SPE 56417

Effect of a Wetting Immobile Phase on Diffusion and Macroscopic Dispersion in

Unconsolidated Porous Media
M.L. Verlaan, SPE; H.F. Dijkgraaf, SPE; and C.P.J.W. van Kruijsdijk, SPE, Delft University of Technology

Copyright 1999, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at the 1999 SPE Annual Technical Conference and
Exhibition held in Houston, Texas, 36 October 1999.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgement of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
The objective of this paper is to determine the effect of a
wetting immobile phase (e.g. connate water) on the diffusion
and macroscopic dispersion in homogeneous unconsolidated
packs. We measured the effective dispersion coefficients of
gases as a function of water saturation, pressure, and velocity
in a laboratory set-up. The dispersion coefficients were 2 to 3
times higher than in the dry porous medium. This effect is also
studied using a pore network model where the immobile phase
blocks a fraction of the pores for fluid flow. A random walk
through the network and counting cluster sizes both gave
approximately the same results, which were in good
agreement with the experimental data.
Introduction
Macroscopic dispersion is a core level mechanism that is
important in miscible displacement processes. It is especially
needed for the proper upscaling of heterogeneous porous
media1, and it controls the mixing in cyclic processes (e.g.
underground gas storage, gas cycling), where dispersion by
large-scale permeability heterogeneity is reversible. Generally
the convectiondiffusion equation is used to model
macroscopic dispersion. The effective diffusion coefficient
(called dispersion coefficient) is experimentally determined,
and can be represented by the empirical relation of Perkins and
Johnston2. These dispersion values are usually measured in
dry sand packs or dry cores, which does not represent the in
situ situation where usually also a second phase is present.
Only a hand full investigated the effect of an immobile
phase, but there is no agreement between the results of their
investigations. Some authors3,4 state that dispersion increases

when an immobile phase is present, but Batycky5 concludes

that it has no effect. Orlob6 even found that dispersion
decreases, but in this case the immobile phase was nonwetting. This raises some fundamental questions as how does
the immobile phase affects macroscopic dispersion, what are
the important parameters involved and how can we quantify it.
The objective of this paper is to understand the mechanism
by which the immobile phase affects macroscopic dispersion.
This paper will limit the questions to the wetting liquid phase
because this is the most encountered case in the field. A
number of laboratory experiments are presented with water as
the immobile phase, gas as the flowing phase and sand or glass
beads as unconsolidated porous medium. The results are
interpreted in terms of dispersion coefficients and
dispersivities. Characterisation of the porous media and
distribution of the immobile phase are used in different
network models and the predicted results are compared to the
laboratory experiments.
Dispersivity
A big problem concerning the comparison of data is the
definition of the dispersion coefficient. In most articles the
following relation is used:

Dl =

Dm
+ U ,.......................................................(1)

where Dl is the longitudinal dispersion coefficient, Dm the

molecular diffusion coefficient, the tortuosity of the porous
medium, the dispersivity, U the interstitial velocity and the
velocity dependence of the dispersion coefficient. It is clear
that using different values for leads to non-comparable
values of . In this paper we assume that equals 1 and
could therefore be velocity dependent.
Using equation (1) we can distinguish three different
regimes. When the first term on the right hand side is larger
than the second, we speak of the diffusion controlled regime.
When the second term prevails, we normally speak of the
convection controlled regime. In between there is a regime
where both diffusion and convection are important.

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
3

calculated values for the tortuosity are shown in Table 2. The

tortuosity in dry unconsolidated porous media is normally
around 1.42 as in our experiments. The higher values in the 9.8
and 15.5% water saturation experiments demonstrate that the
water phase blocks a fraction of the pores, and makes the flow
paths more tortuous. Although the tortuosity becomes not very
high, it is still a significant effect.
Dispersivity
The dispersivities can be obtained over the whole velocity
range from equation (1), but only in the convection dominated
regime accurate values can be obtained. Many authors assume
that the dispersivity is velocity independent and fit a straight
line through the Dl vs. U plot. That this is not always correct
for these experiments is shown in Figure 6, where the
dispersivities for the glass bead experiments are plotted
against the interstitial velocity. This plot shows that the slope
can vary at lower velocities and becomes eventually straight at
higher velocities ( = 1). The straight-line value is taken as the
true dispersivity. Edwards8 observed the same trend in his
experiments, and he gives radial diffusion in the pores as an
explanation.
The values for the dispersivities are shown in Table 3. The
values show more or less the same trend in all cases. Except
the 50 bar series with 14.4% water, where the dispersivity is
much larger. Although the value has a large uncertainty, it is
still significantly larger than the other experiments. In the next
sections an explanation for these results is given.
Porous Medium Characterisation
To find an explanation for the previous experimental results
we need more insight how the immobile water phase is
distributed in the porous medium, and how it effect the flow
through the porous medium. This knowledge is then
incorporated in a network model to simulate the flow through
a porous medium with an immobile water phase. The input
needed for this simulations are the connate water distribution
and the pore size distribution
Connate water distribution
Water, as the wetting phase, can be present in the porous
media as a thin film coating the grains, as a mantle around the
grain contacts or fill whole pores. The thickness of the coating
is normally in the order of nanometers and will not affect the
flow, and the amount of water stored in this way is very small.
The mantles can contain a larger amount of liquid but will
influence the flow only slightly, by decreasing the pore size.
Pores filled with water will be the main storage mechanism
and will have the largest effect on the flow.
To verify this assumption, we simulated the immobile
phase in a sand pack with coloured epoxy resin. The pack was
saturated with the resin and after a while the resin was
displaced with air under gravity drainage. When the epoxy had
become hard, the rest of the open pore space was filled with
transparent epoxy resin. A thin section of this sand pack with
immobile resin is shown in Figure 7. It is clear that most of the

immobile fluid fills the pores and not only the smallest ones.
To get a relation between the water saturation and the fraction
of blocked pores we used a Bethe Lattice network to model
the permeability reduction of a certain pore size distribution
by a random fraction of blocked pores.
Pore size distributions
The pore size distributions were also obtained from thin
sections of the porous media by quantitative image analysis.
The used method is the erosion-dilation technique described
by Yuan9 , and resulted in pore size distributions as shown in
Figure 8. The pore size distribution is narrower for the glass
bead pack which can be expected with these perfectly round
particles. The less spherical sand particles show a wider pore
size distribution. The very small pores (<100 m) are fully
absent in this distribution, because they are nor recognised as
individual pores during the image analysis process.
Network model
Our experimental results have indicated that in the presence of
an inert and immobile phase (longitudinal) dispersion may
increase significantly compared to the single phase results (see
Fig. 3 to 5). The water phase, which we used as the immobile
phase, wets the porous medium. In this section we attempt to
interpret these observations based on percolation concepts. We
assume the porous medium to consist of a network of pore
bodies interconnected by pore throats. We will limit the
discussion to laminar flow and assume all viscous dissipation
to take place in the pore throats. Although most natural porous
media form highly irregular networks with co-ordination
numbers varying from 3 to 15, we chose to use regular
networks based on a constant co-ordination number. This
choice of convenience is justified by the results of Jerauld et
al.10,11 who showed that effective transport properties of
irregular networks are virtually identical to those of regular
networks as long as the average co-ordination numbers are
equal. For a comprehensive review of percolation and other
network models see Sahimi.12,13 It is generally accepted that
the wetting phase will occupy the smaller throats, thereby
blocking them for flow of the other phase (see Fig.7). The
change in (effective) pore morphology as experienced by the
mobile phase may affect all transport properties.
Unfortunately, it is nearly impossible to measure the fraction
of blocked pores directly. The only data we have are the
relative permeability values at the various (immobile) water
saturations and pore size distribution. In a previous paper14 we
showed how to estimate the fraction of blocked pores from
these data. Some details are given in appendix A. The fraction
of closed pores for the different porous media and water
saturations are given in Table 4.
Results Network Model
Many papers2,3 attempt to capture convective dispersion in
terms of the relationship given in equation (1), where a socalled dispersivity, , is introduced. In a homogenous porous
medium the dispersivity is related to the effective particle

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
5

dispersion in a porous medium. First we established the

quantitative effect through laboratory experiments. This was
very cumbersome, because some 350 experiments had to be
carried out and the results were only six dispersivity values.
The error in this value is also quite large as can be seen in
Figure 12, because the dispersivity is not a constant, but
depends on the diffusion and flow velocity. The general trend
is very clear: with increasing water saturation, the dispersion
coefficient increases, in our case with a factor two to three.
The high water saturation experiments at 50 bar looks
awkward in the measured ensemble because of the high value,
and very high error margin. Still it is an ensemble value of 15
experiments, but we can give no reason for the high value.
The network models were derived from the fact that the
immobile water blocks many pores. There are two ways to
look at it: first the effective particle diameter is enlarged by
clustering of adjacent particles and second the blocked pores
can not contribute to the flow. Both methods were explored
and used in a predictive way, without any fitting of
parameters. Also both models gave results which agreed
quantitatively with the experimental data. One big
disadvantage of the cluster model is that it can not be used
with fractions of blocked pores above the percolation
threshold. A very large cluster is formed, and the average
cluster size blows up to infinity. The relation between the
amount of clustered pores and the effective cluster diameter is
still not clear and has to be investigated furthermore.
The random walk in the regular network is better suitable
to model the dispersion, but takes far more computing time to
get reliable results. The network used in this paper was not
capable to model molecular diffusion, but only convection,
with plug flow in the connecting pores. That the results agreed
well with the experimental data, is probably the result of using
the right co-ordination number and an isotropic structure. So
perhaps this is more important than modelling pore scale
phenomena as molecular diffusion and Taylor dispersion.
Further investigation of dispersion in network models is still
needed.
Conclusions
1. An immobile wetting phase with saturations between 10
and 15% increases the macroscopic dispersion in
unconsolidated porous media as glass beads and loose
sand by a factor 2 to 3.
2.

The increased dispersion is a result of pore blockage,

which increases the correlation length of the porous
media.

3.

We have shown how the increased longitudinal dispersion

due to an immobile and inert phase (as observed in
experiments) can be understood on the basis of
percolation concepts. Moreover the numerical results are
in good agreement with our experimental data.

Acknowledgements
This study was supported by a grant from de Nederlandse
Aardolie Maatschappij. We also wish to thank the technicians
of the Dietz laboratory for the technical support during the
experiments.
Nomenclature
C =concentration injected component, mole/L3
d =diameter, L
Dl =longitudinal dispersion coefficient, L/t2
Dm =molecular diffusion coefficient, L/t2
fsize =cluster size function
g =conductivity
keff =effective permeability, L2
L =length, L
n =cluster size
pzero =fraction of blocked pores
Pe = Peclt number
t =time, t
U =average interstitial velocity, L/t
Z =co-ordination number

=dispersivity, L
=velocity dependence of Dl
=Correlation length, L
u2 =Variance of the velocity distribution, L2/t2
t2 =Variance of the arrival time distribution, t2
=porosity
=tortuosity
Subscripts
D =dimensionless
References
1. Lake, L.W. and Hirasaki, G.J.: Taylors Dispersion in Stratified
Porous Media, Soc. Pet. Eng. J. (August 1981) 459-468.
2. Perkins, T.K. and Johnston, O.C.:A Review of Diffusion and
Dispersion in Porous Media, Soc. Pet. Eng. J. (March 1963)
70-84.
3. Legatski, M.W. and Katz, D.L.: Dispersion Coefficients for
Gases Flowing in Consolidated Porous Media, Soc. Pet. Eng. J.
(March 1967) 43-53.
4. Salter, S.J. and Mohanty, K.K.: Multiphase Flow in Porous
Media: I. Macroscopic Observations, paper SPE 11017
presented at the 57th Annual Technical Conference and
Exhibition in New Orleans (1982)
5. Batycky, J.P., Maini, B.B. and Fisher, D.B.: Simulation of
Miscible Displacement in Full-Diameter Carbonate Cores, Soc.
Pet. Eng. J. (October 1982) 647-657.
6. Orlob, G.T. and Radhakrishna, G.N.: The Effects of Entrapped
Gases on the Hydraulic Characteristics of Porous Media, Trans
AGU, Vol.39 (August 1958), 648-659.

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
7

Table 1 Properties and parameters of the porous media.

Length
Diameter
PV
Porosity
Glass beads
Sand pack 1
Sand pack 2

1.84 m
1.84 m
1.84 m

2.75 cm
2.75 cm
2.75 cm

415.9 cm3
408.6 cm3
418.2 cm3

0.383
0.377
0.386

Perm

Grain size

334 D
105 D
108 D

560 800 m
300 850 m
300 850 m

Diffusion
coefficient
2.14 e-6 m2/s
4.28 e-7 m2/s
2.38 e-7 m2/s

Table 2 Experimental tortuosity values

Water saturation
Tortuosity
0%
1.41
9.8%
1.79
15.5%
1.92

Table 3 Experimental derived dispersivities with 95% confidence limits.

10 bar
50 bar
Sw
0%
9.8%
15.5%
0%
11.4%
0.343
0.166
0.256
0.144
0.085
(cm)
0.06
0.034
0.034
0.030
0.014
1
2.07
3.03
1
1.70
/dry (-)
0.36
0.4
0.36

Table 4 Percolation parameters of the different porous media.

Porous medium
Porosity
Sw
Permeability (Darcy)
Glass beads
0.383
0%
334.4
0.383
9.8%
251.5
0.383
15.5%
202.3
Sand Pack 1
0.377
0%
**
0.377
11.4%
**
0.377
14.4%
**
Sand Pack 2
0.386
0%
107.8
0.386
11.2%
78.5
0.386
13.7%
68.0
** No accurate permeability measurements taken

Table 5 Results of the cluster network model.

Sw
Pzero
3 n
n
Glass beads

Sand Pack 2

9.8%
9.8%
15.5%
15.5%
11.2%
11.2%
13.7%
13.7%

0.22
0.18
0.35
0.28
0.22
0.19
0.30
0.25

4.42
2.95
--4.42
3.23
-6.54

1.64
1.43
--1.64
1.48
-1.87

n
2.10
1.72
--2.10
1.80
-2.56

14.4%
0.785
0.2
4.74
1.2

Co-ordination number (Z)

8.19
8.19
8.19
8.32
8.32
8.32
8.14
8.14
8.14

/dry
(-)
1.70
1.70
3.03
3.03
1.58
1.58
2.21
2.21

0%
0.0987
0.012
1

90 bar
11.2%
0.156
0.026
1.58
0.53

Pzero water wet

0
0.22
0.35
**
**
**
0
0.22
0.30

13.7%
0.218
0.028
2.21
0.28

Pzero random
0
0.18
0.28
**
**
**
0
0.19
0.25

M.L. VERLAAN, H.F. DIJKGRAAF, C.P.J.W. VAN KRUIJSDIJK

SPE 56417

Figure 1 Experimental Set-up

1.2

Dimensionless concentration (-)

0.8

0.6

Measurement
Least square fit

0.4

0.2

0
0.6

0.7

0.8

0.9

1.1

1.2

1.3

Dimensionless Time (-)

Figure 2 Fit between convection dispersion model and experimental effluent profile.

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
9

Dispersion coefficient (m 2/s)

0.0001

0.00001

0% water
9.8% water
15.5% water

0.000001

0.0000001
0.00001

0.0001

0.001

0.01

0.1

Interstitial velocity (m/s)

Figure 3 Longitudinal Dispersion coefficient versus interstitial velocity at 10 bar, for 3 different water saturations in a glass bead pack

0.0001

Dispersion coefficient (m /s)

0% water
11.4% water
14.4% water
0.00001

0.000001

0.0000001
0.00001

0.0001

0.001

0.01

0.1

Interstitial velocity (m/s)

Figure 4 Longitudinal Dispersion coefficient versus interstitial velocity at 50 bar, for 3 different water saturations in a sand pack

10

M.L. VERLAAN, H.F. DIJKGRAAF, C.P.J.W. VAN KRUIJSDIJK

SPE 56417

0.0001

Dispersion coefficient (m /s)

0% water
11.2% water
13.7% water
0.00001

0.000001

0.0000001
0.00001

0.0001

0.001

0.01

0.1

Interstitial velocity (m/s)

Figure 5 Longitudinal Dispersion coefficient versus interstitial velocity at 90 bar, for 3 different water saturations in a sand pack

0.005
0.0045
0.004

0% water
9.8% water
15.5% water

Dispersivity (m)

0.0035
0.003
0.0025
0.002
0.0015
0.001
0.0005
0
0.001

0.01

0.1

Interstitial velocity (m/s)

Figure 6 Dispersivity values derived from the10 bar measurements. The dispersivities are not constant at lower velocities.

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
11

Figure 7 Thin section of a with epoxy resin filled sand pack. The grains are white,

the open pores black and the immobile phase is coloured grey.

0.25

0.2

0.15
Density

Sand pack
glass beads

0.1

0.05

00
10

95

90

85

80

75

70

65

60

55

50

45

40

35

30

25

20

15

10

50

Pore diameter (m)

Figure 8 Pore size distributions measured with quantitative image analysis.

12

M.L. VERLAAN, H.F. DIJKGRAAF, C.P.J.W. VAN KRUIJSDIJK

SPE 56417

p=0.05

1.E-01
slope = -2

p=0.1
p=0.15

1.E-02

density

p=0.2
1.E-03

p=0.25

1.E-04
1.E-05
1.E-06
1

10

100

1000

cluster size

<cluster size>

Figure 9 Cluster size distributions for different fraction of blocked pores (p).

9
8
7
6
5
4
3
2
1
0
0

0.1
0.2
fraction blocked pores

0.3

Figure 10 Mean cluster size as a function of the fraction of blocked pores.

0.04

0.035

0.03

/L (-)

0.025

0.02

0.015

0.01

0.005

0
0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Fraction of blocked pores (-)

Figure 11 Dimensionless dispersivity calculated with the network model with uniform pore sizes, versus the fraction of blocked pores.

SPE 56417 EFFECT OF A WETTING IMMOBILE PHASE ON DIFFUSION AND MACROSCOPIC DISPERSION IN UNCONSOLIDATED POROUS
MEDIA
13

10 barexperiments
90 barexperiments
50 barexperiments
Cluster model
Random walk model

/dry (-)

0
0

0.05

0.1

0.15

0.2

water saturation (-)

Figure 12 Experimental results of dispersivities with 95% confidence limits versus water saturation. compared to the dispersivity values
obtained from the cluster and random walk model.