. J. Agric. Sci. 20 : 97-1 16
298-304 Hill, D. E., Parlange, J.-Y. 1972. Soil Sci.
Davies, C. N. 1973. Air Filtration. London : Soc. Am. Proc. 36 : 697-702
Academic. 171 pp. Hillel, D., Gardner, W. R. 1970. Soil Sci.
Deryaguin, B. V. 1940. Trans. Faraday Soc. 109 : 69-76
36 : 204-15 Houpeurt, A. 1974. Mechanique des Fluides
Deryaguin, B. V., Levi, S. M. 1964. Film dans les Milieux Poreux. Paris : Editions
Coating Theory. London : Focal Press. Technip. 385 pp.
190 pp. Huh, C., Scriven, L. E. 1971. J. Colloid
Deutsch, E. R. 1960. Nature 185 : 675-76 Interface Sci. 35 : 85-101
Dietz, D. N. 1953. Proc. K. Ned. Akad. Hutton, B. E. T., Leung, L. S. 1974. Chem.
Wet. Ser. B 56 : 83 Eng. Sci. 29 : 168 1-85
Douglas, J. Jr., Peaceman, D. W., Rachford, Hutton, B. E. T., Leung, L. S., Brooks,
H. H. Jr. 1959. Pet. Trans. Am. Inst. Min. P. C., Nicklin, D. J. 1974. Chem. Eng.
Metall. Pet. Eng. 2 1 6 : 297-308 Sci. 29 : 493-500
Dussan, E. B. V., Davis, S. H. 1974. J. Fluid Institut Frans:ais du Petrole. 1974. L Ex'

Mecho 65 : 71-95 ploitation des Gisements d'Hydrocarbures.


Eisenklam, P., Ford, L. H. 1962. Proc. Symp. Paris : Editions Technip. 290 pp.
Interaction Between Fluids and Particles, Jumikis, A. R. June 1954. Proc. Am. Soc. Civ.
333-42. London : Inst. Chern. Eng. Eng. 80 : Separate No. 445, 445-1-445-14
Elliott, G. E. P., Riddiford, A. C. 1967. J. Jumikis, A. R. 1956. Trans. Am. Geophys.
Colloid Interface Sci. 23 : 389-98 Union 37 : 181-84
EI-Shimi, A., Goddard, E. D. 1974. J. Colloid Jumikis, A. R. 1960. Highw. Res. Board
Interface Sci. 48 : 249-55 Proc. Ann. Meet 39th. Jan. 1960. 619-39
..

Everett, D. H. 1961. Trans. Faraday Soc. 57 : Jumikis, A. R. 1962. Highw. Res. Board Bull.
1541-51 3 1 7, 1-8
Everett, D. H. 1967. Solid Gas Interface, ed. Jumikis, A. R. 1969. Rutgers Univ. NJ Eng.
E. A. Flood, 2 : 1055- 1 1 13. New York : Res. Bull. 49. 71 pp.
Dekker Kay, B. D., Groenevelt, P. H. 1974. Soil Sci.
Evgen'ev, A. E. 1965. Sov. Phys.-Dokl. 10: Soc. Am. Proc. 38 : 395-400
41 1-12 Kennedy, G. F., Lielmezs, J. 1973. Water
Fairbrother, F., Stubbs, A. E. 1935. J. Chem. Resour. Res. 9 : 395-400
Soc. Pt. 1 : 527-29 Kidder, R. E. 1956. J. Appl. Phys. 27 : 154
Fox, H. W., Zisman, W. A. 1952. J. Colloid 48
Sci. 7 : 428-42 Klute, A. 1952. Soil Sci. 73 : 105-16
Frenkel, I. I. 1946. Kinetic Theory of Knight, J. H., Philip, J. R. 1974. Soil Sci.
Liquids. New York : Dover. 488 pp. 1 1 6 : 407-16
Glasstone, S., Laidler, K. J., Eyring, H. 1941. Koopmans, R. W. R., Miller, R. D. 1966.
The Theory of Rate Processes. New Soil Sci. Soc. Am. Proc. 30 : 680-85
York : McGraw-Hill. 6 1 1 pp. Le Fur, B. 1962. J. Mec. 1 : 213-32
Good, R. J. 1973. J. Colloid Interface Sci. Lerner, B. J., Grove, C. S. Jr. 1951. Ind. Eng.
42 : 473-77 Chem. 43 : 2125
Green, W. H., Ampt, G. A. 1911. J. Agric. Leverett, M. C. 1939. Trans. Am. Inst. Min.
272 WOODING & MOREL-SEYTOUX

Metall. Pet. Eng. 142 : 1 52-69 Morel-Seytoux, H. J., Khanji, J. 1975. Pre
Lighthill, M. J., Whitham, G. B. 1955. Proc. diction of Imbibition in a Horizontal
R. Soc. London Ser. A 229 : 317-45 Column, Soil Sci. Soc. Am. Proc. To
Linden, D. R., Dixon, R. M. 1975. Water appear July-August 1975
Resour. Res. 1 1 : 139-43 Morel-Seytoux, H. 1., Khanji, J., Vachaud,
Lobo, W. E., Friend, L., Hashmall, F., G. 1973. Colo. State Univ. Civil Eng. Rep.
Zenz, F. 1945. Trans. AIChE 41 : 693-710 CEP72-73HJM48. 32 pp.
Lockhart, R. W., Martinelli, R. C. 1949. Morel-Seytoux, H. J., Noblanc, A. 1970.
Chern. Eng. Prog. 45 : 39-48 Colo. State Univ. Civil Eng. Pap. CEP70-
Martin, J. C. 1960. Prod. Mon. 24 : 1 8-29 71HJM-AN43. Reprinted 1973 in
McGaw; R. 1972. Highw. Res. Board Proc. Ecological Studies, 4 : 29-42. Berlin :
Ann. Meet., 51st, 1-8 Springer
McLaughlin, B. D., de Bruyn, P. L. 1969. J. Morrow, N. R. 1970a. Ind. Eng. Chem. 62 :
Colloid Interface Sci. 30 : 21-33 32-56
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

McNabb, A., Grant, M. A., Robinson, J. L. Morrow, N. R. 1970b. Chem. Eng. Sci. 25 :
1975. Appl. Math. Div. DSIR Tech. Rep. 1 799-1815
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

No. 34. 38 pp. Morton, F., King, P. J., Atkinson, B. 1964.


Meurs, P. van 1957. Pet. Trans. Am. Inst. Trans. Inst. Chern. Eng. 42 : 149
Min. Metall. Pet. Eng. 210 : 295-301 Mualem, Y. 1973. Water Resour. Res. 9 :
Meurs, P. van, Poel, C. van der 1958. Pet. 1324-31
Trans. Am. Inst. Min. Metall. Pet. Eng. Mualem, Y. 1974. Water Resour. Res. 10:
213 : 103-12 514-20
Miller, R. D. 1965. Phase equilibria and soil Muskat, M. 1949. Physical Principles of Oil
freezing. Permafrost: Int. Conf Proc. Nat. Production. New York : McGraw-Hili
Acad. Sci.-Nat. Res. Council, 1963, Pub!. Nayfeh, A. H. 1972. Phys. Fluids 1 5 : 1 751-
No. 1287 193-97 54
Miller, R. D. 1973. Soil Freezing in Relation Neel, L. 1942. Cah. Phys. 12 : 1-20
to Pore Water Pressure and Temperature. Nielsen, D. R., Biggar, J. W. 1961. Soil Sci.
Presented at Permafrost 2nd Int. Conf., 92 : 192-93
Nat. Acad. Sci. Noblanc, A. 1970. A capillary perturbation
Miller, E. E., Miller, R. D. 1956. J. Appl. study of two-phase flow i'!filtration. MS
Phys. 27 : 324-32 thesis. Dept. Civil Eng., Colo. State
Miller, C. A., Ruckenstein, E. 1 974. J. Colloid Univ., Colo.
Interface Sci. 48 : 368-73 Noblanc, A., More1-Seytoux, H. J. 1972.
Moffatt, H. K. 1964. J. Fluid Mech. 1 8 : 1-18 Proc. Am. Soc. Civ. Eng. 98 : HY9, 1527-
Moore, R. E. 1939. Hilgardia 1 2 : 383-401 41
Morel-Seytoux, H. J. 1967. Dev. Mech. 4 : Noblanc, A., Morel-Seytoux, H. J. 1974.
1 321-35, Proc. Midwestern Conf, 10th, Proc. Am. Soc. Civ. Eng. lOO : HY4, 598-
Colo. State Univ. 608
Morel-Seytoux, H. J. 1969. Flow Through Orr, F. M., Scriven, L. E., Rivas, A. P. 1975.
Porous Media, ed. R. J. M. De Wiest, J. Fluid Mech. 67 : 723-42
Chap. 1 1, 455-516. New York : Academic. Padday, J. F. 1970. Spec. Discuss. Faraday
530 pp. Soc. No. 1 : 64-74
Morel-Seytoux, H. 1. 1972. Proc. Joint Symp. Parlange, J.-Y. 1971a. Soil Sci. 1 1 1 : 134-37
Fundamentals of Transport Phenomena in Parlange, J.-Y. 1971b. Soil Sci. 1 1 1 : 170-74
Porous Media, 2nd, Guelph, Ont., 2 57-65
. Peck, A. J. 1965a. Soil Sci. 99 : 327-34
Morel-Seytoux, H. J. 1 973a. Adv. Hydrosci. Peck, A. J. 1965b. Soil Sci. 100: 44-5 1
9 : 1 19-202 Peck, E. L. 1973. Soil Moisture-a Critical
Morel-Seytoux, H. J. 1973b. Cah. ORS1DM Factor in Snowmelt Runo.ff7 Presented at
Ser. Hydrol. 1 0 : 185-94 Ann. Meet. Am. Geophys. Union, 54th,
Morel-Seytoux, H. J. 1974a. Cah. ORS1DM Washington DC
Ser. Hydrol. 1 1 : 51-59 Penner, E. 1 972. Proc. Ann. Meet. Nat.
Morel-Seytoux, H. 1. 1 974b. Cah. ORS1DM Acad. Sci.-Nat. Res. Council., 51st, 5fr64
Ser. Hydrol. 1 1 : 181-88 Philip, J. R. 1957. Soil Sci. 83 : 345-57
Morel-Seytoux, H. J., Khanji, J. 1974a. Philip, J. R. 1964. J. Geophys. Res. 69 :
Water Resour. Res. 1 0 : 795-800 1553-62
More1-Seytoux, H. J., Khanji, J. 1974b. Philip, J. R. 1969. Adv. Hydrosci. 5 : 215-305
Equation of Infiltration with Compression Philip, J. R. 1970. Ann. Rev. Fluid Mech.
and Counterflow Effects. Colo. State Univ. 2 : 177-204
Civil Eng. Pap. CEP74-75 HJM-JKI4. Philip, J. R. 1972a. J. Appl. Math. Phys.
14 pp. (ZAMP) 23 : 353-72
MULTIPHASE FLUID FLOW THROUGH POROUS MEDIA 273

Philip, J. R. 1972b. J. Appl. Math. Phys. Slattery, I. C. 1974. AIChE J. 20 : 1 145-54


(ZAMP) 23 : 960-68 Smiles, D. E., Vachuad, G., Vauclin, M. 1971.
Philip, J. R. 1973a. Proc. Int. Congr. Theor. Soil Sci. Soc. Am. Proc. 3 5 : 534-39
Appl. Mech Moscow. 13th. ed. E. Becker,
. Smith, C. R. 1966. Mechanics of Secondary
G. K. Mikhailov, 279-94 Oil Recovery. New York : Reinhold. 504
Philip, J. R. 1973b. Soil Sci. 1 1 6 : 328-35 pp.
Philip, J. R. 1974. Geodenna 1 2 : 265-80 Swartzendruber, D. 1974. Soil Sci. 1 1 7 : 272-
Philip, J. R., Knight, J. H. 1974. Soil Sci. 81
1 1 7 : 1-13 Swartzendruber, D., Huberty, M. R. 1958.
Phuc, L. V., Morel-Seytoux, H. J. 1972. Soil Trans. Am. Geophys. Union 39 : 84-93
Sci. Soc. Am. Proc. 36: 237-41 Swartzendruber, D., Youngs, E. G. 1974.
Pirson, S. J. 1958. Oil Reservoir Engineering. Soil Sci. 1 1 7 : 165-67
New York : McGraw-Hill. 735 pp. 2nd ed. Szekely, J., Neumann, A. W., Chuang, Y. K.
1971. J. Colloid Interface Sci. 35 : 273-78
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org

Polubarinova-Kochina, P. Ya. 1969. Re


search Developments in Theory of Flow Talsma, T. 1970. Water Resour. Res. 6 : 964-
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

Through Porous Media in the USSR 70


(1917-1967). Moscow : Izdatelstvo Nauka. Taylor, G. I. 1960. Aeronautics and Astro
545 pp. (In Russian) nautics. ed. N. J. Hoff, W. G. Vincenti,
Poulovassilis, A. 1962. Soil Sci. 93 : 405-12 21-28. New York : Pergamon
Poulovassilis, A. 1969. J. Soil Sci. 20 : 52-56 Taylor, G. I. 1961. J. Fluid Mech. 10: 161-
Poulovassilis, A. 1970. Soil Sci. 109 : 5-12 65
Preisach, F. 1935. Z. Phys. 94 : 277-302 Terwilliger, P. L., Wilsey, L. E., Hall, H. N.,
Raats, P. A. C. 1973. Soil Sci. Soc. Am. Proc. Bridges, P. M., Morse, R. A. 1951. Pet.
37 : 681-84 Trans. Am. Inst. Min. Metall. Pet. Eng.
Raghaven, R., Marsden, S. S. 1971. J. Fluid 192 : 285-96
Mech. 48 : 143-59 Thorpe, S. A. 1969. J. Fluid Mech. 39 :
Richards, L. A. 1931. Physics 1 : 318-33 25-48
Richardson, J. G. 1961. Handbook of Fluid Topp, G. C. 1969. Soil Sci. Soc. Am. Proc.
Mechancis. ed. V. L. Streeter, Chap. 16, 3 3 : 645-51
1-112. New York : McGraw-Hill Topp, G. C. 1971a. Water Resour. Res. 7 :
Rose, W. 1961. Nature 1 91 : 242-43 914-20
Rubin, J. 1968. Soil Sci. Soc. Am. Proc. 32 : Topp, G. C. 1971b. Soil Sci. Soc. Am. Proc.
607-15 35 : 219-25
Ruckenstein, E. 1969. Chern. Eng. Sci. 24 : Topp, G. C., Klute, A., Peters, D. B. 1967.
1223-26 Soil Sci. Soc. Am. Proc. 31 : 312-14
Saffman, P. G. 1959. Q. J. Mech. Appl. Math. Topp, G. c., Miller, E. E. 1966. Soil Sci. Soc.
12 : 146-50 Am. Proc. 30 : 156-62
Saffman, P. G. 1963. Proc. Heat Transfer Turpin, J. L., Huntingdon, R. L. 1967.
Fluid Mech. Inst., ed. A. Roshko, B. AIChE J. 1 3 : 1196-1202
Sturtevant, D. R. Bartz, 176-81. Stanford, Vachaud, G., Thony, J.-L. 1971. Water
Calif: Stanford Univ. Press Resour. Res. 7 : 1 1 1-27
Saffman, P. G., Taylor, G. 1. 1958. Proc. R. Vachaud, G., Vauc1in, M., Wakil, M. 1972.
Soc. London Ser. A 245 : 312-29 Soil Sci. Soc. Am. Proc. 36 : 531-32
Saville, D. A. 1969. Phys. Fluids 12 : 2438-40 Van Dyke, M. 1964. Perturbation Methods
Scheidegger, A. E. 1960. The PhYSics of Flow in Fluid Mechanics. New York : Academic.
Through Porous Media. Toronto : Univ. 229 pp.
Toronto Press. 313 pp. Wallis, G. B. 1969. One Dimensional Two
Schlichting, H. 1960. Boundary Layer Phase Flow. New York : McGraw-Hili.
Theory. New York : McGraw-Hill. 647 pp. 408 pp.
4th ed. Washburn, E. W. 1921. Phys. Rev. 1 7 : 374-
Schwartz, A. M., Rader, C. A., Huey, E. 75
1964. Adv. Chern. Ser. 43 : 250-67 Weekman, V. M., Myers, J. B. 1964. AIChE
Schwartz, A. M., Tejada, S. B. 1972. J. J. 1 0 : 951-57
Colloid Interface Sci. 38 : 359-75 Welge, H. J. 1952. Pet. Trans. Am. Inst. Min.
Settari, A., Aziz, K. 1975. Int. J. Multiphase Metall. Pet. Ellg. 195 : 91-98
Flow 1 : 817-44 West, G. D. 1911. Proc. R. Soc. London A
Shafrin, E. G., Zisman, W. A. 1960. J. Phys. 86 : 20-25
Chern. 64 : 519-24 Whisler, F. D., Bouwer, H. 1970. J. Hydrol.
Sheldon, J. W., Zondek, B., Cardwell, W. T. 1 0 : 1-19
1959. Pet. Trans. Am. Inst. Min. Metall. Wooding, R. A. 1969a. Calif. Inst. Technol.
Pet. Eng. 216: 290-96 Tech. Memo. 69-5. 45 pp.
274 WOODING & MOREL-SEYTOUX

Wooding, R. A. 1969b. J. Fluid Mech. 39 : Yih, C. S. 1965. Dynamics ofNonhomogeneous


477-95 Fluids. New York : Macmillan. 306 pp.
Wooding, R. A. July 1975. Q. Appl. Math. Youngs, E. G. 1964. Soil Sci. 97 : 307-1 1
17 pp. Yuster, S. T. 1951. Proc. World Pet. Congr.
Wright, D. E. 1968. Proc. Am. Soc. Civ. Eng. The Hague. 3rd. Pt. 2. 437-45
94 : HY4, 851-72 Zisman, W. A. 1964. Adv. Chem. Ser. 43 : 1-5 1
Annu. Rev. Fluid Mech. 1976.8:233-274. Downloaded from www.annualreviews.org
by Universidad Autonoma Metropolitana on 07/30/11. For personal use only.

You might also like