SOILS AND FOUNDATIONS Vol. 43, No.6, 63-73, Dec.

2003
1apanese Geotechnical Society

PORE SIZE DISTRIBUTION OF CLAYEY SOILS MEASURED BY MERCURY INTRUSION

POROSIMETRY AND ITS RELATION TO HYDRAULIC CONDUCTIVITY

HIROYUKI T ANAKAi), DINESH R. SHIWAKOTiii), NAOKI 0MUKAiiii), FUSAO Rnoiii),

JACQUES LOCATiv) and MASANORI TANAKAi)

ABSTRACT
Pore Size Distribution (PSD) of naturally deposited clayey soil samples recovered from various parts of the world
and at varying depths was measured by Mercury Intrusion Porosimetry (MIP). In addition to naturally deposited
samples, the artificially mixed soils, namely Singapore clay with Toyoura sand or diatomite, are studied in order to
know the influence of grain size component on the PSD. Characteristics of the PSD were represented by the mean pore
size (Dp50 ). As an application of the PSD characteristics to practical geotechnical engineering, an attempt has been
made to relate the hydraulic conductivity (k) with the PSD. The following interesting findings are obtained:
1) With an increase in the consolidation pressure (p'), Dpso also decreases in the same manner as that of the e-logp'
relation.
2) The range of Dpso for naturally deposited soils is relatively narrow, namely between 0.03 and 2 ,urn, except for
London clay which is an old deposit clay, i.e., Tertiary.
3) It is found from the artificially mixed soils that Dpso generally increases with an increase in silt and sand content.
4) There is a unique relation between the hydraulic conductivity (k) and nD~ 50 , where n is the porosity.

Key words: clayey soil, diatomite, hydraulic conductivity, mercury intrusion porosimetry, permeability, pore size
distribution, Toyoura sand (IGC: D6)

It is not rare to observe that the yield consolidation

INTRODUCTION
It is well recognized that the void ratio (e) is one of very
important and fundamental parameters governing the
geotechnical behavior of a soil, required for geotechnical
design, such as strength, compressibility, rigidity, con-
ductivity and so on. It is usually assumed in most of the
proposed numerical models available in literature, such
as the critical state in the Cam clay model, that a unique
relation exists among the three parameters; p', q and e,
where p' and q are the mean and the deviator stresses,
respectively. However, it is also known that noticeable
differences are observed in characteristics of the strength
or the compressibility of a soil with the same void ratio.
For example, if we compare soils naturally deposited and
reconstituted at laboratory, even though a soil is consoli-
dated under the same overburden pressure in the labora-
tory, the resultant e is different, due, for example to
differences in initial water content before consolidation,
or time duration of the secondary consolidation, so on.
Tsuchida et al. (1991) have pointed out that the tempera-
ture during consolidation also gives an important effect
on the e-log p' relation.
pressure (p;) measured by the oedometer test is some-
what larger than the in situ effective overburden pressure,
(p~ 0 ), in which stress conditions in the past are judged to
be normally consolidated from geological considerations.
It should also be pointed out that the void ratio at such a
site is always larger than that measured from a reconsti-
tuted specimen subjected to the same consolidation pres-
sure as the P~o· For example, Burland (1990) has distin-
guished Sedimentary Compression Line (SCL) from
Intrinsic Compression Line (ICL) in thee-log p' relation.
It has been considered that one of the reasons for such
different behavior is caused by a difference in the arrange-
ment of soil particles, even at the same void ratio. This
arrangement is sometimes called "soil structure", which
has been one of the most controversial themes in
geotechnical engineering, in particular as to how to evalu-
ate it in a quantitative way.
The observation of the soil structure is carried out
usually using Scanning Electron Microscope (SEM).
Tanaka and Locat (1999), for example, have revealed
that Osaka Pleistocene clays contain a large volume of
microfossils, especially the diatoms, and have suggested

il Port and Airport Research Institute (tanaka-h92y2@pari.go.jp).

iil Toa Construction.
iii) Oyo Corporation.
iv) Laval University.
Manuscript was received for review on October 3, 2002.
Written discussions on this paper should be submitted before July 1, 2004 to the Japanese Geotechnical Society, Sugayama Bldg. 4F, Kanda
Awaji-cho 2-23, Chiyoda-ku, Tokyo 101-0063, Japan. Upon request the closing date may be extended one month.

63

This is an Open Access article under the CC-BY-NC-ND license.

NII-Electronic Library Service
64 TAN AKA ET AL.

Table 1. Geotechnical index properties of soils used in this study

Name Depth (m) Sand(%) Silt (%) Clay* (%) WL (%) Wp (%) Wn (%) Reference
Mal3 29 0 50 50 99 36 94 Tanaka et al. (2002)
Mall 112 3 47 50 76 26 55 Tanaka et al. (2002)
Ma9 187 0 26 74 110 41 61 Tanaka et al. (2002)
Kyoto 23 0 25 75 107 34 58 Tanaka et al. (2002)
Bangkok 5-16 0 30-61 39-70 46-101 19-25 30-70 Tanaka et al. (2001a)

Bothkennar 3-17 0 60-72 28-40 55-77 23-32 55-77 Tanaka (2000)

Singapore 14-23 0 10-20 80-90 61-82 22-25 46-60 Tanaka et al. (2001a)

Drammen 4-16 0 36-42 58-64 34-42 19-22 32-43 Tanaka (2000)

Pusan 10 0 33 67 65 24 60 Tanaka et al. (2001b)

Tokyo 10 7 56 47 43 25 34

London 6 0 23 77 83 25 25

Note: * less than 5 .urn

Table 2. Geotechnical index properties of artificially mixed soils

that these microfossils may impart great influences on the
geotechnical properties, in particular, their compressibili-
ty. A problem in the observation of microstructure by the
SEM, however, is how to evaluate the structure in quan-
titative ways. Instead of such .observation, there are
several studies to grasp the structure, such as the pore size
distribution, using mercury intrusion porosimetry (for
example, Delage and Lefebvre, 1984). It is anticipated
that a soil consisting of large pores may behave in differ-
ent ways from that consisting of small pores, even though
they have the same void ratio. For example, the hydraulic
conductivity of the former soil will be much larger than
that of the latter soil because of small resistance for pore
water flow.
This paper presents the pore size distribution of natural
clayey soils recovered from various areas in the world.
The studied clay samples are not only Holocene deposits
but· also Pleistocene clays retrieved from depths of deep
as about 200m. In addition to these naturally deposited
soil samples, artificially mixed soils with sand or diatom-
ite soil are used to examine the influences of grain size
components on the pore size distribution. The paper also
makes an attempt to correlate the pore size distribution
with the hydraulic conductivity.
SOILS USED IN THE INVESTIGATION AND THEIR
TESTING METHODS
Soils Tested
(1) Naturally deposited soils
The geotechnical index properties of soils used in this
study are given in Table 1, more detailed information on
these soils is available in references indicated in the Table.
Ma13, Mall, Ma9 and Kyoto clay belong to the Osaka
Group, which are numbered according to deposited age.
For example, Ma13 is the youngest Holocene clay deposit
(see ltoh et al., 2000; Tanaka et aL, 2002). The deposi-
tion age of Kyoto clay is relatively old, and corresponds
3
Soil Ps (g/cm ) WL (%) Wp (%)
Singapore 2.77 82.5 22.7
Diatomite 2.26-2.37 NP NP
Toyoura 2.65 NP NP
to Ma4. All intact samples, except for Pleistocene clays of
the Osaka Group, were recovered by the fixed piston sam-
pler specified by the 1apanese Geotechnical Society
(JGS). Since the Pleistocene clays (Mall, Ma9 and
Kyoto) were very stiff, the rotary core sampler was used
to retrieve them (see for example, Tanaka et al., 2002).
Tokyo clay was recovered at Minato ward, Tokyo, and
according to chronological study, it is a late Pleistocene
clay. London clay is well-known clay in geotechnical
engineering literatures and was taken from the Heathrow
Airport, London, UK. Only this clay is a Tertiary deposit.
(2) ,Artificially mixed soils
To study the influence of grain size component, arti-
ficially mixed soils with Toyoura sand and diatomite soil
were prepared. As a base material, Singapore clay was
used. Their main properties, including the grain size
distribution, are indicated in Table 2 and Fig. 1. The
Toyoura sand is a well-documented material in Japan
because researchers of the JGS have extensively used it in
their study as a standard granular material. Its grain size
distribution is shown in Fig. 1. While the diatomite soil
used in this study was recovered from the Hiruzenbara
site, Okayama, which is the same material the authors
previously used to study the influence of diatom on
geotechnical properties (Shiwakoti et al., 1999; Tanaka
and Locat, 1999; Shiwakoti et al., 2002). Although the
particle size of the diatomite soil is mainly silty, as shown
in Fig, 1, its void ratio or water content is very large
because diatom particles have large pores, which can hold
a large volume of water. As a result, if the diatomite soil

NII-Electronic Library Service

 PORE SIZE DISTRIBUTION HYDRAULIC CONDUCTIVITY
100
80
g special chamber under a vacuum (less than - 90 kPa) and
60
....
Q)
c:
u.... 40
20
0 applying pressure. The diameter of voids is calculated
0.001 0.01 0.1
Grain Size (mm)
Fig. 1. Grain curves used for artificially mixed soils
is mixed with usual soil, its index properties such as liquid
limit ( wL) or plastic limit ( Wp) increase in spite of a
decrease in clay particle content. This tendency is com-
pletely opposite to the soil mixed with the Toyoura sand.
The selected clayey soil, namely Singapore clay, was
mixed with the Toyoura sand or the diatomite soil, using
a mixer at high water content. The mixed soils were
prepared at several proportions: namely, 0, 25%, 50%
and 75%, in terms of the ratios of dry weight of the
Toyoura sand or the diatomite soil to the Singapore clay.
Hereafter, the mixed soils are named such as S +Sa(50).
The first letter of S indicates the name of Singapore clay.
The latter letters of Sa and D mean the mixing material of
the Toyoura sand and the diatomite soil, respectively.
The number in the blanket indicates the mixing ratio as
explained above.
The mixed soil was poured into a cylindrical consolida-
tion cell having the diameter of 10 em. The final consoli-
dation pressure was 200 kPa and the completion of con-
solidation was confirmed by the so-called 3t method: i.e.,
the duration for the consolidation is three times longer
than the time for getting the primary consolidation
defined by the bending points in the settlement curve plot-
ted against time in the logarithm scale. Then, the consoli-
dation pressure was removed, and the reconstituted
sample was extruded from the cell.
(3) Consolidated intact soils
To study the influence of consolidation, some speci-
mens were consolidated under specified pressures. The
consolidation was performed using a conventional
oedometer ring, whose initial height and diameter are
20 mm and 60 mm, respectively. The load was applied
step by step to the final pressure, and the duration of
consolidation at the final pressure was 24 hours. Then the
load was removed, and the specimen was extruded from
the oedometer ring.
Mercury Intrusion Porosimetry
For the Mercury Intrusion Porosimetry (MIP), every
void in the specimen should be replaced by air: i.e., the
specimen needs to be dried. In this study, the dry freezing
method was employed to avoid changes in the pore size
distribution in the process of drying due to the volume
65
change. The specimen was cut into a cube about 5 mm,
which was then submerged into the liquid nitrogen to
freeze it. To dry the frozen specimen, it was set in a
low temperature ( -10°C). In this process, water in the
voids was sublimated without any volume change of the
specimen.
The dried specimen was then placed in the apparatus
of the MIP. Mercury was intruded into the voids by
according to the capillary force theory. The maximum
pressure of mercury intrusion is 225 MPa, which cor-
responds to a pore size diameter of about 0.005 ,urn. It is
usual to calculate the void ratio of the specimen from the
natural water content (wn), unit weight of the soil (y1) and
unit solid weight of the soil particle (p The calculated e
5 ).
in this manner should agree with the total sum of pores
measured by the MIP. However, if pores in the specimen
get shrunken in the process of the specimen preparation
or if cracks develop in the specimen, then the total pores
measured by the MIP are not identical to the e obtained
by the above conventional calculation method. There-
fore, if the discrepancy between these two values of e was
found to be more than 10%, then such test data were not
used in this study.
It should be kept in mind that the diameter of pores
measured by the MIP is somewhat different from the
impression brought by the SEM observation. In addition
to the assumption of the pore shape as cylindric, the pore
diameter obtained by the MIP corresponds to that at the
entrance of pores, but not the inside diameter of the
pores, as discussed by Lapierre et al. (1990).
Constant Rate of Strain (CRS) Oedometer Test
CRS test was carried out to obtain the e-log p' relation
as well as the hydraulic conductivity (k) of the soil sam-
ple. The size of the specimen is the same as the conven-
tional incremental load oedometer test: i.e., the initial
height and the diameter of the specimen are 20 mm and
60 mm, respectively. The upper boundary of the speci-
men was kept drained and the pore water pressure was
measured at the bottom. The effective stress (p') and k
were calculated by Eqs. (1) and (2), respectively.
p' = p- 2u/3 (1)
k=h·yw·H/2u (2)
where p and u are applied pressure (total pressure) and
the pore water pressure measured at the bottom, respec-
tively. h, Hand Yw are the displacement rate, the height of
the specimen and the unit weight of pore water, respec-
tively.
The strain rate applied for the natural soils was 0.02% I
min (3.3 x 10~ 6 /s). As the k value for artificially mixed
soils with sand or diatomite is large, the strain rate of
these materials was faster than that of the natural soils, so
as to measure the pore water pressure more accurately
(i.e., for these soils, the strain rate of 0.2% /min was
used). A back pressure of 100 kPa was applied to get high
degree of saturation.
NII-Electronic Library Service
66 TAN AKA ET AL.

A B
I
I I

0.8
Intact Sample
'bQ
::::::.
..5 0. 6
C>
0
b.O
~
"C 0. 4
........
~

0.2

0. 01 0. 1 10
Pore Diameter, DP <~-tm)
1.2

Fig. 2. Typical examples of pore size distribution measured by mercu- a: 1.0

C)
ry intrusion porosimetry (MIP) 0 0.8
...J
:!:2 0.6
>
'"0 0.4

PORE SIZE OF INTACT NATURAL SOILS 0.2

0.0
Pore Size Distribution (PSD)
Figure 2 shows typical examples of Pore Size Distribu-
tion (PSD) for intact natural soils. The vertical axis in the
figure is plotted as dV/dlogDp, where DP is the entrance
pore size diameter and dV is the volume of pores whose
diameter is Dp, expressed in terms of the unit dry weight
of soil. Therefore, the area bounded by certain Dps
indicates the volume of pores between these Dps, as
shown in the figure. If the volume of pores with DP less
than 0.005 ,urn is ignored, then the total area bounded by
the measured curve by the MIP corresponds to the void
ratio, e.
The samples indicated in Fig. 2 were selected so that ,..-,--n1'T>Tir-->..,..,.,.m.r-'"I'""M-~--r-r''i't'tmt-T'~D-207 (383.65m)

their void ratios were nearly in the same order. As shown
in Fig. 2, however, their shapes of the PSD curve are con-
siderably different for each soil. That is, the pore size of
the Bothkennar clay is significantly larger than that of
Osaka bay Pleistocene clay (Mall). It seems that this
difference is caused by the consolidation pressure,
because the Bothkennar clay was recovered from a depth
of 17 m, while the Mall of Osaka bay Pleistocene clay
was retrieved from a depth of 93 m.
Tanaka and Locat (1999) have shown the PSD for a
series of the Pleistocene clays in the Osaka bay as shown
in Fig. 3. It can be observed that as depth increases, the
area surrounded by the curve of the PSD decreases, since
the void ratio decreases due to an increase in P~o· It also
can be seen that the diameter of major pores becomes
small with increase in depth, although some specimens do
not follow this trend. These observations reveal that the
PSD is not only dependent on consolidation pressure, but
also several factors, for example, the inherent soil prop-
erties such as index properties as well as grain composi-
tion, and environmental conditions during and after
deposition. In the following sections, these anticipated
factors are discussed in more· detail.
0.001 0.01 0.1 1.0 10.0 100.0
Entrance Pore Radius (R, JJ.m)
Fig. 3. Pore size distribution of Osaka bay Pleistocene clays (after
Tanaka and Locat, 1999)
Effects of Consolidation and Representative Value for
Pore Size Distribution
Using the specimens consolidated in the oedometer
ring, the influence of consolidation on the PSD was stud-
ied. The PSD was measured in the samples at four differ-
ent consolidation pressures: a) the condition before
consolidation pressure (p') was applied: i.e., the intact
condition of naturally deposited soil; b) the condition
where the p' is about the yield consolidation pressure
(p~); c) the Normally Consolidated (NC) condition,
where the p' is about three times as large asp~; d) the p'
of 10 MPa (=the maximum capacity of the oedometer
apparatus used in this study) was applied. Typical
changes in the PSD due to consolidation for Mal3 and
Kyoto .clays are shown in Figs. 4(a) and (b), respectively.
The PSD for both clays at a) and b) states, where p' is
relatively lower compared top~, is nearly identical, except

NII-Electronic Library Service

 PORE SIZE DISTRIBUTION HYDRAULIC CONDUCTIVITY 67

Osaka Clay Ma13

Kyoto Clay
1.5 0. 8
---- a:O (intact)
-e- b: 78kPa 'liD
-+- c:2511kPa ::::- ---- a:O (intact)
- - x - - d: 10042kPa 0. 6
-5 -e- b:941kPa
Q.
c
tlD -+- c: 3530kPa
~
-.:> 0. 4
........
::;
0. 5
0. 2

0
0.1 10 0. 01 0. 1 10

Pore Diameter, DP <~-tm) Pore Diameter, DP (~)

(a) Osaka clay (Ma13) (b) Kyoto clay (Ma4)

Fig. 4. Typical examples of change in pore size distribution due to consolidation

Osaka Clay Ma13
Assumed that all pores
are equally shrunk.
1.5
---a 0
-e- b 78kPa
-+- c 2511kPa
- - x - - d 10042kPa
0. 5
0~~~~~~~~~~~~--~
0. 01 0. 1
Pore Diameter, DP <t-tm)
Fig. 5. Calculated pore size distribution of Osaka Clay Ma13, assum-
ing that the size of all pores is equally reduced by consolidation
at large Dp range. That is, the curves of the PSD at a) and
b) states are identical at .Dp s smaller than the Dp at the
peak, for example, when Dp is smaller than 1 ,urn for
Ma13 (see Fig. 4(a)). On the other hand, some volume
change takes place at pores larger than Dp s. When the
stress state is entirely in the NC state, however, the PSD
gets completely shifted from the original position to the
left side. This fact indicates that the large pore is first
compressed, then the small pores are gradually affected as
p' increases.
Figure 5 indicates how the PSD of Ma13 changes due
to consolidation as calculated from the original PSD,
assuming that volume of all pores equally decreases with
increase in pressure, regardless of their pore size. Each
calculated PSD corresponds to the consolidation stage, as
indicated in Fig. 4(a): i.e., the total area surrounded by
the calculated PSD is identical to that of measured one. It
can be seen that there are clear differences in the PSD
between observed (Fig. 4(a)) and calculated (Fig. 5)
values. Although the peak of Dp for the calculated PSD is
shifted to the left side asp' increased, the amount of the
shift is significantly smaller than that in the observed
PSD. This discrepancy suggests that the pores are not
only compressed, but also breakdown into small pores, as
p' increases.
Figures 6(a) and (b) show number of pores with differ-
ent Dp, which was calculated, assuming that the pore
shape is spherical. Let us examine what happens to pores
due to the increase inp', as taken Ma13 in Fig. 6(a) as an
example. As long asp' is slightly larger than p~, the num-
ber of pore at each Dp does not noticeably change, which
is also supported by the earlier consideration. With the
10
increase in p', however, the pores whose diameter is
larger than 0.5 ,urn get broken, and consequently become
transformed into smaller pores having diameter between
0.2 and 0.5 ,urn. Similarly, in the process of p' increasing
from 2511 kPa to 10042 kPa, pores with diameter be-
tween 0.3 and 0.5 ,urn get collapsed and become further
smaller pores. The same trend can be observed in Kyoto
clay (see Fig. 6(b)). It is found from these observations
that the increase in p' causes the reduction in the pore
volume (the void ratio) as well as decrease in the pore
diameter. Several other researchers have also reported
similar observations (for example, Griffiths and Joshi,
1989; Onizuka and Yamagami, 1998; Miura et al., 1999).
This finding is very important in considering the reduc-
tion in hydraulic conductivity due to consolidation, as
will be discussed later.
We have described the PSD as a relation between Dp
and dVI d log DP. However, it is inconvenient to quan-
titatively characterize the PSD in this way, and it is neces-
sary to show the PSD in term of a more representative
parameter. Figure 7 shows the relation between Dp and
the cumulative volume normalized by the dry soil weight,
where the cumulative pore at the infinitive large Dp cor-
responds to the total void volume, i.e., e. In this figure,
three pore diameters are defined as Dp3o, Dpso and Dp60·

NII-Electronic Library Service

68 TANAKA ETAL.

15
t;Q
'-
:z
0
0 Kyoto Clay
Osaka Clay (Ma13) ><
~

:;:; 10
"1::>
Q) --e--a 0
-e- b 78kPa
"1::> ---+- c 2511kPa
--X-- d 10042kPa
Q)
c. --a 0
.,
Q)
-e- b 941kPa
0
---+- c 3530kPa
a.. - -X- - d 10042kPa

~
Q)
..0
§
:z

0. 2 0. 4 0. 6 0. 8
0 0. 2 0. 4 0. 6 0. 8
Pore Diameter, DP (!1m)

(a) Osaka clay (Ma13) (b) Kyoto clay

Fig. 6. Change in number of pores due to consolidation

0.6

~ 0.5
sQ) Bothkennar Clay
E 0.4 'E
::I
0 .E-
>
~
0.3 0
0.

0
c. i
"C 0.2 0 0. 1
Q) 0.
1ii D =1.3!-lm ~
:; p50
E 0.1 ..;
::I 0
(.)
s::
--+- Ma13
~Mall
0 "'
Q)
_.,_Ma9
::E
----*-Kyoto
0.001 0.01 0.1 10 100 --e-- Pusan
---13---- Tokyo
Pore Diameter, D (!lm)
p
0. 01
100 1000
Fig. 7. Typical cumulative pore size and definitions of three pore di-
ameters: DP30 , Dpso and DP60 Consolidation Pressure, p' (kPa)

Fig. 9. Change in Dpso due to consolidation

I
. I

0
0. For example, the Dp3o is the diameter where 30% of the
') . total cumulative pore is attained, as shown in the figure .
0
• • Figure 8 shows how the ratio of Dp3o IDpso or Dp6o I Dpso
0

~
0
"

0
••
•
•
• •
#.
• ;
.'
•••••
. ,. . . . . . • ••••••
varies with Dpso. It can be seen that these ratios are scat-
tered in some narrow ranges and they do not change with
Dpso. This means that the pattern of the PSD curve is
Q)

Q)
e ••• • • nearly the same and the PSD can be represented in term
0"' I of Dpso as long as the range of the PSD is between 30 and
"'0 0. 9 0
c.. cOO 0 cPo 60% of the total cumulative pore. It should be kept in
0. 8 0
~
0
oo mind, however, that since these ratios are somewhat scat-
0 0. 7
tered, there can be some doubt whether the Dpso truly
"'
a:: 0. 6 represents the PSD, because pores with large Dp, even
0. 5 • ' I ' I if their volume is small, may control entire hydraulic
0. 1 conductivity. This point will be discussed later again in
Mean Pore Di arne t er, DP50 (f.tm) more detail.
Figure 9 shows how the Dp changes due to consolida-
Fig. 8. Ratio of DP30 /Dpso or DP60 /Dpso for various soils tion, using the newly defined Dp 50 • Since the tested speci-
men has different p{ values due to various sampling

NII-Electronic Library Service

 PORE SIZE DISTRIBUTION HYDRAULIC CONDUCTIVITY 69

10
•0 Osaka
Bangkok
•
X
Bothkennar
Singapore
+ Drammen e
6 Pusan ~
... Tokyo 0

10 r-"1 •
133
London
Louis ev i I I e
"T""T"": 0 "'

1-
(After Lapierre et al., 1990)
.. ~
Q)
.._
0

e ~ . ';
-E- 0
·~. -----osaka Ma13
0 "' 1 ~ • •
0+
6
lo •• -=
.,;
·-
0 0. 1 ~Mall
---+--- Ma9
Q)
1-
+
+ + >so<
X
. c:
"-
... xx 133
"'
Q) ~Kyoto
0
a.
)<:
X \ I • ::;;:
~Pusan
':; ---'f- Tokyo
..; o Intact Clays
·-
0
0. 1 ~
F 0.01
•
.:
c::
~
0 0.5 1.5 2. 5
::;;;
. Void Ratio, e
. All specimens are intact

0. 01 . . I L ..I. .1. . I . I . Fig. 11. Relation between e and Dpso caused by consolidation
0. 5 1.5 2 2.5

Void Ratio, e .. .. .
.

Fig. 10. Relation between e and Dpso for natural soils at in situ stress
•
0
Osaka Group
Osaka Canso.
conditions 2. 5 - • Bangkok
Bo thkennar
Pusan
• •• -

o\~x
X
+ Singapore
•... Drammen
Others
0

depths, some tests start from smallp' value (for example, - -.

Pusan or Ma13), while other samples with relatively high 0
1-
1- e e ~ X
.
p~ start to consolidate at high pressure. Nevertheless,
"'
1.5 ~ if<.++ -.
..•
•
0::
1- 0 X
with the increase in p', the value of Dpso becomes smaller -c
0
1-
London
0
+oat
X
.
and a liner relation exists between them in logarithm
scales. However, the relation between p' and Dpso varies
with samples. For example, even after the specimens were
>
1--
1-
0
0\ ~
o o•
ore.
;· ... -
consolidated at 10 MPa, Dpso of the Tokyo clay is 4 times 0. 5 p e Tokyo -
larger than that of the Kyoto clay. This test result indi-
cates that the variation in the Dpso cannot be explained 0
10-12
All specimens except for Osaka Conso. are intact
. ...
10-11
. ..
10-10
... .
10-8
only by the consolidation pressure, though it is recog-
nized that the magnitude of p' is a very important factor Hydraulic Conductivity, k (m/s)
in governing the PSD.
Fig. 12. Relation between k and e for various natural soils
Relation between Dpso and e
Figure 10 shows the relation between e and Dpso for all soils (e is nearly equal to 0.5), the values of Dpso for
intact soils at their in situ stress conditions. The relation these clays are considerably different from each other.
for Louiseville clay is referred from the literature
(Lapierre et al., 1990) and plotted in this figure. It is
observed, as a general trend, that with the decrease in e, RELATION BETWEEN HYDRAULIC
the Dpso slightly decreases. However, there is a large scat- CONDUCTIVITY AND PORE SIZE
ter in this relation. For example, relations for Bangkok, The hydraulic conductivity (k) of a soil is one of the
Bothkennar and Pusan clays are located in the upper part most important geotechnical parameters. Mostly, the
of the relation for the Osaka bay clays. This means that value of k is not directly used in geotechnical engineering
these clays possess relatively larger pore diameters com-. design, instead, it is incorporated as the coefficient of
pared with that for Osaka bay clays, even at the same consolidation (cv) for the prediction of the rate of settle-
void ratio. The relations for Louiseville, Singapore and ment with the time. However, recently with increasing
Drammen clays are located between them. concern of environmental problems, the value of k itself
Figure 11 shows the relation between e and Dpso in the has become a very important parameter in seepage prob-
process of consolidation, where the data are the same as lems of polluted fluid. It is well established from previous
in Fig. 9. As e decreases due to consolidation, so does the studies that k decreases with a decrease in e and there is a
Dpso. The relation between e and the Dpso for consolidated unique relation between e and k of a soil, regardless of its
sample follows the relation measured at the intact condi- stress history. However, if the relation between e and k is
tion. It is very interesting to note that although the void compared for various soils, some scatters exist in data as
ratio after consolidation to 10 MPa is nearly the same for indicated in Fig. 12, which were obtained in the present

NII-Electronic Library Service

70 TAN AKA ET AL.

• Pusan
0 Singapore "' •
0
Osaka Group
Osaka Conso. • X
5 fE Bangkok
0
c.
• Bangkok Pusan ~
•
0
+
Ariake
Bothkennar
X
+
1:1
A
Bo thkennar
Singapore
Drammen
Others
~XX
x• ~ x
>i 0
,
~ 4 ['; Louiseville 0
·s;: All specimens except for
u
<(
Osaka Conso. are intact.
+
+ + + /::,~
;:- 0 0 0. 1 i+•
+u <( + +
3 0
<(
::::::::-
c0 • ljJ+
~0~ •

u::-u +
0 ce o
() ctl
0
c. Tokyo
-;ooU:
.._..
2 c: 0. 01
0
If;. [';
~
0

0 cWdil 0 fE 0 00
London
1.L
() 0~0 k=6.54*10-11 ((e /CF/A +1W c
c:
0

0 c 0. 001 iioO-......L-.........."""""&.&&..---1.--L....&..&.I........._~"""'-i..&.&.J,;...___._..&....I.ooloolol.&.l.l
10-12 1o-1o

0 Hydraulic Conductivity, k (m/s)

10·10 10·9

Hydraulic Conductivity (m/s}
Fig. 13. Applicability of Mesri et al. 's proposal for permeability meas-
ured by authors
study. If we consider the relation for sand and clay
materials, then it is understandable that k is not simply
related to e. We know the fact that k of sandy soil is, in
general, much higher than that of clayey soil, in spite of
the fact that the e of sand is significantly smaller than that
of clay.
Mesri et al. (1994) have considered the grain size com-
ponent as an influencing factor to define the k values.
This concept is quite acceptable because water cannot
smoothly flow in a soil with small void diameter, when
the soil consists of small size particles. They have also
considered the activity of clay particles because the size
of clay particle is different according to clay mineral.
Finally, they have proposed the following equation:
4 figure, a relatively good correlation is seen between k and
k= 6.54 x 10- 11 (e/CF )
Ac+1
where Ac, CF, k are activity, clay fraction (not percent)
and hydraulic conductivity (cm/s), respectively.
The above relation is applied to the authors' data and
the results are shown in Fig. 13. It can be noted that
considerably large scatters still exist in this relation.
Until now, several researchers have tried to establish
the correlation between the PSD measured by MIP and k
(for example, Garcia-Bengochea et al., 1979; luang and
Holtz, 1986). Watabe et al. (2000) have also attempted to
correlate k with the PSD measured by the suction tests.
Their material for research is compacted soil, and their
concern is difference in k especially between wet and dry
sides of the compaction curve. Lapierre et al. (1990) have
als.o studied the PSD and k .relation for the Champlain
clay, which is widely distributed in Quebec, Canada. In
their paper, they reviewed several models to relate the
PSD to k, such as Capillary, Hydraulic-radius and
Mashall probabilistic models. For example, Capillary
model obtained by Garcia-Bengochea et al., (1979) is
expressed by Eq. (4).
Fig. 14. Relation between k and nD~ 50 for natural soils (n is the
porosity of soil)
- ny 2
k- f.l Iij(Dpi)Dpi (4)
32
where, n is the porosity, which is equal to e/(1 +e), y is
the unit weight of the permeant fluid, f.1 is the absolute
fluid viscosity and f(Dpi) is volumetric frequency at the
Dpi and its summation is unity as shown in Eq. (5)
(5)
As already discussed, it is found that the shape of the
PSD is roughly identical for every soil, and may be
represented by the Dpso. Therefore, proposed models
relating with DP and k, including Capillary model in Eq.
(4), may be simply related by nD~ 5 o. The relation between
nD~ 50 and k is shown in Fig. 14, where the relations for
both intact and consolidated states are plotted. In the
(3) nD~ 50 , except for Bothkennar and Bangkok clays. The k
value for these clays is somewhat small in spite of large
nD~ 50 parameter. Since these clays have relatively large
Dpso compared to e, as shown in Fig. 10, the reason for
large nD~ 50 is large Dpso, not n. Therefore, even though
Bothkennar and Bangkok clays have large pore diameter,
it can be said that their conductivity is small, probably
due to lack of continuity of pores or different shape of
pores.
PORE SIZE OF ARTIFICIAL MIXTURE
Let us consider the relation between k and nDi;Jso
in more detail, using artificially mixed soils. Before
examining this relation for the artificial mixture, it will be
shown how diatomite or sand content affects the PSD.
Figures 15(a) and (b) show the PSD for diatomite and
sand mixtures in Singapore clay with different mixing
ratios, respectively. In case of diatomite mixture, with the
increase in diatomite content, void ratio as well as Dp
increases monotonously. Comparison of the two figures
also reveals the tremendously large volume of voids in the

NII-Electronic Library Service

 PORE SIZE DISTRIBUTION HYDRAULIC CONDUCTIVITY 71

0.5
3
:§ 0.4 ~8+8a(O)
:§
I ~8+0(0) I
- e - 8+8a(25)
--6--- 8+8a(50)
o"" 2 --e---8+0(25) 0 "" 0.3
0) ~8+8a(75)
0 -:--*-- 8+0(50) Cl
=a ---+--- 8+0(75) ~
0

>"0 >
"t:l
0.2

0.1

o~~~~~~~~~MMJ 0
0.01 0.1 10 100 0.01 0.1 10 100
Pore Diameter, 0 p (!lm) Pore Diameter, D (!lm)
p

(a) Mixture of Singapore clay and diatomite (b) Mixture of Singapore clay and Toyoura sand

Fig. 15. Pore size distribution curve for artificially mixed soils

10 •
diatomite mixture. On the other hand, the PSD for the • • • I • • I • • • I • I • • • •
:
mixture of Singapore clay and Toyoura sand looks more
complicated as shown in Fig. 15(b). Unlike the mixture of
•S\:crease + Diat~ase ~
diatomite, with the increase in sand content, the void • -It:>~~.
ratio decreases (this trend can be more clearly seen in • ""~t:a tiP' • co
1 =- ~
.
Fig. 16). However, the Dpso increases with increase in
sand content, although the shape of the PSD is not so 0 lSI
~lSI
QIS!sJ 0
..:
ISiog
smooth as that of the diatomite mixture. When the con- 0
0. !•

tent of sand become 75%, the shape of the PSD suddenly 0 0 Osaka Group
0 Pusan
changes and the pore size is distributed in a relatively 0 Tokyo
X X London Clay ~
narrow range. This may be caused by the development of + Bangkok
aggregates, which consist of bridges of sand particles 6 Bothkennar
lSI Singapore
glued by a small quantity of clay. As already mentioned, •• Mixture S+D
Mixture S+Sa
the shape of the PSD is not so much different for most I I ••
0. 01 I I I I •• I

natural soils because Dp6o I Dpso or Dp3o IDpso is nearly con-

stant as shown in Fig. 8. It should be kept in mind,
however, that this finding cannot be applied to the sand
mixture soil.
Figure 16 shows the relation between Dpso and e for
naturally deposited soils and artificially mixed soils.
Again this figure clearly shows that in case of sand mix-
ture, void ratio decreases with increase in sand content.
The opposite trend is observed in the diatomite mixture,
since diatomite has large voids in its particles. However,
the Dpso of sand and diatomite mixtures increases with
increase in their content. Especially, it is likely that sand
content plays an important role to increase the Dpso in
spite of a small change in e, compared with the diatomite
mixture. From this investigation, the reason for relatively
large Dpso for the Bothkennar clay is inferred to be due to
the large content of silt and sand particles, as shown in
Table 1; on the other hand, clay content of Singapore and
Louiseville clay (for Louiseville clay, it is about 80%,
according to Lapierre et al., 1990) is quite large, therefore,
their Dpso is small. However, the reason for a large Dpso of
Bangkok clay as well as Osaka clays cannot be explained
only by grain component. It may be concluded that
though the grain component is an important factor in
governing the PSD, there must be other equally im-
portant factors.
0 4
Void Ratio
Fig. 16. Relation between Dpso and void ratio for artificially mixed
soils
Figure 17 shows the relation between k and e, measured
by the CRS test for the artificially mixed soils. In the
curve of k and e relation, there is a bending point indi-
cated by an arrow. The sample was reconstituted under
the consolidation pressure of 200 kPa. When p' is smaller
than the reconstituted pressure, the measurement of the
pore water pressure is not accurate enough to calculate
reliable value of k. Therefore, the relation between k and
e should be considered only in the part of the left hand
side from the arrow. It is very interesting to note that as
void ratio increases or decreases due to increase in con-
tent of diatomite and sand respectively, the value of k
increases. This fact also confirms that k cannot be deter-
mined only by void ratio.
Since the PSD of the mixture was measured after pre-
consolidating at 200 kPa, the k value of the sample may
correspond to that indicated by the arrow in Fig. 17.
Similarly, as in the case of the natural soils, the relation

NII-Electronic Library Service

72 TANAKA ET AL.

. '''"I• . .. .. ....... . . ...

~

Dp from MIP. This fact is easily recognized when we

- e - S(O)
observe microstructure· of soil by Scanning Electron
-11- S+D~25~

/'~:'1
~
- -+- S+D 50
- - _.- - S+D 75
- Microscope (for example, Tanaka and Locat, 1999).
- o · - S+Sa ~25~ ..
4 1- ---<>-- S+Sa 50
· -6- S+Sa 75 -.. CONCLUSIONS
,, <,i ..
-c
. •
~

}
_.
Ji.. - Pore size distribution (PSD) for clayey soils was meas-
0
ured by Mercury intrusion porosimetry (MIP). To study
>

- Y- - the influence of grain composition, the PSD was meas-

~~~"'
ured in artificially mixed soils consisting of Singapore
- clay+ Toyoura sand and Singapore clay+ diatomite soil.
~
~ ~-- ;;;;;_..·?{.- -
--- ~ 0 ' . --fs· The following conclusions are obtained:
~ .. .. .. .... . .. : 1) Due to consolidation, the PSD is changed. The larger
J Q-10 size pores get broken first, before the breakage of
Hydraulic Conductivity, k (m/s) smaller pores, into several smaller pores in accord-
ance with the reduction of the pore volume.
Fig. 17. Relation between void ratio and hydraulic conductivity for 2) From the relation between cumulated pore volume
artificially mixed soils and pore diameter (Dp), the Dp3o, Dpso and Dp 6o are
defined by the corresponding Dp at the cumulated
10 .... • ,, 'I ,. ... .. . volume with 30, 50 and 60% of the total volume,
~
~
• II respectively. It is found that the ratios of Dp3oiDpso or
+
6
Iii
Dp6o I Dpso are nearly identical for every soil. This
~
+
means that the shape of the PSD can be considered to
1=- • +
611
11>6 6
B "':
be not so different from each other, and the charac-
• r9
6
Lf:J
teristics of the PSD can be represented by Dpso, the
~E
ISl mean diameter of pores.
ISl [S]ISJ
~ 3) The value of Dpso decreases with the increase in
0. 1 :- :
131Sl
qg 0
ISl
0
0 0 consolidation pressure, and its relation is linear in
0
0
Osaka Group
Pusan
log-log plots.
0
X
Tokyo
London CI ay
4) There is a trend that the Dpso for both intact and
0. 01 ~
+
6
Bangkok
Bothkennar
consolidated specimens decreases with a decrease in
rsJ
1111
Singapore
Mixture SD
e, although there is a large scatter in this relation.
X Intact and mixtured Samples
.. .. • .. Mixture SS 5) A good correlation exists between the hydraulic con-
0. 001 ductivity (k) and nD~ 50 , where n is the porosity, ei(I
+e), for both the intact and consolidated specimens.
Hydraulic Conductivity (m/s) However, k for Bothkennar and Bangkok clays is
relatively small in spite of their large nD~ 50 •
Fig. 18. Relation between nDpso and hydraulic conductivity for arti- 6) Using artificially mixed soils, the influence of grain
ficially mixed and natural soils size component on the PSD was studied. From these
two mixtures, completely opposite trend has been
obtained; i.e., for the sand mixture, the e decreases
between k and nD~so for artificially mixed soils is plotted with the increase in sand content, while for diatomite
in Fig. 18. A good relation exists between k and nD~so in mixture, the e increases with the increase in diatomite
spite of so large a scatter of k and e relation. However, content, because of the existence of large pore in
the relation for the artificial mixtures is located in the particles of the diatomite. For both mixture of diat-
same range as that of the Bangkok and Bothkennar clays, omite and Toyoura sand, however, Dpso increases
which are located in the relatively upper part compared to with the increase in their contents. It is considered
other natural clays. The reason for this discrepancy may that the content of silt or sand particles has an im-
be attributed to 1) the· representation of the PSD as Dpso portant effect on the PSD.
and-2) derivation of the pore diameter measured by MIP: 7) The value of k for both the sand and the diatomite
i.e., assumption that the shape of pores is cylindrical. As mixtures increases with the increase in their content,
already seen in Fig. 15(b), the pattern of the PSD for sand even though the their e and k relations are completely
mixture is considerably different compared to that of different.
natural soils or diatomite mixture. The pattern of the 8) Even for the artificial mixture, a clear relation exists
PSD for even natural soils are slightly different from each between nDpso and k. This relation is close to that of
other, if we carefully examine the ratio of Dp 3oI Dpso or Bothkennar and Bangkok clays.
Dp6oiDpso in Fig. 8. In addition, as pointed by Lapierre et
al. (1990), the shape of a real pore is somewhat different
from cylindrical, which has been assumed to obtain the

NII-Electronic Library Service

 PORE SIZE DISTRIBUTION HYDRAULIC CONDUCTIVITY
REFERENCES
1) Burland, J. B. (1990): On the compressibility and shear strength of
natural clays, Geotechnique, 40(3), 329-347.
2) Delage, P. and Lefebvre, G. (1984): Study on the structure of a
sensitive Champlain clay and its evolution during consolidation,
Can. Geotech. J., 21, 21-35.
3) Garcia-Bengochea, I., Lovell, C. W. and Altschaeffi, A. G. (1979):
Pore distribution and permeability of silty clays, J. of the Geotech.
Engrg., ASCE, 105, 839-856.
4) Griffiths, F. J. and Joshi, R. C. (1989): Change in pore size distribu-
tion due to consolidation of clays, Geotechnique, 39, 159-167.
5) Itoh, Y., Takemura, K., Ishiyama, T., Tanaka, Y. and Iwaki, H.
(2000): Basin formation at a contractional bend of a large transcur-
rent fault. Plio-Pleistocene subsidence of the Kobe and northern
Osaka basins, Japan, Tectonophysics, 321, 327-341.
6) Juang, C. H. and Holtz, R. D. (1986): Fabric, pore size distribu-
tion, and permeability of sandy soils, J. of the Geotech. Engrg.,
ASCE, 112, 855-868.
7) Lapierre, C., Leroueil, S. and Locat, J. (1990): Mercury intrusion
and permeability of Louiseville clay, Can. Geotech. J., 27,
761-773.
8) Mesri, G., Kwan Lo, D. 0. and Feng, T. W. (1994): Settlement of
embankments on soft clays, Proc. of Settlement '94. ASCE,
Geotechnical Special Publication (40), 8-56.
9) Miura, T., Yamadera, A. and Hino, G. (1999): Consideration on
compression properties of marine clay based on the pore size
distribution measurement, J. of Geotech. Engrg., JSCE, (624/
111-47), 203-215 (in Japanese).
10) Onizuka, T. and Negami, T. (1998): Microstructure of undisturbed
and reconsolidated Ariake clay, Tsuchi-to-Kiso, 46{4), 17-20 (in
Japanese).
73
11) Shiwakoti, D. R., Tanaka, H., Locat, J. and Goulet, C. (1999):
Influence of microfossils on the behaviour of cohesive soil, Proc. of
the 11th Asian Regional Conf., Seoul, South Korea, 23-26.
12) Shiwakoti, D. R., Tanaka, H., Tanaka, M. and Locat, J. (2002):
Influences of diatom microfossils on engineering properties of soil,
Soils and Foundations, 42(3), 1-17.
13) Tanaka, H. (2000): Sample quality of cohesive soils: lessons from
three sites, Ariake, Bothkennar and Drammen, Soils and Founda-
tions, 40(4), 57-74.
14) Tanaka, H. and Locat, J. (1999): A microstructural investigation of
Osaka bay clay: the impact of microfossils on its mechanical behav-
iour, Can. Geotech. J., 36, 493-508.
15) Tanaka, H., Locat, J., Shibuya, S., Tan, T. S. and Shiwakoti, D.
R. (2001a): Characterization of Singapore, Bangkok and Ariake
clays, Can. Geotech. J., 38, 378-400.
16) Tanaka, H., Mishima, 0., Tanaka, M., Park, S. Z., Jeong, G. H.
and Locat, J. (2001b): Characterization of Yangsan clay, Pusan,
Korea, Soils and Foundations, 41(2), 89-104.
17) Tanaka, H., Ritoh, F. and Omukai, N. (2002): Quality of samples
retrieved from great depth and its influence on consolidation prop-
erties, Can. Geotech. J., 39, 1288-1301.
18) Tsuchida, T., Kobayashi, M. and Mizukami, J. (1991): Effect of ag-
ing of marine clay and its duplication by high temperature consoli-
dation, Soils and Foundations, 31(4), 133-147.
19) Watabe, Y., Leroueil, S. and Le Bihan, J.-P. (2000): Influence of
compaction conditions on pore-size distribution and saturated
hydraulic conductivity of a glacial till, Can. Geotech. J., 37,
1184-1194.
20) Yamaguchi, H., Hashizume, Y. and Ikenaga, H. (1992): Change in
pore size distribution of peat in shear process, Soils and Founda-
tions, 32(4), 1-16.
NII-Electronic Library Service