Yanuka (1992) PDF

You might also like

You are on page 1of 18

Transport in Porous Media 7: 265-282, 1992.

265
9 1992 Kluwer Academic Publishers. Printed in the Netherlands'.

Percolation Theory Approach to Transport


Phenomena in Porous Media
M. Y A N U K A
Department of Physical Chemistry and The Fritz Haber Research Centre for Molecular Dynamics,
The Hebrew University of Jerusalem, 91904, Israel

(Received: 28 March 1990; in final form: 31 January 1991)

Abstract. The percolation theory approach to static and dynamic properties of the single- and two-phase
fluid flow in porous media is described. Using percoIation cluster scaling Iaws, one can obtain functional
relations between the saturation fraction of a given phase and the capillary pressure, the relative
permeability, and the dispersion coefficient, in drainage and imbibition processes. In addition, the scale
dependency of the transport coefficientis shown to be an outcome of the fractal nature of pore space and
of the random flow pattern of the fluids or contaminant.

Key words. Percolation processes, pore space topology and geometry, capillary displacement, phase
entrapment, relativepermeability,dispersioncoefficient,hydrodynamicdispersion, transport coefficients-
scaling laws of, anomalous diffusion, random flow in network model, contaminant spreading, drainage,
imbibition, fractal dimension.

1. Introduction
Transport phenomena in porous media are of great interest and a challenge to
theoreticians and to hydrology, petroleum and ecology engineers. In m a n y cases we
are interested to know the magnitude of spreading of water, spills of oil or
contaminants in the top soil or in deeper porous and fractured geological material,
as a function of time and distance. The estimation of such transport properties is
extremely difficult, since the structures of porous media are not regular and in m a n y
cases are random. The flow pattern of the fluid is then governed by the structure of
the alia and by hydrodynamics, so that the solution of the differential equations
inve.,et., together with the appropriate boundary conditions, becomes a very
complicated task and in m a n y cases is an impossible one. One possible way for
solving single- or multiphase flow regime problems is by employing the equations of
continuity and regarding the media as continuous and statistically representable by
a REV (representative elementary volume) [1].
Miscible contaminant displacement and hydrodynamic dispersion [1] are exten-
sions to the single-phase flow problem where convective and diffusive transport
regimes are considered. However, we can look also at cases of two-phase flow
problems where miscible displacement occurs in each phase, or partial miscibility of
one phase by the other is possible [2].
266 M. YANUKA
The mass conservation in a continuous media approach is phenomenological in
nature, in the sense that in order to satisfy the continuity equation we need to
determine transport coefficients, such as the permeability, k, for a single-phase flow,
and the relative permeability, kr, for a multiphase flow. When dealing with
hydrodynamic dispersion, we need to know the dispersion coefficient, De, which is
a combination of the convective and diffusive transport coefficients. Recently it has
been argued that the transport coefficients are scale dependent [3-10]. This means
that the magnitude of the transport coefficient depends upon the measuring unit we
use. Thus, an assumption regarding the size of a statistically uniform elementary
volume will determine the magnitude of the transport coefficient. The reason for the
scale dependency of the transport coefficients is in general strongly related to the
fractal nature of pore space. The porosity scales as 9 ~ (~/L) D- Dj, where ~ is some
characteristic length scale, L is the length of the sample and D and Df are the
Euclidean and fractal dimensions [11]. Furthermore, any random walk or a path in
the pore space will result in tortuous scale dependent length Lf = ~1-DsLDs [4]. The
transport coefficients can also be shown to be scale dependent using percolation
concepts [3, 5, 12-15]. The difficulties in determining the transport coefficient, in
the classical continuum transport approach, are related to the determination of the
REV or, in other words, the question asked is whether such an element can be
determined together with the appropriate coefficients, [16, 17]. According to perco-
lation theory [13, 18], the answer to this question is that as long as our medium is
random and not very far from some critical value (critical porosity or critical
saturation), the transport coefficients are scale dependent.
In this review, we describe the percolation approach to transport phenomena in
porous media where the statics and dynamics of single- and two-phase system
properties will be considered. Here, the two phases can be solid phase and pore
(fluid) phase, or two fluids coexisting phases.

2. The Percolation Process Analogy


We regard a porous medium as a network of pores (sites), which are connected by
capillary tubes (bonds) (Figure 1). Percolation theory deals with networks of sites
and bonds where the topology of the system is determined by a number of
connections, Z, that each site has with its neighboring sites.
In percolation processes, we determine a fraction of open sites pS and open bonds
pb which can be regarded in analogy to porous media in the two following cases: (1)
~p = (pS + pb) is the pore space fraction which is capable of conducting (open) fluid
and q~s = [(1 - p ' ) + (1 _pb) is the fraction of solid phase. This is a single-phase
flow problem. (2) ~nw = (P~ +pb) is the fraction which is capable of conducting
(open) one phase and ~w = [( 1 - pS) + 1 - pb)] is the fraction capable of conduct-
ing (open) a second phase. This is a two-phase flow problem, where ~,w + ~w = ~p-
Percolation theory provides us with the tools to investigate the static saturation
capillary pressure relation [19-23] in two-phase problems, and also the fluid
PERCOLATION THEORY APPROACH 267

blocked site / p e n site

Ji,l

open bond
t blocked bond
Fig. 1. A square network showing sites and bonds in open and blocked clusters.

dynamic properties, e.g., relative conductivity kr [5, 14, 24] and the dispersion
coefficient DL [3, 12, 15, 25].
Here we would like to present the conceptual aspects needed for the application
of percolation theory to the dynamics and statics of single or two phases in porous
media.

3. The Static Properties of Two Coexisting Phases in Pore Space


As was mentioned before, we consider a network of conducting sites, which are
junctions, and bonds, which are the connections between the sites. Fraction pS and
pb of these are respectively open. We can look at the case where only the sites are
considered (site problem, where pb = 1), or only the bonds are considered (bond
problem, where p~ = 1), or both (mixed bond-site problem, [22, 23]).
When we look at flow under a capillary pressure regime, in the bond problem the
bonds restrict flow (Figure 2a), in the site problem the sites restrict flow (Figure
2b), and in the mixed problem both the sites and the bonds restrict flow (Figure 2c).
Since pS or pb are the fraction of elements open for transport, we have, according
to percolation theory, critical fractions p~ and p~ below which flow of a given phase
will not exist. This phase will be trapped (Figure 3). We can see in Figure 3 that the
268 M. YANUKA
bond problem

controling bond

(c)

site problem

controling site

(b)

mixed bond-site problem

controling bond

~ controling site
(a)
Fig. 2. A bond problem (2a), site problem (2b) and mixed problem (2c) analogy to capillary
displacement.
PERCOLATION THEORY APPROACH 269
Square lattice above percolation threshold
p = .75

Infinit~e cluster

Square lattice under percolation threshold


p = .25

finite clusters

Fig. 3. Infinite and finite clusters below and above the critical percolation point. Note that the black
and white regions are complementary.

trapped phase forms small and finite clusters, while the trapping phase forms an
infinite continuous cluster. Here, it can be emphasised that within this trapping
mechanism, one fluid can trap another fluid, or the solid phase can trap pockets of
voids which are not connected to each other, such as in impermeable rocks.
The probability that an open site (or bond) belongs to the infinite cluster of a
given phase is the percolation probability

P(p~) ~ (p~ - p~)~, (la)

P(p') ~ (p~ -p~)~, (lb)

where fi ( = 0.44 for D = 3) is a universal critical exponent which is not dependent


on the geometry and the topology of the network [18].
270 M. YANUKA
The mixed percolation probability is related to the bond (la) and site (lb)
percolation probabilities [26] as
p(1,pbpS) < p(pb, p~,) < p(pbp~, 1). (lc)
Thus, it can be seen that the mixed critical percolation probability, (Pcm,
b P scm), is
bound between the bond and site critical percolation probabilities as
< PcmPc,~, < P~." (ld)
We can summarize that P(p) is the fraction of pores belonging to an infinite and
continuous chain of pores, and flow of a phase occupying these pores occurs. In this
sense, P(p) is proportional to the saturation fraction of an invading phase.
The average size (number of sites) of a finite (trapped) cluster of a given phase
can be described as follows:
( s ) - (p - p c ) -~, (2)
where 7 ( = 1.8 for D = 3) is a universal critical exponent and ( S ) diverges near Pc,
i.e., the phase is continuous near Pc. However, when p >Pc, there remain trapped
clusters of this phase (Figure 3). It should be noted that in three dimensions two
phases can form simultaneously continuous infinite clusters where for both phases
trapped finite clusters can be found.
The characteristic length (rms of pairwise distances between sites) of the finite
clusters varies as
~ (p - p A v, (3)
where v ( = 0 . 8 8 for D = 3) is a universal critical exponent. Here, we also see that
diverges near p~. Cluster scaling [ 13, 18] relates the size of the cluster to the linear
length of measurement (scale) as
S ~ LD~; (4)
where S is the number of sites counted in a box of linear length L and Ds is the
fractal dimension of the cluster. Self similarity implies that Ds is a constant property
of the cluster, independent of its size.
Other cluster properties can be scaled accordingly using the fractal dimension Df
and the Euclidean dimension D:
( S ) ~ ~2~: ~, (5)
p(p) ~ ~D: o. (6)
It follows also that B/v = D -- Df and 7/v = 2 D f - D, where near Pc, Df = 1.9, 2.5
for D = 2, 3. The above relations can give us the appropriate tools for describing
the static capillary pressure saturation curves. If we define

p" = f~,(R), (7a)

pb = f,o(R), (7b)
PERCOLATION THEORY APPROACH 271

where f ~ ( R ) and fa(R) are the size distributions of the sites and the bond, and the
capillary pressure ~ is related to R by

= ~ cos(0)/R, (8)

where as is the surface tension and 0 is the contact angle.


The saturation curve as a function of the capillary pressure for the invading
nonwetting phase (S~) during the drainage process can be written using (lc):

Sn W ~ (p,~pb __ Pdc),a. (9a)


The saturation curve during the imbibition process is
A S , w ~ (pSpb _ p~c)~ + ~, (9b)

where Sw + S,,w = 1, ASnw = Snw - S,,wc, and S~w~~ p~.


Physically, P J ~ ( P ~b, P ~ ms ) is a determined property, the fraction of pores of size
R > R~, corresponding to the critical capillary pressure ~ . In this sense, the
drainage process, the invasion of nonwetting phase, can be described accurately by
percolation processes. The reverse process imbibition, or the invasion of the wetting

I f I r I r

x x .

I I I [ I I

Fig. 4. Demonstration of invading phase saturation, S,w (black circles), inside a withdrawing phase.
Note that the withdrawing phase saturation, Sw, is the entire region of the white circles and pulses.
272 M. YANUKA

I I I I

~9
I,i

primary drainage
m
m
.m

secondary i m b i b i t i o n - " " ~

Snwe
i i ~ ~/

0 0.2 0.4 0.6 0.8 1


Sw
Fig. 5. Capillary hysteresis loop calculated from (9) and Figure 6.

I I I I I I I

sites

~9

~9
&

~ f
I I i ~ I I I

R
Fig. 6. Site and bond size distributions used in the calculation of Figure 5.
PERCOLATION THEORY APPROACH 273

phase, does not behave in the same critical manner. The wetting phase can flow
on surface corners etc., so that there is no physical meaning to Pc. Instead, during
imbibition we look at the withdrawal of the nonwetting phase; this gives a
different e x p o n e n t / / + l in Equation (gb) {27]. P~c is then the fraction of sites and
bonds corresponding to the pressure at which residual nonwetting phase satura-
tion occurs. It is worth noting that the residual wetting saturation has less
physical meaning than the nonwetting residual saturation for the reason just
given.
Figure 4 demonstrates the basic difference between the fractions S,,w and p,
where a penetrating fluid ( S ~ = black circles) displaces another fluid (Sw = open
circles and pulses). The fraction of open pores (p = open circles and black circles)
are the pores of size R > R*. In Figure 4, we can see that at this stage there are
trapped clusters of both phases.
Figure 5 gives an example of the hysteresis loop, describing imbibition and
drainage processes including the residual S,w~ nonwetting phase fraction, i.e., the
trapped clusters fraction. The calculation was done, using lognormal distributions
(Figure 6) and Equations (7 9).

4. The Dynamic Properties


The dynamic properties of a single- or two-phase flow can also be addressed with
the use of percolation concepts and scaling laws [5, 14, 24]. We look again at a
network of conducting elements, such that the element between junctions i and j
has a randomly distributed conductivity go and a pressure 6~'ij is applied on the
element. With Kirchhoff's law, we can determine the flux in each element by
solving simultaneously the set of equation gij 6~b~j= 0 (Figure 7). The resultant
permeability of the system can then be scaled (using Equation (3)) as

kw ~ (p - p c ) ~ ~ ~/'~. (10a)

The relation of the relative permeability to the saturation fraction during the
drainage process can be expressed as a combination of Equations (9a) and (10a)
[27]:

kn w ~ s~/~, k~ ~ ( 1 - snw) ~"~, ( l 0b)

and during the imbibition process we have

k,,w ~ AS~/w(~ + '), kw ~ ( 1 - AS, w)~'/(/~+ ~), (10c)

where # ( = 2.0 for D = 3) is a critical transport exponent.


It follows that the relative permeability is scale dependent (10a), and has an
hysteresis response (10b) and (10c). Figure 8 gives the calculated relative perme-
abilities as a function of saturation.
274 M. YANUKA

rpbc

/ "r b c l

r a n d o m walk starts here


Fig. 7. A network analogue for calculating the permeability of pore space. Each bond between i and
j sites in the network is assigned a random conductivity gii and pressure boundary conditions are
imposed as shown by ~bpbc (periodic boundary conditions) and ~bbo1> ~bbc2. The solution of the network
gives the pressures at each site ~bi, and the flux in each bond Vo.. In the dispersion phenomena
simulation, we let a particle move in a random walk pattern as shown. This walk is governed by the local
fluxes ill the network.

5. Contaminant Transport and Miscible Displacement


T h e h y d r o d y n a m i c d i s p e r s i o n [1] is related to the s p r e a d i n g o f miscible c o m p o n e n t s
in a given phase. This s p r e a d i n g is d e p e n d e n t on the local i n h o m o g e n e i t y o f flow
a n d is u s u a l l y described b y the c o n v e c t i o n diffusion e q u a t i o n

OC Dr ~2C
c3~- = ~x2 + DrVZC- VC, (11)

where DL a n d D r are the l o n g i t u d i n a l a n d transverse d i s p e r s i o n coefficient, which


describe the rate o f c h a n g e o f the c o n t a m i n a n t d i s p l a c e m e n t D/. = crz/2t, where
a 2 = ( x 2) - ( x ) 2. DL is a s s u m e d to be p r o p o r t i o n a l to the average velocity V o f
the fluid c a r r y i n g the c o n t a m i n a n t (miscible c o m p o n e n t s ) . T h e s o l u t i o n o f the
c o n v e c t i o n diffusion e q u a t i o n (11) is possible, p r o v i d i n g t h a t D c is k n o w n a n d
PERCOLATION THEORY APPROACH 275

1 I I I _ _

0.8
I,i
k~

i~ 0 6 -

0.4

0 ~ i I [~"

0 0.2 0.4 0.6 0.8 1


Sw
Fig. 8. Relative permeability of the wetting and nonwetting phases during the primary drainage and
secondary imbibition calculated using (10).

constant. Recent studies indicate that the dispersion coefficient is scale dependent
[3-10], and that the application of (11) is questionable [16, 17].
Random walk and particle tracking techniques have been used extensively for
simulation of the dispersion of a tracer [3, 12, 25, 28, 29]. At the same time, very
extensive work has been done to relate percolation cluster scaling concepts to the
dispersion coefficient and to obtain a possible scale dependent solution of (11)
[3, 12, 15, 25]. Here, we will present results from random walk simulations of
dispersion in a two-dimensional square lattice.
First, we consider a Fickian diffusion process, where a particle can move in any
desired random direction with no limitations. A walk starts at a fixed point for N
steps and M realizations. A result of such a simulation is given in Figure 9.
We expect for this process that

d ( x 25
(X 2) ~ t and D L =- ~ constant.
dt

If, however, we allow a fraction p" of the sites in the lattice to be open, i.e. to
conduct the contaminants, then every time the walk approaches a site that is not
open it 'waits' and a new attempt is made in a different direction. Figure 10 gives
a spreading pattern of such a case when pS is close to p~. Here, the diffusion process
276 M. YANUKA

/
Fig. 9. A random walk in a square lattice simulation of a Fickian diffusion pS= 1, i.e., all sites are
open. Numbers on contour lines are the number of times each site was visited, on a logarithmic scale.
There were N = 1 0 4 time steps and M = 1 0 4 realizations. Note that the source point was not located in
a symmetric location, however the entire network is uniformly accessed.

behaves a n o m a l o u s l y a n d one finds that ( x 2) ~ t 2/(2 +0), with 0 = (# - f l ) / v [30]. I n


this case we can see that

d(x2) t -0/(2+~ DL ~ (x) -~


D L =- dt

a n d also for ( x ) < ~, we expect that DL ~ ~ 0


T o conclude, we see that if the spreading of c o n t a m i n a n t is diffusion d o m i n a t e d ,
then n e a r Pc the diffusion coefficient vanishes with distance or, in other words, it is
scale d e p e n d e n t . This can be rewritten using (3) as follows:

DL~(P--pc)~ ~ -~ (12)

Figure 11 gives a c o m p a r i s o n of calculated DL((X)) using the m e t h o d described


earlier ( F i g u r e s 9 a n d 10). The results show that, as expected, for F i c k i a n diffusion
PERCOLATION THEORY APPROACH 277
I I I I I I I I I I I I I I I I I I I I I I I I I

Fig. 10. A random walk in a square Iattice simulation of an anomalous diffusion pS = 0.6. Numbers on
contour lines are the number of times each site was visited, on a logarithmic scale. N = 104 time steps
and M = 1 0 4 realizations.

Dr ~ c o n s t a n t , b u t in the a n o m a l o u s case we o b t a i n D L ~ < x ) (-~ (this is close


to the e x p o n e n t in (12), since in 2 D # = 1.3,/~ = 0.14, v = 1.33 a n d 0 = 0.87 [18]).
Secondly, we w o u l d like to l o o k at the s i t u a t i o n where, in a d d i t i o n to the
diffusion flux, we have a convective flux which is the d o m i n a t i n g one. H e r e we
e m p l o y the following m e t h o d . Starting with a square lattice, we d e t e r m i n e ran-
d o m l y the conductivities g~ o f each b o n d in the lattice ( F i g u r e 7) a n d a p p l y pressure
b o u n d a r y c o n d i t i o n s ~bc. T h e n we solve the linear system gij 6~ = 0, as before. T h e
result is a deterministic pressure discharge a n d fluxes in each site a n d b o n d ,
respectively. I n i t i a t e d at a fixed p o i n t ( F i g u r e 7), a r a n d o m w a l k continues,
weighted by the discharge o f each b o n d . T h e w a l k stops at the b o u n d a r i e s o f the
lattice. M realizations are taken. H e r e also, we e x a m i n e a s i t u a t i o n where the
fraction o f o p e n c o n d u c t i n g b o n d s pb = 1 a n d where p is close to p~. F i g u r e s 12 a n d
13 s h o w an e x a m p l e o f the s p r e a d i n g p a t t e r n in these cases. Scaling o f the
278 M. YANUKA

p = 1.0
"~0.1
0 o

0.01 1
1 10 100
log(<x>)
Fig. 11. S i m u l a t i o n results in a s q u a r e lattice of D L = ~ 2 / 2 t for diffusion (p~ = 1.0) a n d a n o m a l o u s
diffusion ( p ~ = 0.6). N = 10 3 time steps a n d M = l 0 4 realizations.

dispersion coefficient can be done here also. By dimensional considerations we have


~x2)/V2 T2 and DL ~ z V2, where V is the averaged macroscopic velocity and z is
the maximum residence time it takes a particle to stay in dead-end regions (pockets
or pores). From our previous analysis (10) we know that the macroscopic perme-
ability scales as k ~ ~-*v. The macroscopic average velocity is proportional to an
effective permeability, V ..~ k/Snw, where Snw is the nonwetting phase saturation
fraction which scales (using (3) and (9)) as S~w ~ ~-~/~. The residence time scales
as ~ ~ ~2+0. From these relations we can see that [12, 15]

DL~zV 2~2+~176 ~2 0. (13)

According to the above mentioned analysis we can see that when the convection
term is taken into consideration, the dispersion coefficient diverges with length
scale. Furthermore, we can rewrite the above expression as:

DL~(p_pc), ~ 2~, (14)

or, in other words, the dispersion coefficient diverges as p ~Pc. These results can be
seen when comparing Figures 12 and 13, which correspond to the cases of
dispersion in a full square lattice connected network ( p b = 1), and partially con-
nected square network (pb= 0.5). Also in Figure 14 is given a comparison of
PERCOLATION THEORY APPROACH 279

TU=
i
8

I11111111111111111111111111111111111|IIIIIIIIIII

Fig. 12. A r a n d o m walk in a square lattice simulation of convective dispersion at p6 = l. N u m b e r s on


c o n t o u r lines are the n u m b e r of times each site was visited, on a logarithmic scale. Lattice size 51,51,
M = 104 realizations.

DL((x), where we find that when pb = 1, DL ~ constant and when pb is near p~, we
obtain DL ~ ( x ) L2 (which is what we expect from (13)).
The relation of the dispersion coefficient and the saturation fraction is a result of
(9) and (14) and gives in the drainage process

DL ~ S}~. # - 2~)/#, (15a)

and in the imbibition process

Dc ~ AS(~- # 2t,)/(/~+ i) (15b)

In these relations the dispersion coefficient diverges as saturation approaches the


residual saturation.
280 M. Y A N U K A

Fig. 13. A random walk in a square lattice simulation of convective dispersion at pb = 0.5. Numbers
on contour lines are the number of times each site was visited, on a logarithmic scale. Lattice size 35*35,
M : 104 realizations.

6. Conclusions
In this work we have presented the percolation approach to transport phenomena
in porous media. It has been shown that the scaling laws of percolation clusters as
well as of the fractal nature of pore space may lead to the understanding of:
(1) The relation between saturation and capillary pressure, the phase entrap-
ment and its magnitude, understanding of the hysteresis effect during the
draining and imbibition process from the point of view of percolation
processes.
(2) The scaling aspects of the permeability and the hysteresis in the relation
connecting relative permeability and saturation fraction during drainage and
imbibition.
PERCOLATION THEORY APPROACH 281

0.4 = . -

0.3

0.2

0.1

0 I I [ I I I I [ I---'-

0 10 20 30 40 50
<X>
Fig. 14. Simulation results of D r = a2/2t for convective dispersion (pb= 1.0) and super-convective
dispersion ( y ' = 0.5). M = 104 realizations.

(3) The scaling aspects of the dispersion coefficient in the diffusion dominated
spreading and the convection-diffusion dominated spreading.
(4) The relation between the saturation fraction and the dispersion coefficient
and the hysteresis effect.

Finally, a great deal of theoretical and simulation work has been done on these
issues. The results are encouraging, especially on the ground of continued evidence
from field measurements on the validity of these frameworks [7-10]. However,
extensive and thorough experimental work must be done, which will then test the
theoretical results. A new scope of possibilities is also opened in which a statistical
assessment of contamination in porous and fractured media can be estimated, based
on small field experiments.

References

1. Bear, J., 1972, Dynamic of Fluids in Porous Media, Elsevier Science, New York.
2. Corapcioglu, M. Y. and Baehr, A. L., 1987, A compositional multiphase model for groundwater
contamination by petroleum products 1. Theoretical considerations, Water Resour. Rea. 23, 191
200.
3. Sahimi, M. and Imdakm, A. O., 1988, The effect of morphological disorder on hydrodynamic
dispersion in flow through porous media, ]. Phys. A: Math. Gen. 21, 3833-3870.
4. Wheatcraft, C. W. and Tyler, S. W., 1988, An explanation of scale-dependent dispersivity in
heterogeneous aquifers using concepts of fractal geometry, Water Resour. Res. 24, 566-578.
282 M. YANUKA

5. Thompson, A. H., Katz, A. J., and Krohn, C. E., 1987, The microgeometry and transport properties
of sedimentary rock, Adv. in Physics 36, 625-694.
6. Ross, B., 1986, Dispersion in fractal fracture networks, Water Resour. Res. 22, 823 827.
7. Seiler, K. P., 1985, Results of field experiments of hydrodynamic dispersion in quaternary gravels of
southern Germany, in Scientific basis for water resources management (Proc. of the Jerusalem
Symposium, September 1985), IAHS Publ. no. 153, pp. 351 360.
8. Sudicky, E. A., Cherry, J. A., and Frind, E. O., 1983, Migration of contaminants in groundwater at
a landfill: A case study, J. Hydrol. 63, 81 108.
9. Molz, F., Guven, O., and Melville, J. G., 1983, An examination of scale-dependent dispersion
coefficients, Ground Water 21, 715-725.
10. Pickens, J. F. and Grisak, G. E., 1981, Scale-dependent dispersion in a stratified granular aquifer,
Water Resour. Res. 17, 1191-1211.
11. Baudet, C. et al., 1985, Scaling concepts in porous media, in Scaling phenomena in disordered
systems (proceedings of the NATO conference, April 20-21, Geilo, Norway), pp. 399 422.
12. Sahimi, M., 1987, Hydrodynamic dispersion near the percolation threshold: scaling and probability
densities, J. Phys. A: Math. Gen. 20, L1293-L1298.
13. Aharony, A., 1986, Percolation, in G. Grinstein and G. Mazenko (eds) Directions in Condensed
Matter Physics, World Scientific, pp. I 50.
I4. Feng, S., Halperin, B. I., and Sen, P. N., 1987, Transport properties of continuum systems near
percolation threshold, Phys. Rev. B. 35, 197-214.
15. De Gennes, P. G., 1983, Hydrodynamic dispersion in unsaturated porous media, J. Fluid Mech. 136,
189-200.
I6. Matheron, G. and de Marsily, G., 1980, Is transport in porous media always diffusive? A counter
example, Water Resour. Res. 16, 901-917.
17. Berkowitz, B. and Bachmat, Y., 1988, A scale-dependent equation for solute transport in porous
media, Transport in Porous Media 3, 199-205.
18. Stauffer, D., 1985, Introduction to Percolation Theory, Taylor & Francis, London.
19. Yanuka, M., Dullien, F. A: L., and Elrick, D. E., 1986, Percolation processes and porous media I.
A geometrical and topological model of porous media using three-dimensional joint pore-size
distribution, J. Coll. I. Sci. 112, 24 41.
20. Yanuka, M., 1989, Percolation processes and porous media IL Computer calculations of percolation
probabilities and cluster formation, J. Coll. L Sci. 127, 35-47.
21. Yanuka, M., 1989, Percolation processes and porous media III. Prediction of capillary hysteresis
loop from geometrical and topological information of pore space, J. Coll. L Sci. 127, 48 57.
22. Yanuka, M., I990, The mixed bond-site percolation problem and its application to capillary
phenomena in porous media, J. Coll. L Sci. 134, 198 205.
23. Yanuka, M. and Englman, R., 1990, Bond-site percolation: analytical representation of critical
probabilities, J. Phys. A: Math. Gen. 23, L339-345.
24. Kirkpatrick, S., 1983, Percolation and conduction, Rev. Mod. Phys. 45, 574-588.
25. Koplik, J., Redner, S., and Wilkinson, D., 1988, Transport and dispersion in random networks with
percolation disorder, Phys. Rev. A. 37, 2619-2636.
26. Hammersley, J. M., 1980, A generalization of McDiarmid's theorem of mixed Bernoulli percolation,
Math. Proc. Camb. Phil. Soc. 88, 167-170.
27. Wilkinson, D., 1986, Percolation effects in immiscible displacement, Phys. Rev. 34, 1380-I386.
28. Kinzelbach, W. and Uffink, G., 1989, The random walk method and extensions in groundwater
modelling, in Nato Advanced Studies Institute On Transport Process in Porous Media, July 9-18,
Pullman, Washington, U.S.A.
29. Schwartz, W. and Smith, L., 1988, A continuum approach for modeling mass transport in fractured
media, Water Resour. Res. 24, 1360-1372.
30. Gefen, Y., Aharony, A., and Alexander, S., I983, Anomalous diffusion on percolating clusters, Phys.
Rev. Let. 50, 77-79.

You might also like