Soils and Foundations 2015;55(1):63–73

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

www.sciencedirect.com
journal homepage: www.elsevier.com/locate/sandf

Modification of triaxial apparatus for permeability measurement

of unsaturated soils$
S.G. Goha, H. Rahardjob,n, E.C. Leongc,1
a
Coastal Engineering, Infrastructure & Land Survey, Surbana International Consultants Pte. Ltd., 168, Jalan Bukit Merah, Surbana One, Singapore 150168,
Rep. of Singapore
b
School of Civil & Environmental Engineering, Nanyang Technological University, Block N1, #1B-36, Nanyang Avenue, Singapore 639798, Rep. of Singapore
c
School of Civil & Environmental Engineering, Nanyang Technological University, Block N1, #1C-80, Nanyang Avenue, Singapore 639798, Rep. of Singapore
Received 16 July 2012; received in revised form 22 September 2014; accepted 20 October 2014
Available online 2 January 2015

Abstract

Both the shear strength and the permeability of unsaturated soils are important engineering properties that are required in numerous
geotechnical designs. Many studies on the shear strength of unsaturated soils have been reported; however, only a limited number of studies on
the permeability of unsaturated soils have been presented. This might be due to the fact that the time and the costs associated with unsaturated
permeability measurements are excessive. The purpose of this paper is to introduce the modification of a triaxial apparatus for the direct
measurement of permeability in conjunction with shear strength tests on unsaturated soils under multiple cycles of drying and wetting. Detailed
designs and modifications are carried out to allow the unsaturated permeability and the shear strength of a soil to be measured for the same soil
specimen. The test results are found to be more consistent as both measurements are conducted on the same specimen. A series of unsaturated
permeability tests and then a series of unsaturated consolidated drained (CD) triaxial tests, under multiple cycles of drying and wetting, are
conducted on three different soils, and the results are shown to be compatible with the theory reported in literature.
& 2015 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Permeameter; Hysteresis; Permeability; Shear strength; Multi-cycle; Drying; Wetting; Unsaturated soils

1. Introduction clay barriers for the containment of contaminated soils, water

Many geotechnical and geoenvironmental problems invol-
ving unsaturated soils require an understanding of the unsatu-
rated permeability of the soils. These problems include the
stability of slopes, road and railway embankments, earth dams,
n
Corresponding author. Tel.: þ65 67905246; fax: þ 65 67910676.
E-mail addresses: Q060001@e.ntu.edu.sg (S.G. Goh),
chrahardjo@ntu.edu.sg (H. Rahardjo), cecleong@ntu.edu.sg (E.C. Leong).
URL: http://www.ntu.edu.sg/cee/ (H. Rahardjo).
1
Tel.: þ65 67904774; fax: þ 65 67910676.
Peer review under responsibility of The Japanese Geotechnical Society.
management structures, contaminant transport in unsaturated
soil zones and many more. The permeability of soil is a soil
property which expresses the rate of water flow through the
soil. It refers to the Darcy's coefficient of permeability, k, or the
simplified term, coefficient of permeability, in civil engineer-
ing (Holtz and Kovacs, 1981). The permeability of a saturated
soil, with respect to the water phase, is a function of the void
ratio of the soil. However, in unsaturated soil, the permeability
of the soil, with respect to the water phase, is a function of
both the void ratio and the water content of the soil (Fredlund
and Rahardjo, 1993). Various studies have shown that the

http://dx.doi.org/10.1016/j.sandf.2014.12.005
0038-0806/& 2015 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.
64 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
water in soil can only flow through the voids that are filled
with water in a continuous path. As a result, the permeability
of an unsaturated soil is dependent on two stress state
variables, i.e., the net normal stress and the matric suction,
which control the water content of an unsaturated soil.
The matric suction has a dominant influence on the amount
of water in a soil. The permeability of an unsaturated soil
decreases significantly as the water content decreases when the
matric suction of the soil increases. The coefficient of
permeability of soil can vary by several orders of magnitude
when the matric suction varies in the range of practical interest
to engineers. The unsaturated permeability of a soil has been
described to have a close relationship to the Soil–Water
Characteristic Curve (SWCC) of the soil during the drying
and wetting processes (Fredlund et al., 1994). In other words,
the unsaturated permeability of a soil is hysteretic as it depends
on the soil properties associated with either the drying or
wetting paths (Fredlund and Rahardjo, 1993; Fredlund, 2006;
Gallage et al., 2013). The shear strength of unsaturated soils is
also a crucial engineering property in various geotechnical
problems. Many studies have been carried out in order to
understand the shear strength behavior of unsaturated soil (Han
et al., 1995; Vanapalli et al., 1996; Rahardjo et al., 2004).
Numerous permeability and shear strength results of unsa-
turated soils have been reported in the literature. However,
these results were obtained using different specimens and
different apparatuses. Furthermore, most of the results were
limited to the initial drying and/or wetting paths. Therefore, it
is appropriate to develop an apparatus that can perform direct
measurements of both the shear strength and the permeability
of unsaturated soil using the same specimen in order to achieve
cost effectiveness and to shorten the time needed for both tests.
The test results would be more meaningful if both the
permeability and the shear strength results could be directly
obtained from the same specimen. The objective of this paper
is to describe the modification of a triaxial apparatus for both
the permeability and the shear strength testing of unsaturated
soils under multiple cycles of drying and wetting.
2. Determination of permeability for unsaturated soil
There are two approaches for obtaining the permeability of
an unsaturated soil, i.e., direct and indirect approaches (Leong
and Rahardjo, 1997a). Direct permeability measurements can
be conducted in a laboratory or in the field. Although field
measurements are always considered to be more representa-
tive, laboratory measurements are usually preferred due to the
lower costs and fewer uncertainties involved in laboratory
measurements (Benson and Gribb, 1997). On the other hand,
the indirect approach to the permeability determination of an
unsaturated soil refers to the use of a permeability function
associated with particular soil properties, e.g., SWCC, to
estimate the permeability of the unsaturated soil.
The laboratory permeability measurements of an unsaturated soil
can be performed using a steady-state method or an unsteady-state
method. The steady-state method (also called constant-head method
or constant-flow method) is performed by maintaining a constant
hydraulic head gradient across a specimen (Fredlund and Rahardjo,
1993). A steady-state water flow across the specimen is created
while the matric suction and the water content of the specimen are
maintained as constants. Benson and Gribb (1997) commented that
more accurate results can be produced by the steady-state method
using Darcy's law, although it is always more time-consuming, as
compared to the unsteady-state method. The unsteady-state methods
(i.e., variable–head method, infiltration techniques, instantaneous
techniques, etc.) can be used either in the laboratory or in-situ. The
unsteady-state methods have several variations. The differences are
mainly found in the flow process and in the measurement of the
hydraulic head and the flow rate (Fredlund and Rahardjo, 1993;
Krisdani, et al., 2009). The flow process can be a wetting process,
where water flows into the specimen, or vice versa. When using the
variable-head method to measure the permeability, difficulties are
often encountered in maintaining the stress state of the specimen
during the test (Agus et al., 2003)
In general, there are two types of permeameters used to
measure the unsaturated permeability of soil, i.e., the rigid wall
permeameter and the flexible wall permeameter. Klute (1965),
Gan and Fredlund (2000), Lu et al. (2006), Vanapalli et al. (2007)
and Gallage et al. (2013) developed rigid wall permeameters to
measure the unsaturated permeability of soil. On the other hand,
Barden and Pavlakis (1971), Huang et al. (1998), Agus et al.
(2003) and Moncada and Campos (2010) developed flexible wall
permeameters to measure the unsaturated permeability of soil.
The main advantage of using a flexible wall permeameter over a
rigid wall permeameter is that the changes in the volume of the
soil, during consolidation, drying and wetting processes, can be
monitored and determined. Another advantage of a flexible wall
permeameter is that the shrinkage of specimens during drying
does not affect the accuracy of the permeability measurement.
Generally, the permeability of soil remains relatively constant at
matric suctions below AEV and starts to decrease significantly at
matric suctions beyond AEV. The shape of the permeability
function is similar to the shape of SWCC (Fredlund et al. 1994;
Gan and Fredlund, 2000). Numerous permeability functions for
unsaturated soils, e.g., Richards (1931), Brooks and Corey (1964),
Mualem (1976), Kunze et al. (1968), van Genuchten (1980) and
Leong and Rahardjo (1997a), have been introduced to estimate the
unsaturated permeability of soil. Most of the equations were
developed based on the relationship among permeability, SWCC,
pore sizes and the pore-size distribution of the soil.
The permeability functions are able to provide a quick
approximation of the permeability of soil; however, the
estimations using some of the permeability functions may
significantly underestimate the actual unsaturated permeability
of certain types of soil (van Genuchten, 1980; Fredlund et al.,
1994; Chiu and Shackelford, 1998). Therefore, it is always
recommended that the direct permeability measurement of an
unsaturated soil be conducted, even though it is more time-
consuming (Fredlund and Rahardjo, 1993).
3. Modified triaxial apparatus for unsaturated soil tests
A modified triaxial apparatus has been developed for the
direct measurement of unsaturated permeability before a
 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
Fig. 1. Modified triaxial apparatus for permeability measurement before shearing of unsaturated soils.
shearing test on the same specimen. The apparatus was
modified from the triaxial apparatus described by Fredlund
and Rahardjo (1993) and Rahardjo et al. (1995), which was
originally developed to determine the shear strength on the
initial drying path only. In this study, further improvements
and modifications were performed on the triaxial apparatus in
order to allow the unsaturated permeability and the shear
strength of a soil to be measured on the same specimen under
multiple cycles of drying and wetting. In order to prevent the
cavitation of water (at approximately  100 kPa water pres-
sure) in the pore-water pressure system, the axis-translation
technique (Hilf, 1956) was used. Both pore-air, ua, and pore-
water, uw, pressures were controlled in order to apply the
matric suction, (ua–uw), on the soil specimen. The modified
triaxial apparatus mainly consists of a triaxial cell and a base,
pressure lines and a measurement system, flushing lines, a
volume change measurement system, a shearing and load
measurement system as well as a data acquisition system.
Fig. 1 shows the layout of the improved triaxial apparatus for
unsaturated soil testing.
3.1. Triaxial cell and base
The apparatus has a maximum working pressure of
1700 kPa, which is much higher than the pressure required
in this study. The designs of the ceramic disks, pedestal,
top cap and triaxial base, as described in Fredlund and
Rahardjo (1993), were modified to accommodate the specimen
65
of 50 mm in diameter for the unsaturated permeability and
shearing tests. With this improved design, the time required for
saturation and consolidation may also be shortened.
The modified top cap and pedestal have similar designs,
which can be used to apply and control air and water pressures
simultaneously. The modified pedestal was made of stainless
steel, while the top cap was made of aluminum in order to
minimize the self-weight of the modified top cap on the soil
specimen. Spiral grooved water channels were etched into the
pedestal and the top cap. Two water pressure outlets are
located on the base of the spiral groove, which are designed for
applying water pressure into the water compartment, and
subsequently, into the specimen through the modified high
air-entry ceramic disk. A protruding air pressure outlet was
constructed on top of the grooves, which are designed for
applying pore-air pressure by allowing the air to pass through
the modified ceramic disk and then distributing it into the
specimen through the porous disk. Drawings of the modified
pedestal, modified top cap and modified triaxial cell base are
shown in Figs. 2, 3 and 4, respectively. As shown in Fig. 4,
there are 7 outlets at the modified triaxial cell base to be
connected to valves A to G (see Fig. 1). Fig. 5 shows the
modified top cap and pedestal with spiral grooves as well as
the modified ceramic disk with a porous disk.
The saturated high-air entry ceramic disk is the key element
that separates the air and the water phases in unsaturated soil
testing. A ceramic disk, with an air-entry value higher than the
maximum matric suction applied during the test, is necessary
66 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73

Fig. 2. Drawing of modified pedestal.

Fig. 3. Drawing of modified top cap.

for the measuring, the applying and the controlling of the pore-
water pressure in the soil specimen. A 5-bar high air-entry
ceramic disk with a thickness of 7.14 mm was selected by
considering the applied matric suction range during the test
and the saturated permeability of the ceramic disk. Four
equivalent grooves were constructed on top of the ceramic
disk (see Fig. 5). An opening was constructed at the edge of
the base of the grooves to allow air, which is supplied from the
protruding air pressure outlet from the spiral grooves (see
Fig. 5) of the pedestal, to pass through the ceramic disk. The
opening of the ceramic disk was designed to fit over the
protruding air pressure outlet from the spiral grooves, while the
spaces between the grooves of the ceramic disk were designed
to be filled with porous metal. The main purposes of the
ceramic and the porous disks are to serve as a medium for
uniformly distributing the water pressure and the air pressure,
 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
Fig. 4. Drawing of modified triaxial cell base.
Water outlets
Air outlet
Fig. 5. Modified top cap and modified pedestal with spiral grooves.
respectively, into the soil specimen as well as to provide a flat
surface of contact with the soil specimen for the shearing test.
The modified high air-entry ceramic disk was sealed on the
modified pedestal by applying slowly setting epoxy glue along
its circumference. During the test, air pressure was supplied
from valves C and G (see Fig. 1), while water pressure was
supplied from valves A and F (see Fig. 1).
The thermal sensor (LM35), from RS Components Pte Ltd.,
was installed at the outside of the triaxial cell wall (at the
middle height of the triaxial cell) to monitor the ambient
temperature. The accuracy of the LM35 thermal sensor was
7 0.3 1C. Another submersible thermal sensor, from Wetec Pte
Ltd., was installed inside the triaxial cell to monitor the cell
water temperature. The accuracy of the submersible thermal
sensor was 7 0.2 1C. Both thermal sensors were calibrated
within the range of 20–30 1C. Correction to the total volume
change was performed using the recorded temperature.
67
3.2. Pressure and flushing lines
The water pressure and flushing lines of the pedestal were
connected to valves A and B, while the air pressure line of the
pedestal was connected to valve G (see Fig. 1). On the other
hand, the water pressure and flushing lines of the top cap were
connected to valves E and F, while the air pressure line of the
top cap was connected to valve C. The cell pressure line was
connected to valve D.
Three pressure transducers were attached to valve B for the
pore-water pressure measurement, valve C for the pore-air
pressure measurement and valve D for the cell pressure
measurement. All transducers are able to measure pressures
with an accuracy of 7 0.3 kPa and pressure up to 1000 kPa.
All pressure transducers were calibrated within the range of
0–600 kPa with intervals of 20 kPa.
Flushing of the apparatus was scheduled and performed in
order to remove the air bubbles which were trapped in the
water compartments during the testing. Due to the high pore-
air pressure in the specimen and the long duration of the test,
air can diffuse through the ceramic disks and reappear as air
bubbles in the water compartments. Air bubbles can affect the
accuracy of the pore-water pressure and volume change
measurements and impede the flow of water into and out of
the specimen (Fredlund and Rahardjo, 1993).
3.3. Volume change measurement system
Three DPVCs (pressure range: 0–2000 kPa; volumetric
capacity: 200 cc; maximum flow rate: 9  10  4 m3/h) from
GDS Instruments Limited, England, were used to apply and
control the water pressure and the confining pressure of the
specimen. The confining pressure was applied and controlled
using DPVC-2 through valve D during all stages. During the
68 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
drying and wetting processes, as well as during the shearing
stage, the water compartments in the modified pedestal and the
modified top cap were connected to DPVC-1 through valves A
and F. Water in the modified ceramic disks, the soil specimen
and the water compartments of the modified pedestal were in a
continuous condition. In other words, the pore-water pressure
and the volume in the specimen could be controlled and
measured directly by DPVC-1. During the unsaturated perme-
ability test, valves A and F were connected to DPVC-1 and
DPVC-3, respectively, in order to control the different water
pressures at the top and bottom of the specimen as well as to
accurately measure the flow of water through the specimen.
3.4. Shearing and load measurement system
A 50-kN compression machine from Tritech was used to
apply load at a constant strain rate on the specimen. The range
in strain rate that this compression machine can apply is
between 0.00001 mm/min and 9.9 mm/min. In this research,
0.0009 mm/min of strain rate was used to conduct the CD
triaxial test. This rate was used in the studies of Rahardjo et al.
(1995), Rahardjo et al. (2004) and Goh et al. (2010), for soils
with properties similar to those of the mixtures used in
this study.
A linear variable differential transformer (LVDT) was
attached to the loading ram for measuring the axial deforma-
tion of the specimen during shearing. The measuring range of
the LVDT (displacement LVDT) was 25 mm, while the
maximum axial strain of the soil specimen was set to 20%
(approximately 20 mm). The displacement LVDT was cali-
brated within the range of 0–20 mm with an interval of 1 mm.
The sensitivity of the displacement LVDT was 0.01 mm. On
the other hand, another LVDT, with a measuring range of
10 mm, was attached to the 2-kN load ring in order to
digitalize and record the applied load that was exerted on the
soil specimen during shearing automatically and continuously.
The load ring LVDT was well-calibrated within the range of
0–2 mm (approximately 0–1.5 kN) with an interval of
0.05 mm before being installed at the top of the triaxial cell.
The sensitivity of the load ring LVDT was 0.002 mm.
3.5. Data acquisition system
An 8-channel data logger and a personal computer were
used as part of the acquisition system. The pressure transdu-
cers, thermal sensors and LVDTs were connected to the data
logger that was connected to the personal computer, while
three DPVCs were directly connected to the personal computer
for data collection and monitoring. All readings were taken at
2-min intervals using the Triax 4.0 data acquisition program
(Toll, 1999).
4. Experimental study
Identical statically compacted sand–kaolin specimens were
used in this study. Specially mixed soils were chosen, since the
soil properties can be controlled and they are reproducible, to
ensure that all the results from different specimens are
comparable. The sand–kaolin mixtures were mixed from
kaolin, which is produced by Kaolin Malaysia SDN BHD
(Malaysia), and Ottawa sand, which is furnished by U.S. Silica
Company. Three different sand–kaolin mixtures, SK-5, SK-10
and SK-17, which comprise 15% sand and 85% kaolin, 35%
sand and 65% kaolin and 55% sand and 45% kaolin,
respectively, were used. The Ottawa sand was sieved and
categorized into different particle sizes, i.e., 600–1250 mm,
300–600 mm, 150–300 mm and 75–150 mm, for better control
of the grain-size distribution of the mixtures.
4.1. Soil properties
Basic property tests were performed according to the ASTM
standards. The SK-5, SK-10 and SK-17 are classified as MH,
CL and SC, respectively, in accordance with the Unified Soil
Classification System (ASTM D2487-93, 1997). The max-
imum dry densities, ρdmax, of SK-5, SK-10 and SK-17 are
1.50 Mg/m3, 1.67 Mg/m3 and 1.86 Mg/m3, respectively, while
the optimum water contents, wopt, of SK-5, SK-10 and SK-17
are 25%, 19% and 14%, respectively. Table 1 summarizes the
basic soil properties of SK-5, SK-10 and SK-17.
All sand–kaolin specimens were prepared at the maximum
dry density and the optimum water content. The triaxial
specimens, 100 mm in height and 50 mm in diameter, were
statically compacted in ten equal layers with a thickness of
10 mm. The SWCC specimens, 30 mm in height and 54 mm in
diameter, were statically compacted in three equal layers of
with a thickness of 10 mm.
4.2. Permeability measurement before shearing test
Each statically compacted triaxial specimen was placed on
the saturated ceramic disk, while filter papers were placed in
between the specimen and the porous disk in order to prevent
fine particles from being trapped inside the pores of the porous
disk. Vacuum grease was applied at the circumference of the
modified top cap and modified pedestal; and subsequently, the
rubber membrane was put on the specimen with O-rings being
mounted at the pedestal and the top cap to prevent leakage
during the test.
All specimens were saturated by applying a cell pressure
and a back pressure from DPVCs. The saturation of the
specimen at the beginning of the test is done to ensure that
all the specimens have uniform initial conditions and to reduce
the matric suction to a low value. A small net confining
pressure, i.e., 10 kPa, was maintained to prevent significant
swelling of the specimen until a pore-water pressure parameter
B of larger than 0.95 was achieved, as suggested by Head
(1986). The soil specimen was then isotropically consolidated
to the designated net confining pressure once the saturation
process had been completed.
Matric suction was applied to the specimens using the axis
translation technique after the completion of consolidation.
The drying and wetting of each specimen were conducted by
following the designated incremental and decremental steps of
 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73 69

Table 1
Soil properties of SK-5, SK-10 and SK-17.

Characteristics of sand–kaolin mixture SK-5 SK-10 SK-17

Specific gravity, Gs 2.66 2.67 2.66

Maximum dry density, ρdmax (Mg/m3) 1.50 1.67 1.86
Optimum water content, wopt (%) 25 19 14
Liquid limit, LL (%) 51.1 41.9 29.3
Plastic limit, PL (%) 31.2 23.3 16.2
Plasticity Index, IP (%) 19.9 18.6 13.1
Composition:
Ottawa sand (by dry weight) (%) 15 35 55
Kaolin (by dry weight) (%) 85 65 45
Grain size distribution:
Sand ( 475 μm) (%) 15.0 35.0 55.0
Silt (2–75 μm) (%) 73.6 44.5 39.0
Clay ( o2 μm) (%) 11.4 20.5 6.0
D60 (μm) 12.5 10.0 203.0
D50 (μm) 10.5 6.0 130.0
D30 (μm) 6.0 2.7 12.0
D10 (μm) 1.7 1.3 3.7
Cu (D60/D10) 7.353 7.692 54.865
Cc ((D30)2/(D60D10)) 1.694 0.561 0.192
Unified Soil Classification System, USCS MH CL SC
ks (m/s) @ (σ uw)¼50 kPa 6.02  10  9 7.67  10  9 1.63  10  8
c0 (kPa) 29.5 8.5 8.2
ϕ0 (1) 26.8 26.9 33.2

matric suction. During drying, once the change in water 10000 4

Water flow (UP-50-1D100-III)
volume had reached equilibrium, the matric suction was then
Volume of water flow,Qw (mm3)

Water flow (UP-50-1W100-III)

Total volumetric strain, εv (%)

increased to the next step by decreasing the pore-water
pressure while keeping the confining pressure and the pore-
air pressure constant. Equilibrium was achieved for the matric
suction when the excess pore-water pressure had fully dis-
sipated and the change in water volume was less than 0.04%
per day, as suggested by Sivakumar and Wheeler (1993). Upon
reaching the maximum matric suction, i.e., 440 kPa, the
wetting stage of the specimen was conducted by increasing
the pore-water pressure while maintaining constant confining
and pore-air pressures. The water compartments were flushed
to remove diffused air before the matric suction was changed.
The unsaturated permeability test was conducted after the
designated number of cycles of drying and wetting processes
and matric suction value had been reached. DPVC-3 was
connected to valve F, while DPVC-1 remained connected to
valve A. Two DPVCs were used to apply and control different
water pressures to the top and bottom of the specimen in order
to create a hydraulic head gradient for the water flow in the
specimen. The applied water pressure from DPVC-1 was a
pressure of 15 kPa higher than the pore-water pressure in the
specimen, while the applied water pressure from DPVC-3 was
a pressure of 15 kPa lower than the pore-water pressure in the
specimen. Therefore, a difference in water pressure head of
30 kPa was created during the unsaturated permeability test.
The same difference in water pressure head was used in all the
unsaturated permeability tests in this research.
An upward water flow through the soil specimen was
established under the application of the difference in water
pressure head. The inflow and outflow of water passing
Total volumetric strain (UP-50-1D100-III)
7500 2
Total volumetric strain (UP-50-1W100-III)
5000 0
y = 4.54x + 185.28
R² = 1.00 y = 1.40x - 126.05
2500 R² = 1.00 -2
0 -4
0 400 800 1200 1600
Elapsed time, t (min)
Fig. 6. Water flow measurements and total volumetric strain during perme-
ability tests at matric suction of 100 kPa on drying and wetting paths of
first cycle.
through the specimen within a given time were measured
continuously. A graph of water volume against time was
plotted and the flow rate of water, qt (m3/s), was calculated
(see Fig. 6). A steady-state condition was established where
the inflow and outflow rates were approximately the same. The
flow rate was found to vary in the beginning, but it became
constant after the difference in water pressure was evenly
distributed along the soil specimen (see Fig. 6). The test was
stopped when a constant water flow rate for a given period of
time was achieved. Valve F was then reconnected to DPVC-1.
The water pressure applied to the specimen through valves A
and F was adjusted back to the water pressure before the
permeability test. The matric suction was then increased or
70 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
decreased to the next targeted matric suction. The same
procedures were repeated to measure the permeability of the
soil under the targeted matric suction. Upon completion of
the permeability measurements, a CD triaxial test was then
performed on the same specimen at the designated matric
suction. More details on the unsaturated permeability measure-
ments and CD trixial tests under multiple cycles of drying and
wetting are presented in Goh (2012).
4.3. SWCC test
The SWCCs of the sand–kaolin mixtures under zero net
confining pressure were obtained using a pressure plate
apparatus. After saturation of the specimen had been com-
pleted, a matric suction of 1 kPa, with air pressure of 10 kPa
and water pressure of 9 kPa (equivalent to 90 cm water head),
was first applied to all specimens. The water pressure remained
unchanged throughout the SWCC test. The measurements of
the drying and wetting SWCCs were performed by following
the designated incremental and decremental steps of matric
suction. The mass of the specimens was recorded daily until it
had reached equilibrium before increasing the matric suction to
the next value. In order to provide good contact between the
saturated ceramic disk and the specimen, a little de-aired water
was sprayed on the surface of the ceramic disk. The water
compartment was flushed every time after the readings were
taken to ensure that the ceramic disk was in a saturated
condition and that the soil specimen was in contact with the
water compartment. After the first cycle of drying and wetting
processes had been completed, the SWCC tests were continued
with the second and third cycles of the drying and wetting
processes. At the end of the test, the final water contents of the
specimens were determined.
4.4. Determination of water coefficient of permeability
The permeability of an unsaturated soil can be calculated
using Darcy's law (Childs and Collis-George, 1950; Fredlund
and Rahardjo, 1993), which has the following relationships:
h
qt ¼ vA ¼ kw iA ¼ k w A ð1Þ
L Multiple cycles of drying and wetting SWCCs of SK-5, SK-
where, qt is the total flow rate through the cross-sectional area,
v is the flow velocity, A is the cross-sectional area, kw is the
Darcy coefficient of permeability (water coefficient of perme-
ability), i is the hydraulic gradient, i.e., i is the h/L, h is the
head loss, and L is the length of sample.
In the analysis of the permeability test results obtained from
the improved triaxial apparatus, the impedance of the ceramic
disks was considered. The permeability (or water coefficient of
permeability) of the soil was determined by considering the
arrangement of the disk–soil–disk as three different layers
(Agus et al., 2003). The flow velocity, v, is the same for every
layer, while the total head loss, hT, is the summation of the
head loss in each layer (top disk, soil specimen and bottom
disk). The flow velocity and head loss for each layer have the
following relationships:
v ¼ v t ¼ vs ¼ vb ð2Þ
hT ¼ ht þ hs þ hb ð3Þ
where, vt is the flow velocity through the top ceramic disk, vs is
the flow velocity through the soil specimen, vb is the flow
velocity through the bottom ceramic disk, ht is the head loss
along the top ceramic disk, hs is the head loss along the soil
specimen, hb is the head loss along the bottom ceramic disk.
By substituting Eq. 1 into Eq. 3, the following equation is
obtained:
vT LT vt Lt vs Ls vb Lb
¼ þ þ ð4Þ
kT kt kw kb
where, kw is the water coefficient of permeability of soil
specimen, kT is the water coefficient of permeability of the
disk–soil–disk system, kt is the water coefficient of perme-
ability of top ceramic disk, kb is the water coefficient of
permeability of bottom ceramic disk, Ls is the height of the soil
specimen, Lt is the thickness of the top ceramic disk, Lb is the
thickness of the bottom ceramic disk, and LT is the height of
the disk–soil–disk system, i.e. LT ¼ Ls þ Lt þ Lb.
Since v is the same for each layer, Eq. 4 can be rearranged
and written as follows:
L
kw ¼  s  ð5Þ
ðLT =kT Þ ðLt =kt Þþ ðLb =kb Þ
Therefore, the water coefficient of permeability for the soil
specimen can be calculated using Eq. 5. Meanwhile, the water
coefficient of permeability for the disk–soil–disk system, kT,
which was obtained from the permeability test, was determined
using the following equation:
Q
kT ¼ w ð6Þ
iAt
where, Qw is the volume of water flow through the soil
specimen, and t is the elapsed time.
5. Results and discussions
5.1. Drying and wetting SWCCs
10 and SK-17 were obtained in this study. Drying and wetting
SWCCs were fitted using the Fredlund and Xing (1994)
equation with the correction factor, C(ψ), taken as 1, as
suggested by Leong and Rahardjo (1997b). Table 2 sum-
marizes the SWCC parameters of SK-5, SK-10 and SK-17.
The AEVs of all the sand–kaolin specimens were determined
using the technique described in Goh et al. (2011). One set of
the primary drying and wetting SWCCs of the sand-kaolin
specimens, i.e., SK-5, along with the scanning curves defined
during drying and wetting under zero net confining pressure, is
illustrated in Fig. 7.
In general, the drying and wetting SWCCs of all the sand–
kaolin mixtures were found to be distinctly different, where the
volumetric water contents of the soils at the drying SWCCs are
 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73 71

Table 2 1.0E-10
Initial ρd = 1.50 Mg/m3
Summary of the SWCC results for SK-5, SK-10 and SK-17 under zero net
confining pressure from pressure plate tests. SK-5
1.0E-11
SK-5 SK-10 SK-17

qt/i (m3s-1)
First Drying SWCC 1.0E-12
Saturated volumetric water content, θs 0.503 0.466 0.359
AEV, (ua uw)b (kPa) 30.0 41.0 26.0
Residual matric suction, (ua  uw)r (kPa) 1000.0 1200.0 1000.0 1.0E-13
Fredlund and Xing (1994) fitting parameters UP-50-1D100-I
ad (kPa) 68.946 81.829 64.757
UP-50-1W300-I
nd 1.412 1.578 1.422 1.0E-14
md 0.793 0.363 0.711 10 20 30 40 50
First Wetting SWCC Hydraulic gradient, i
Wetting saturated volumetric water content, θsw 0.441 0.452 0.325
Wetting saturated point, (ua  uw)bw (kPa) 11.0 13.0 6.2 Fig. 8. Relationship between (qt/i) and hydraulic gradient, i.
Water-entry value, (ua uw)w (kPa) 550.0 730.0 340.0
Fredlund and Xing (1994) fitting parameters
aw (kPa) 22.180 21.123 11.150 1.0E-08
nw 1.284 1.282 1.493 Initial ρd = 1.50 Mg/m3
mw 0.520 0.266 0.400 SK-5
1.0E-09
drying
kw (ms-1)
0.6
Initial d = 1.50 Mg/m3 Experiment (1st drying) 1.0E-10 Measured @ 1st drying
Experiment (1st wetting)
w

0.5 Measured @ 1st wetting wetting

Experiment (2nd drying)
Volumetric water content,

Measured @ 2nd drying

0.4
0.3
0.2
0.1
SK-5
0.0
0.1 1 10
Matric suction, (ua- uw) (kPa)
Fig. 7. Multi-cycled SWCCs of SK-5 specimen under zero net confining
pressure.
higher than those at the wetting SWCCs in all cycles of drying
and wetting. All the sand–kaolin mixtures exhibit hysteresis
during drying and wetting. A detailed description and discus-
sion of the characteristics of multiple cycles of drying and
wetting SWCCs are presented in Goh et al., (2010, 2014).
5.2. Drying and wetting permeability of soil
The average coefficient of permeability of the modified ceramic
disk with a groove design is 5.02  10  10 m/s. The volume of
water passing through the specimen within a given time was
measured and plotted as the volume of the water flow, Qw, versus
the time elapsed, t, as illustrated in Fig. 6. The flow rate through
the cross-sectional area, qt, was calculated as the slope of the plot
after the constant flow of water was observed (Fig. 6). On the other
hand, as shown in Fig. 6, the total volume of the specimens
generally remained unchanged and the water volume was not
affected by the room temperature during the test. The permeability
test was stopped after the water had flowed through the specimen
steadily for a given period of time.
Experiment (2nd wetting)
1.0E-11 Measured @ 2nd wetting
Experiment (3rd drying)
Experiment (3rd wetting) Fitted @ 1st drying
Fitted @ 1st wetting
1st drying fitting
1.0E-12
1st wetting fitting 0.1 1 10 100 1000 10000 100000
2nd drying fitting
Matric suction, (ua- uw) (kPa)
2nd wetting fitting
3rd drying fitting Fig. 9. Measured water coefficient of permeability of SK-5 specimens and
3rd wetting fitting fitting curves using Leong and Rahardjo (1997b) permeability function.
100 1000 10000 100000
A suitable difference in water pressure head, used in the
unsaturated permeability measurements, was determined by vary-
ing the applied difference in water pressure head of the perme-
ability tests at a given matric suction. Fig. 8 illustrates the
relationship between qt/i and hydraulic gradient, i. As shown in
Fig. 8, when the hydraulic gradient is approximately equal to or
higher than 28, no significant change in qt/i is observed with the
variation in the applied hydraulic gradient. Therefore, the hydraulic
gradient of approximately 28 (equivalent to approximately water
pressure head difference of 30 kPa) was adopted in this study for
all the unsaturated permeability tests.
Series of permeability tests were conducted to study the
permeability characteristics of the three different sand–kaolin
mixtures on multiple cycles of drying and wetting paths. The
unsaturated permeability on the drying and wetting paths was
fitted with Leong and Rahardjo (1997a)'s permeability func-
tion. Figs. 9–11 present the permeability characteristics of SK-
5, SK-10 and SK-17 specimens, respectively, on the drying
and wetting paths of different cycles.
The permeability of the sand–kaolin mixtures on the drying
and wetting paths are shown to be different. The permeability
on the drying paths is generally higher than those on the
wetting paths for all cycles of drying and wetting. The
differences between the permeability on the first cycle drying
72 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73

1.0E-08 Leong and Rahardjo (1997a)'s permeability function, for fitting

Initial ρd = 1.67 Mg/m3
the coefficient of permeability of SK-5 on the 1st cycle drying
SK-10
and wetting paths, are 4.230 and 4.483, respectively. The fitting
1.0E-09
drying parameters, p, for fitting the water coefficient of permeability of
SK-10 on the 1st cycle drying and wetting paths are 10.786 and
kw (ms-1)

1.0E-10 Measured @ 1st drying 11.123, respectively. On the other hand, the fitting parameters, p,
Measured @ 1st wetting wetting for fitting the water coefficient of permeability of SK-17 on the
Measured @ 2nd drying 1st cycle drying and wetting paths are 6.920 and 6.923,
1.0E-11 Measured @ 2nd wetting respectively. These values of the fitting parameters agree with
Fitted @ 1st drying
the findings reported in Fredlund et al. (2001).
Fitted @ 1st wetting
1.0E-12 After the unsaturated permeability measurements, the tests
0.1 1 10 100 1000 10000 100000 were continued to the unsaturated CD triaxial tests. The CD
Matric suction, (ua- uw) (kPa) triaxial tests of SK-5, SK-10 and SK-17 on multiple cycles of
Fig. 10. Measured water coefficient of permeability of SK-10 specimens and drying and wetting paths were conducted and reported in Goh
fitting curves using Leong and Rahardjo (1997b) permeability function. et al. (2010, 2014).
The characteristics of the unsaturated permeability of the
1.0E-07 sand–kaolin mixtures are presented in this study and agree
Initial ρd = 1.86 Mg/m3
SK-17
with those in the literature. In addition, unsaturated shear
1.0E-08 strengths on multiple cycles of drying and wetting paths were
also obtained and presented in Goh et al. (2010, 2014) using
drying
1.0E-09 Measured @ 1st drying the same specimen as that used for the permeability test. The
kw (ms-1)

Measured @ 1st wetting modified triaxial apparatus presented in this study reduces
Measured @ 2nd drying
1.0E-10 the effort and time needed to determine the permeability and
Measured @ 2nd wetting
Measured @ 3rd drying the shear strength of a specimen under multiple cycles of
wetting
1.0E-11 Measured @ 3rd wetting drying and wetting. Furthermore, the test results are more
Fitted @ 1st drying consistent as both measurements were conducted on the same
Fitted @ 1st wetting
1.0E-12 specimen. In addition, both the permeability and the shear
0.1 1 10 100 1000 10000 strength of soils under multiple cycles of drying and wetting
Matric suction, (ua- uw) (kPa) paths are valuable and useful for understanding the character-
Fig. 11. Measured water coefficient of permeability of SK-17 specimens and
istics of unsaturated soils.
fitting curves using Leong and Rahardjo (1997b) permeability function.
6. Conclusions

and wetting paths were found to be more significant than the
differences between the permeability on the subsequent cycles
of drying and wetting paths (see Figs. 9–11). It was noticed
that the characteristics of permeability under the drying and
wetting of a soil are analogous to the characteristics of the
drying and wetting SWCCs of the soil. This is attributed to
the fact that water in unsaturated soil can only flow through the
soil voids that are filled with water in a continuous path
(Fredlund and Rahardjo, 1993). As the matric suction of a soil
increases (drying process), the water content in the soil
decreases, causing a decrease in the permeability of the soil.
When the matric suction of a soil decreases (wetting process),
the water content in the soil increases, causing an increase in
the permeability of the soil. Therefore, the permeability of the
sand–kaolin specimens is different during drying and wetting
due to hysteresis. Similar findings on the permeability char-
acteristics of soil, on the first cycle of drying and wetting
paths, were reported in other studies, e.g., Fredlund and
Rahardjo (1993), Gan and Fredlund (2000) and Agus et al.
(2003). Likewise, the different water contents of a soil at the
same matric suction on different cycles of drying and wetting
paths, cause the permeability of the soil to be different.
The permeability results were fitted using Leong and Rahardjo
(1997a)'s permeability function. The fitting parameters, p, used in
A modified triaxial apparatus has been developed for the
direct measurements of the permeability and the shear strength
of unsaturated soil. The apparatus has been shown to be able to
measure the permeability and the shear strength of the same
specimen for three different sand–kaolin mixtures in multiple
cycles of drying and wetting paths. The water volume change,
the total volume change, the pore-air, the pore-water and the
cell pressures, as well as the room and water temperatures,
were also measured and recorded using the modified triaxial
apparatus during the tests. Therefore, the time and effort
needed to determine more consistent results for the perme-
ability and the shear strength of a soil under multiple cycles of
drying and wetting were reduced as both measurements were
conducted on the same specimen.
The permeability of sand–kaolin specimens on the drying
paths is shown to be higher than that of the sand–kaolin
specimens on the wetting paths. This could be attributed to the
hysteresis effects on the soils during drying and wetting. The
differences between the permeability in the first cycle drying
and wetting paths are shown to be more significant than those
on the subsequent cycles of drying and wetting paths. The
results show that the characteristics of the permeability of
sand–kaolin specimens are similar to the characteristics of the
drying and wetting SWCCs of the sand–kaolin specimens.
 S.G. Goh et al. / Soils and Foundations 55 (2015) 63–73
This is due to the fact that the water in unsaturated soil can
only flow through the soil voids that are filled with water in a
continuous path.
Acknowledgments
This study was supported by a research grant (M4060100)
from a collaborative project between the Housing and
Development Board (HDB), Rep. of Singapore and
Nanyang Technological University (NTU), Rep. of Singa-
pore. The first author would also like to acknowledge the
research scholarship provided by NTU. The assistance and
advice from the faculty and staff of the School of Civil and
Environmental Engineering, NTU in this study are highly
appreciated.
References
Agus, S.S., Leong, E.C., Rahardjo, H., 2003. A flexible wall permeability for
measurements of water and air coefficient of permeability of residual soil.
Can. Geotech. J. 40, 559–574.
ASTM, 1997. Standard Classification of Soils for Engineering Purposes
(Unified Soil Classification System)”, Annual Books of ASTM Standard
D2487-93. vol. 04–08. ASTM, Philadelphia217–227 (Section 4).
Barden, C.H., Pavlakis, G., 1971. Air and water permeability of compacted
unsaturated cohesive soils. J. Soil Sci. 22, 302–318.
Benson, C.H., Gribb, M.M., 1997. Measuring unsaturated hydraulic conduc-
tivity in the laboratory and field. Unsaturated Soil Engineering Practise,
Geotechnical Special Publication, vol. 68, pp. 113–168.
Brooks, R.H., Corey, A.T., 1964. Hydraulic Properties of Porous Media,
Hydrology Paper, No. 3, Colorado State University, Fort Collins, Colo.
Childs, E.C., Collis-George, G.N., 1950. The permeability of porous materials.
Proc. R. Soc. Lond. Ser. A 201, 392–405.
Chiu, T., Shackelford, C.D., 1998. Unsaturated hydraulic conductivity of
compacted sand–kaolin mixtures. J. Geotech. Geoenvironmental Eng.,
ASCE 124 (2), 160–170.
Fredlund, D.G., 2006. Unsaturated soil mechanics in engineering practise. J.
Geotech. Geoenvironmental Eng. 132 (3), 286–321.
Fredlund, D.G., Fredlund, M.D., Zakerzadeh, 2001. Predicting the permeability
function for unsaturated soils, Clay Science for Engineering. Balkema,
Rotterdam215–222.
Fredlund, D.G., Rahardjo, H., 1993. Soil Mechanics for Unsaturated Soils.
John Wiley and Sons Inc., New York.
Fredlund, D.G., Xing, A., 1994. Equations for the soil–water characteristic
curve. Can. Geotech. J. 31 (3), 521–532.
Fredlund, D.G., Xing, A., Huang, S., 1994. Predicting the permeability
function for unsaturated soils using the soil–water characteristic curve.
Can. Geotech. J. 31 (3), 533–546.
Gallage, C.P.K., Kodikara, J., Uchimura, T., 2013. Laboratory measurement of
hydraulic conductivity functions of two unsaturated sandy soils during
drying and wetting processes. Soils Found. 53 (3), 417–430.
Gan, J.K.M., Fredlund, D.G., 2000, A new laboratory method for the
measurement of unsaturated coefficient of permeability of soils. In:
Proceedings of the Asian Conference on Unsaturated Soils – From Theory
to Practise, Singapore, 18–19 May, pp. 381–386.
Goh, S.G., 2012. Hysteresis Effects on Mechanical Behaviour of Unsaturated
Soils (Doctor of Philosophy Thesis). Nanyang Technological University.
Singapore.
Goh, S.G., Rahardjo, H., Leong, E.C., 2011. Closure to shear strength
equations for unsaturated soil under drying and wetting. J. Geotech.
Geoenvironmental Eng. 137 (12), 1313–1315.
73
Goh, S.G., Rahardjo, H., Leong, E.C., 2014. Shear strength of unsaturated soils
under multiple drying-wetting cycles. J. Geotech. Geoenvironment Eng.
140 (2), 06013001.
Goh, S.G., Rahardjo, H., Leong, E.C., 2010. Shear strength equations for
unsaturated soil under drying and wetting. J. Geotech. Geoenvironmental
Eng. 136 (4), 594–606.
Han, K.K., Rahardjo, H., Broms, B.B., 1995. Effect of hysteresis on the shear
strength of a residual soil. In: Proceedings of the First International
Conference on Unsaturated Soil, vol. 2, Balkema, Rotterdam, The Nether-
lands, pp. 499–504.
Head, K.H., 1986. Manual of Soil Laboratory Testing, vol. 03. John Wiley and
Sons Inc., New York, 925–945.
Hilf, J.W., 1956. An Investigation of Pore-water Pressure in Compacted
Cohesive Soils (Doctor of Philosophy Dissertation, Tech. Memo. no. 654).
U. S. Department of the Interior, Bureau of Reclamation, Design and
Construction Division, Denver, CO.
Holtz, R.D., Kovacs, W.D., 1981. An Introduction to Geotechnical Engineer-
ing. Prentice-Hall, New Jersey.
Huang, S., Fredlund, D.G., Barbour, S.L., 1998. Measurement of the
coefficient of permeability for a deformable unsaturated soil using a
triaxial permeameter. Can. Geotech. J. 35, 426–432.
Klute, A., 1965. Laboratory measurement of hydraulic conductivity of
unsaturated soils. In: Black, C.A., Evans, D.D., White, J.L., Esminger,
L.E., Clark, F.E. (Eds.), Method of Soil Analysis, vol. 1. Merican Society
of Agronomy, Madison, Wis, pp. 253–261 (Monograph 9).
Krisdani, H., Rahardjo, H., Leong, E.C., 2009. Use of instantaneous profile and
statistical methods to determine permeability functions of unsaturated soils.
Can. Geotech. J. 46, 869–874.
Kunze, R.J., Uehera, G., Graham, K., 1968. Factors important in the calculation of
hydraulic conductivity. Proc. Soil Sci. Soc. Am. 32, 760–765.
Leong, E.C., Rahardjo, H., 1997a. Permeability functions for unsaturated soils.
J. Geotech. Geoenvironmental Eng. 123 (12), 1118–1126.
Leong, E.C., Rahardjo, H., 1997b. Review of soil–water characteristic curve
equations. J. Geotech. Geoenvironmental Eng. 123 (12), 1106–1117.
Lu, N., Wayllace, A., Carrera, J., Likos, W.J., 2006. Constant flow method for
concurrently measuring soil–water characteristic curve and hydraulic
conductivity function. Geotech. Test. J. 29 (3), 1–12.
Moncada, M.P.H., Campos, T.M.P., 2010. A new permeameter for the
determination of the unsaturated coefficient of permeability. In: Proceed-
ings of the Firth International Conference of Unsaturated Soil, UNSAT
2010, Barcelona, Spain, 6–8 September, pp. 701–706.
Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of
unsaturated porous media. Water Resour. Res. 12, 513–522.
Rahardjo, H., Lim, T.T., Chang, M.F., Fredlund, D.G., 1995. Shear
strength characteristics of a residual soil in Singapore. Can. Geotech.
J. 32, 60–77.
Rahardjo, H., Ong, B.H., Leong, E.C., 2004. Shear strength of a compacted
residual soil from consolidated drained and constant water content triaxial
tests. Can. Geotech. J. 41, 421–436.
Richards, L.A., 1931. Capillary conducting of liquids through porous medium.
Physics 1, 318–333.
Sivakumar, V., Wheeler, S.J., 1993. Elasto-plastic volume change of unsatu-
rated compacted clay. In: Proceedings of Sessions Sponsored by the
Subcommittee on Shallow Foundations of the Geotechnical Engineering of
the American Society of Civil Engineer, Dallas, Texas.
Toll, D.G., 1999. A data acquisition and control system for geotechnical
testing. In: Kumar, B., Topping, B.H.V. (Eds.), Computing Developments
in Civil and Structural Engineering. Civil-Comp, Edinburgh, UK, pp.
237–242.
van Genuchten, M.T., 1980. A closed form equation predicting the hydraulic
conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892–898.
Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., 1996. The relationship between
the soil–water characteristic curve and the unsaturated shear strength of a
compacted glacial till. Geotech. Test. J. 19 (3), 259–268.
Vanapalli, S.K., Garga, V.K., Brisson, P., 2007. A modified permeameter for
determination of unsaturated coefficient of permeability. Geotech. Geol.
Eng. 25, 191–202.