Soils and Foundations 2016;56(4):732–744

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

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

Technical Paper

Effect of saturation on liquefaction resistance of iron ore fines

and two sandy soils
Hailong Wanga,n, Junichi Kosekib, Takeshi Satoc, Gabriele Chiarod, Jaylord Tan Tianb
a
OYO Corporation, Formerly Research fellow of Institute of Industrial Science, The University of Tokyo, Japan
b
Department of Civil Engineering, The University of Tokyo, Japan
c
Integrated Geotechnology Institute Ltd., Japan
d
Department of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand
Received 19 April 2015; received in revised form 3 March 2016; accepted 20 May 2016
Available online 18 August 2016

Abstract

Over the past several years, the International Maritime Organization (IMO) has become increasingly concerned about the liquefaction of
unsaturated solid bulk cargo (e.g. iron ore fines) during maritime transportation. This concern has arisen due to several accidents including the
capsizing of vessels. In addition, although the resistance against liquefaction of ordinary unsaturated soils is higher than for saturated soils,
possible key parameters governing the liquefaction resistance of unsaturated soils (RL,unsat) have not yet been clearly identified. Therefore, in this
study, undrained cyclic loading tests of saturated and unsaturated iron ore fines and two sandy soils were conducted using a triaxial apparatus to
reveal the liquefaction behavior of iron ore fines and to find the key parameters governing RL,unsat. Through comparisons, it was found that the
liquefaction behavior of iron ore fines is similar to that of sandy soils. The degree of saturation and potential volumetric strain, which have been
proposed as the governing parameters of RL,unsat, were examined based on experimental data obtained in this study and by other researchers. It
was shown that neither of the two parameters correlate with the liquefaction resistance ratio (LRR), a ratio of RL,unsat to the liquefaction resistance
of the saturated soils (RL,sat) with a unique relationship, especially when considering soils with considerable fines content. Following the concept
of potential volumetric strain, which considers the compressibility of pore air in the unsaturated soils, volumetric expansion due to the reduction
in confining pressure during cyclic loading is further considered, and a new index, the volumetric strain ratio (Rv) is proposed in this study.
According to the experimental data obtained in this study, Rv exhibits a much better correlation with LRR than the two former parameters.
& 2016 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Iron ore fines; Unsaturated soils; Liquefaction; Volumetric strain ratio; Triaxial test

1. Introduction transportation (Isacson, 2010a, b; LPC, 2011; Gard, 2012 etc.).

Awareness was recently raised among the International
Maritime Organization (IMO) due to the substantial losses
caused reportedly by liquefaction of solid bulk cargo e.g. iron
ore fines, nickel ore, sinter feed bauxite etc. during maritime
n
Corresponding author.
E-mail address: whlxy2002@gmail.com (H. Wang).
Peer review under responsibility of The Japanese Geotechnical Society.
Though such solid bulk cargo is usually loaded into the vessels
under the unsaturated condition, the capsizing of vessels due to
the liquefaction of such cargo subjected to ocean wave motion,
for example, cannot be prevented in some cases due to a lack
of knowledge about the liquefaction of these materials.
In addition, in earthquake-prone countries, such as Japan,
the liquefaction resistance of unsaturated soils is a much-
researched topic in the field of geotechnical engineering. A
great number of liquefaction sites were reported in the Tohoku

http://dx.doi.org/10.1016/j.sandf.2016.07.013
0038-0806/& 2016 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744 733

Nomenclature qcyclic single amplitude of vertical cyclic stress

RL resistance against liquefaction
CSR cyclic stress ratio (¼ qcyclic/2s00 ) RL,sat RL of the saturated soil
Dc compaction degree before applying cyclic loading RL,unsat RL of the unsaturated soil
Dr relative density before applying cyclic loading Rv volumetric strain ratio (¼ εν,air/εν,s0 )
e void ratio S matric suction ( ¼ ua  uw)
emax maximum void ratio Sr degree of saturation
emin minimum void ratio ua pore air pressure
DA ¼ 5% 5% double amplitude of axial strain uw pore water pressure
DA ¼ 5% criterion criterion for liquefaction based on the wopt optimum water content
condition of DA ¼ 5% Δu excess pore pressure, denoting either Δua or Δuw
Fc fines content Δua excess pore air pressure
Gs specific gravity Δuw excess pore water pressure
IMO International Maritime Organization Δu ¼ 0.9s00 a condition when Δu equals 90% s00
LRR liquefaction resistance ratio, a ratio of RL,unsat to Δu ¼ 0.9s00 criterion criterion for liquefaction based on the
RL,sat condition of Δu¼ 0.9s00
LRRDA ¼ 5% LRR, in which RL is determined based on εν,air volumetric strain caused by pore air compression
DA ¼ 5% criterion εν,0.9air volumetric strain caused by pore air compression
LRRΔu ¼ 0.9s00 LRR, in which RL is determined based on when Δua ¼ 0.9s00
Δu ¼ 0.9s00 criterion ε*ν,air potential volumetric strain, namely, εν,air when
NDA ¼ 5% number of cycles to trigger liquefaction based on Δua ¼ s00
DA ¼ 5% criterion εν,s0 volumetric strain caused by reduction of s0
NΔu ¼ 0.9s00 number of cycles to trigger liquefaction based εν,0.9s0 εν,s0 due to reduction of s0 by 90%
on Δu¼ 0.9s00 criterion εν,τ volumetric strain caused by contraction (or nega-
p total mean principal stress tive dilatancy)
p-constant condition a condition to maintain p constant ρdmax maximum dry density obtained from
during applying vertical cyclic loading compaction test
p ua net mean principal stress of unsaturated soil s0 confining pressure
p uw effective mean principal stress of unsaturated soil s 00 initial confining pressure
p u effective mean principal stress denoting either sh cell pressure
p  ua or p  uw

district (Yamaguchi et al., 2012) and the Kanto district,
including the Tokyo Bay area (Yasuda et al., 2012; Towhata
et al., 2014) after the Great East Japan Earthquake Disaster in
2011. Severe damage to houses (e.g. tilting, cracking etc.),
roads and lifeline facilities was incurred in the liquefaction
affected areas. As a possible countermeasure against liquefac-
tion, which can be applied to the narrowly constructed
residential areas, ground desaturation by either dewatering
the ground or the injection of micro air bubbles has been given
particular attention in recent years due to its low cost
(Okamura et al., 2006; NILIM, 2011, 2013).
To the authors' knowledge, starting from the early work by
Sherif et al. (1977), it has been repeatedly shown that soils
under the unsaturated condition show higher resistance against
liquefaction than those under the saturated condition. How-
ever, the governing parameters determining the liquefaction
resistance of unsaturated soils remain unclear. Attempts were
made to determine the role played by the degree of saturation
Sr (Yoshimi et al., 1989; Goto and Shamoto, 2002), pore-
pressure coefficient B value (Yoshimi et al., 1989; Unno et al.,
2008; Arab et al., 2011), elastic wave velocity (Huang et al.,
1999; Ishihara et al., 2001; Tsukamoto et al., 2002; Yang,
2002; Yang et al., 2004) and potential volumetric strain
(Okamura and Soga, 2006). It seems that the first three
parameters can work well only for a given soil under similar
test conditions, and the last parameter needs to be verified on
soils with considerable fines content.
In this study, undrained cyclic loading tests were conducted
on one type of iron ore fines and two types of sandy soils by
employing a stress-controlled triaxial apparatus. The liquefac-
tion behaviors of iron ore fines were compared with those of
ordinary sandy soils. Based on the test data obtained in this
study and that reported in the literature, two parameters
proposed to be the governing factors of liquefaction resistance
of the unsaturated soil are discussed. Finally, following
relevant findings from past studies, a new governing index is
proposed.
2. Apparatus
Both saturated and unsaturated specimens were tested on the
same apparatus, while extra components were added to the
apparatus for the unsaturated tests as schematically illustrated
in Fig. 1. Sinusoidal vertical cyclic loading with frequency of
0.1 Hz is applied by a double action cylinder controlled by a
function generator and an E/P regulator. There is another set of
734 H. Wang et al. / Soils and Foundations 56 (2016) 732–744

Fig. 1. Layout of the apparatus employed for undrained cyclic loading tests on unsaturated specimens.

function generator and E/P regulator to control the cell

pressure (sh) in order to maintain the total mean principal
stress (p) constant (i.e. p-constant condition: Huang et al.,
1999; and Tsukamoto et al., 2002). By introducing the
membrane filter technique (Nishimura et al., 2012; Wang
et al., 2014) to the pedestal instead of the traditional ceramic
disk for unsaturated specimens, the test duration can be
reduced and the pore water (uw) can be measured promptly
by the connected pressure transducer. The pore air pressure
(ua) is measured by another pressure transducer connected to
the top cap, on which a hydrophobic filter is glued. Volume
change of the unsaturated specimens is monitored by the inner
cell system. A detailed description of the apparatus can be
found in Wang et al. (2015).

3. Testing programs Fig. 2. Particle size distribution of tested soils.

As testing materials, Iron ore fines type B (IOF-B), Toyoura
sand and Inagi sand were used. Their gradation curves are
shown in Fig. 2. Iron ore fines (IOF) is tentatively defined by
the IMO as iron ore containing 10% or more particles less than
1 mm and 50% or more particles less than 10 mm (IMO,
2013). IOF-B (fines content Fc ¼ 24%) used in this study and
Inagi sand (Fc ¼ 30%) can be classified as non-plastic sands.
Their specific gravity (Gs), maximum dry density (ρdmax) and
optimum water content (wopt) are 4.444, 2.79 g/cm3 and 12%
for IOF-B, and 2.656, 1.66 g/cm3 and 20% for Inagi sand,
respectively (the compaction tests were conducted following
the method A of JIS A 1210 with the similar compaction
energy to the standard Proctor method, where JIS stands for
the Japanese Industrial Standard). On the other hand, Toyoura
sand is a clean sand with negligible fines content (Fc o0.5%).
Its maximum void ratio (emax), minimum void ratio (emin) and
Gs are 0.989, 0.611 and 2.652, respectively (emax and emin were
measured based on JIS A 1224).
All the specimens were 50 mm in diameter and 100 mm in
height. The undrained cyclic loading test conditions are
described in Table 1. The specimens of IOF-B and Inagi sand
were molded by one-dimensional compression with an initial
water content of 12% and 22%, respectively. The double
vacuum method (Ampadu and Tatsuoka, 1993) was applied to
saturate the specimens. The unsaturated specimens of IOF-B
with Dc of 93%, s00 of 100 kPa and Sr of 72% were tested by
keeping the initial water content (12%) and all the other
unsaturated specimens of IOF-B were flushed by distilled
water (without de-airing) prior to the cyclic loading. For the
unsaturated specimens of Inagi sand, a pre-determined amount
of water was added from the top of the specimens to obtain the
desired Sr and curing of the specimens was undertaken for
approximately 10 h in the mold in order to achieve a uniform
water distribution in the specimens. Medium dense specimens
(Dr: 56–68%) of Toyoura sand were made by the air pluviation
method for the saturated specimens, while the moisture
tamping method was used for the unsaturated specimens to
prevent the dry sand from absorbing water from the membrane
filter fixed on the pedestal, which would result in difficulty
taking uw measurements. The effect of specimen preparation
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744 735

Table 1
Conditions of undrained cyclic loading tests.

Types of Density s00 (kPa) Sr (%) RL (DA¼ 5%) RL (u¼ 0.9s00 )

soils

IOF-B Dc: 93% 100 100a 0.315 0.315

(91 94%) 84 0.475 0.508
72b 0.546 –
50 100a 0.334 0.324
81 0.532 0.579
Dc: 87% 100 100a 0.228 0.228
(87 88%) 85 0.400 0.408
Dc: 81% 100 100a 0.121 0.118
(75 84%) 76 0.201 0.204
50 100a 0.120 0.120
71 0.200 0.201
Inagi sand Dc: 76% 60 100a 0.140 0.140
(73 77%) 84 0.189 0.189
73b 0.266 0.273
68a,b 0.250 0.255
Toyoura Dr: 63% 60 100a 0.160 0.160
sand (56 68%) 92 0.351 0.351
Dr:  4% 60 100a 0.079 0.079
(-2 -12%) 79 0.221 0.223

Notes: Dr and Dc: average values and ranges (in parentheses) of relative density and compaction degree before applying cyclic loading (Dc ¼ ρdtest/ρdmax  100,
ρdtest: dry density of the tested specimens); s00 and Sr: initial confining pressure and degree of saturation before applying cyclic loading; RL (DA¼ 5%) and RL
(u¼ 0.9s00 ): liquefaction resistance defined by DA ¼ 5% and u¼ 0.9s00 criteria.
a
The tests were conducted without the p-constant condition.
b
Suction of these specimens was measured before and during cyclic loading (the suction values for Inagi sand and IOF-B specimens were about 4 kPa and 1 kPa,
respectively before applying cyclic loading).

Fig. 3. Typical pore water, pore air and suction measurements on (a) Inagi sand and (b) Iron ore fines.

methods on the resistance against liquefaction is discussed in a
later section. The extremely loose specimens (Dr:
 2  12%) of Toyoura sand were all made by moisture
tamping following the instructions provided by Ishihara
(1993). The specimens of Toyoura sand were either flushed
by CO2 and de-aired water followed by applying back pressure
of 200 kPa to achieve the saturated condition (B value
Z 0.95), or only flushed by distilled water (without de-airing)
to remain in the unsaturated condition. In addition, the volume
change of extremely loose Toyoura sand specimens caused by
the saturation or water flushing processes was taken into
account in the interpretation of the test results.
The cell pressure (sh) condition during the application of
vertical cyclic loading was kept constant (without the p-constant
736 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
Fig. 4. Typical behaviors of saturated IOF-B specimens with Dc of 93%
(a) stress strain relationship and (b) effective stress path.
Fig. 5. Typical behaviors of saturated IOF-B specimens with Dc of 87%
(a) stress strain relationship and (b) effective stress path.
condition) for all the saturated specimens and regulated to keep
the p-constant condition for most of the unsaturated specimens.
In addition, the sh of the specimens of Inagi sand with Sr of
68% was kept constant for comparison purposes.
4. Test results
The apparatus performed well when measuring the suction
(S¼ ua  uw) of unsaturated specimens, as described in detail
Fig. 6. Typical behaviors of saturated IOF-B specimens with Dc of 81%
(a) stress strain relationship and (b) effective stress path.
Fig. 7. Typical behaviors of saturated Inagi sand specimens (a) stress strain
relationship and (b) effective stress path.
by Wang et al. (2015). With some unsaturated Inagi sand and
IOF-B specimens, relatively low suction values were mea-
sured, as indicated in Table 1, and the typical measurement
results during cyclic loading are shown in Fig. 3. On the other
hand, for specimens for which no suction measurements were
taken due to their relatively high Sr values, it would be
appropriate to assume their suction values were nearly zero.
The effect of suction on the cyclic behaviors of soils is not
discussed in this paper. It is sufficient to note that the term
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
Fig. 8. Typical behaviors of saturated medium dense specimens of Toyoura
sand (a) stress strain relationship and (b) effective stress path.
Fig. 9. Typical behaviors of saturated extremely loose specimens of Toyoura
sand (a) stress strain relationship and (b) effective stress path.
p  u is used to represent the effective stress state of the
specimens, where u¼ ua for specimens with suction measure-
ments and u¼ uw for specimens without suction measurements.
4.1. Effective stress paths and stress strain relationships
4.1.1. Saturated specimens
Figs. 4–6 show the typical cyclic stress strain relationships
and effective stress paths of saturated IOF-B specimens with
Dc of 93%, 87% and 81% under s00 of 100 kPa, respectively.
737
Fig. 10. Typical behaviors of unsaturated IOF-B specimens with Dc of 93%
(a) stress strain relationship and (b) effective stress path.
Fig. 11. Typical behaviors of unsaturated IOF-B specimens with Dc of 87%
(a) stress strain relationship and (b) effective stress path.
For the sake of comparison, Figs. 7–9 report the cyclic
behaviors of saturated specimens of Inagi sand and Toyoura
sand (medium dense and extremely loose conditions),
respectively.
For saturated IOF-B, zero effective stress is reached regard-
less of density. It is also observed that the assigned deviator
stress amplitude (qcyclic) cannot be achieved for relatively loose
IOF-B (e.g. Dc ¼ 81% in Fig. 6) at large axial strain amplitude
after the effective stress reaches zero, while qcyclic is reached at
limited strain for the relatively dense specimen (i.e. Dc ¼ 93%
738 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
Fig. 12. Typical behaviors of unsaturated IOF-B specimens with Dc of 81%
(a) stress strain relationship and (b) effective stress path.
Fig. 13. Typical behaviors of unsaturated Inagi sand specimens (a) stress strain
relationship and (b) effective stress path.
in Fig. 4). As can be seen in Figs. 8 and 9, similar behavior
was also observed for ordinary Toyoura sand: once zero
effective stress is achieved, the cyclic mobility behavior
changes to a flow type liquefaction behavior as the densities
of the specimens decrease from a medium dense to an
extremely loose condition.
4.1.2. Unsaturated specimens
Figs. 10–12 show the typical cyclic stress strain relation-
ships and effective stress paths of unsaturated IOF-B
Fig. 14. Typical behaviors of unsaturated medium dense specimens of
Toyoura sand (a) stress strain relationship and (b) effective stress path.
Fig. 15. Typical behaviors of unsaturated extremely loose specimens of
Toyoura sand (a) stress strain relationship and (b) effective stress path.
specimens with a Dc of 93%, 87% and 81%, respectively.
Figs. 13–15 illustrate similar relationships for unsaturated
Inagi sand and Toyoura sand (medium dense and extremely
loose), respectively.
While much larger cyclic stress loading amplitudes than
those applied on the corresponding saturated specimens were
applied on the unsaturated specimens in order to trigger
liquefaction, not all the unsaturated specimens reach a zero
effective stress state before the end of the tests (e.g. IOF-B in
Figs. 10 and 11, Inagi sand in Fig. 13). It is worth mentioning
that brittle failure at the extension side accompanied by large
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
Fig. 16. Relationship between CSR and the number of cycles for Toyoura sand specimens (a) medium dense condition and (b) extremely loose condition.
vertical deformation was observed for IOF-B specimens with a
Dc of 93% and 87%.
From the stress stain relationships, it can be seen that most
of axial strain of unsaturated IOF-B with Dc of 93% (Fig. 10)
and medium dense Toyoura sand specimen (Fig. 14) accumu-
lated in the extension side (i.e. negative strain). On the other
hand, the accumulated strain is rather symmetric in the
compression (i.e. positive strain) and extension sides for
IOF-B with Dc of 81% and Inagi sand (Figs. 12 and 13).
For unsaturated specimens with a relatively loose condition
(IOF-B specimens with a Dc of 81%, Inagi sand specimens and
extremely loose Toyoura sand specimens in Figs. 12, 13 and
15, respectively), one of the primary characteristics is that the
cyclic mobility behavior (Seed, 1979) is observed clearly,
while for the corresponding saturated specimens, the behavior
resembles flow type liquefaction at the late stage of the tests
(Figs. 6, 7 and 9).
4.1.3. Liquefaction resistance
The relationship between the cyclic stress ratio (CSR ¼
qcyc/2s00 ) applied to the specimens and the number of cycles
required to induce liquefaction is discussed in this section.
Liquefaction is defined by either a 5% double amplitude of
axial strain (DA ¼ 5% criterion) or by the development of the
excess pore water pressure (or excess pore air pressure for the
specimens with positive suction measurement) to 90% the
initial confining pressure (Δu ¼ 0.9 s00 criterion) for both
saturated and unsaturated specimens. Though neither of these
definitions guarantees that the tested materials reach the state
of liquefaction defined by Castro (1975) and Castro and Poulos
(1977), they can be conveniently applied to the widely
accepted framework proposed by Seed and Idriss (1971) and
Seed (1979) to evaluate the resistance against liquefaction (RL)
of the soil. Note that the relationship between CSR and the
number of cycles before Δu ¼ 0.9 s00 changes when
739
considering the effect of suction on the effective stress
variables (e.g. s00 ). This issue will be discussed later in the
context of the results of unsaturated Inagi sand specimens.
Fig. 16(a) and (b) shows the relationship between CSR and
the number of cycles (i.e. NDA ¼ 5%, DA ¼ 5% criterion in solid
points and NΔu ¼ 0.9s00 , Δu ¼ 0.9s00 criterion in open points)
for Toyoura sand specimens with medium dense (Dr ¼ 63%)
and extremely loose conditions (Dr ¼  4%), respectively. It
can be seen that NDA ¼ 5% and NΔu ¼ 0.9s00 of each test are
similar except in the case of medium dense specimens
subjected to relatively large CSR. The liquefaction resistance
curves are parallel for the saturated and unsaturated cases
under otherwise similar conditions. The liquefaction resistance
of saturated soils (RL,sat) is defined here as the corresponding
CSR value on the liquefaction resistance curve that results in
either DA ¼ 5% or Δu ¼ 0.9s00 after 20 cycles of loading. The
same definition is applied to the liquefaction resistance of the
unsaturated specimens (RL,unsat). Clearly, RL,unsat is much
higher than RL,sat for Toyoura sand specimens prepared at
the similar density conditions. As was mentioned, Toyoura
sand specimens with medium dense (Dr ¼ 63%) were prepared
by the air pluviation method for the saturated case and the
moisture tamping method for the unsaturated case. Tatsuoka et
al. (1986) reported that for Toyoura sand specimens with a Dr
of 60%, the RL,sat of specimens prepared by air pluviation
method was 69% that of specimens prepared by moisture
tamping method. The solid star symbol in Fig. 16(a) shows the
position of 69% of RL,unsat of the test results obtained in this
study. On the other hand, Huang et al. (1999) reported on the
RL of saturated and unsaturated Toyoura sand specimens with
a Dr of 60% prepared by air pluviation method as shown in
Fig. 16(a). Though the significant effect of the preparation
method on RL,sat has also been reported by some other
researchers (Mulilis et al., 1977; Sze and Yang, 2014), it can
be seen from Fig. 16(a) that the significance may be less in the
740 H. Wang et al. / Soils and Foundations 56 (2016) 732–744

Fig. 17. Relationship between CSR and the number of cycles for Inagi sand specimens (a) saturated specimens and unsaturated specimens with the p-constant
condition, (b) unsaturated specimens with/without the p-constant condition and (c) unsaturated specimens with different definitions of effective stress variables.

Fig. 18. Relationship between CSR and the number of cycles for the saturated Fig. 19. Relationship between CSR and the number of cycles for the IOF-B
IOF-B specimens with different density conditions. specimens with Dc of 93%.

case of unsaturated specimens. Therefore, the effect of the
specimen preparation method is not considered in the
discussion.
Fig. 17(a) shows the relationship between the CSR and the
number of loading cycles for saturated Inagi sand specimens
and unsaturated Inagi sand specimens with the p-constant
condition. The NDA ¼ 5% and NΔu ¼ 0.9s00 for each saturated
specimen are similar, while it seems that the development of
DA ¼ 5% takes place earlier than Δu¼ 0.9s00 as Sr decreases.
Also in the case of Inagi sand, RL,unsat is larger than RL,sat. Fig.
17(b) compares the results between the unsaturated specimens
with and without the p-constant conditions. It implies that the
liquefaction resistance under the p-constant condition
(Sr ¼ 73%) is slightly higher than that without the p-constant
condition (Sr ¼ 68%). This may be attributed to the total stress
path and the current cyclic stress ratios (Koseki et al., 2005) of
the two test conditions. Details on this issue will be presented
elsewhere. The liquefaction resistance curve of unsaturated
Inagi sand is shown in Fig. 17(c). The curve is obtained by
using the state variables s0 ¼ p  ua, and s0 ¼ (p ua) þ χ
(ua  uw) defined by Vanapalli et al. (1996), and the residual
degree of saturation (Sr0 ¼ 58%) was obtained by Wang et al.
(2015). It can be seen that the difference caused by the
different definitions of state variables is very small. In this
study, as stated before, the effect of suction is not considered.
It is acknowledged, however, that the effect of suction should
be further studied.
Fig. 18 shows the CSR versus the number of cycles of
saturated IOF-B specimens under different density and con-
fining pressure conditions. It can be seen that RL,sat values
increase as the density increases, while the confining pressure
seems to have a negligible or a minor effect on RL,sat in the
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
Fig. 20. Relationship between CSR and the number of cycles for the IOF-B
specimens with Dc of 87%.
Fig. 21. Relationship between CSR and the number of cycles for the IOF-B
specimens with Dc of 81%.
tested confining pressure range. It also shows that the
liquefaction resistance changes from a linear relationships to
a rounded curve as density increases, which is in good
agreement with the behavior of Toyoura sand as described
by Tatsuoka et al. (1982) and Tatsuoka et al. (1986).
Figs. 19–21 compare the resistance against liquefaction
between saturated and unsaturated IOF-B with Dc of 93%,
87% and 81%, respectively. It can be seen that in general,
under similar density conditions, the resistance against lique-
faction increases as Sr decreases. As was the case with Inagi
sand, a noticeable difference in the number of cycles (between
NDA ¼ 5% and NΔu ¼ 0.9s00 ) required for liquefaction to take place
was also observed for the unsaturated IOF-B specimens,
particularly in those under the relatively dense condition
(Fig. 19 with Dc of 93%). Note that for the specimens with
a Dc of 93% and Sr of 72%, the pore water pressure did not
reach Δu¼ 0.9s00 before the end of the tests (maximum Δuw
values of the specimens was less than 0.3s0’). In Fig. 19, the
741
Fig. 22. Relationship between Sr and liquefaction resistance ratio defined by
DA¼ 5% criterion (LRRDA ¼ 5%).
Fig. 23. Relationship between potential volumetric strain (ε*ν,air) and lique-
faction resistance ratio defined by DA¼ 5% criterion (LRRDA ¼ 5%).
liquefaction resistance curves of the IOF-B specimens with a
Dc of 93% defined by the DA ¼ 5% criterion tend to intersect
each other as the increase in CSR value, while the curves with
different values of Sr are rather parallel when using the
Δu ¼ 0.9s00 criterion. The liquefaction resistance curves
between the saturated and unsaturated conditions in Figs. 20
and 21 are generally parallel for the same density condition
regardless the liquefaction criterion.
4.2. Discussion on the effect of Sr on liquefaction resistance
To evaluate the effect of Sr on the RL, liquefaction resistance
ratio (LRR) is defined as the ratio of RL,unsat to RL,sat under
either the DA ¼ 5% criterion (LRRDA ¼ 5%) or the Δu ¼ 0.9s00
criterion (LRRΔu ¼ 0.9s00 ) for the same soil with otherwise
similar conditions. Fig. 22 shows the relationship between Sr
and LRRDA ¼ 5% for the three types of granular media tested in
this study and other soils used in previous studies by other
742 H. Wang et al. / Soils and Foundations 56 (2016) 732–744

Table 2
Conditions of isotropic consolidation tests.

Media types Dc or Dr (%) s00 (kPa) εν,0.9s00 (%)

Toyoura sand Dr 76 60 0.33

Dr 4 60 0.47
Inagi sand Dc 73 60 1.37
IOF-B Dc 92 100 0.79
50 0.73
Dc 87 100 1.05
Dc 75 100 1.25
50 1.54

Note: s00 here denotes the confining pressure immediately before unloading.

researchers. Masa sand (Yasuda et al., 1999) and Niigata sand
(Ishihara et al., 2001) in Fig. 22 are clean sands with an Fc of
less than 5%. It can be seen that the test results of Toyoura
sand in this study match with some of the results of previous
studies. Though Fig. 22 may imply a monotonic increase trend
of liquefaction resistance as the reduction in Sr for all tested
soils, the relationship between Sr and LRRDA ¼ 5% is not unique
even only for the clean sands under different testing condi-
tions. It also reveals that the LRRDA ¼ 5% values of Toyoura
sand are much higher than those of materials with considerable
fines content (i.e. Inagi sand and IOF-B) under the same Sr.
Okamura and Soga (2006) considered the effect of the
compressibility of pore air in the unsaturated soils on
LRRDA ¼ 5% and proposed a parameter, the potential volumetric
strain (ε*ν,air) to correlate LRRDA ¼ 5%. ε*ν,air is regarded as
the volumetric strain of the specimens due to pore air
compression when the excess pore air pressure equals the
initial confining pressure (Δua ¼ s00 ). It is obtained by applying
Boyle's law:
s0 0 e
εv;air ¼ 0
ð1 Sr Þ
p b þ s0 1þe
where pb is the absolute value of back pressure (kPa) and e is
the void ratio.
In Fig. 23, the relationship between ε*ν,air and LRRDA ¼ 5%
is plotted. It is clear that the relationship between LRRDA ¼ 5%
and ε*ν,air is rather unique for clean sands. However, the data
of Inagi sand and IOF-B do not follow the trend curve
proposed by Okamura and Soga (2006), distributing under
the curve instead.
The components of volumetric strain of saturated and
unsaturated soils under undrained conditions can be written as:
εv;τ þ εv;s’ ¼ εv;air
where εν,τ, εν,s0 and εν,air are volumetric strains caused by the
contraction (or the negative dilatancy) behavior of the soil
subjected to a cyclic shear loading, a reduction in confining
pressure (s0 ) and in the compressibility of pore air ( εν,air ¼
0 for Sr ¼ 100%), respectively. System compliance, membrane
penetration etc. are not considered in Eq. (2). εν,τ may be
regarded as the motion breaking the contact between soil
particles during cyclic loading, while both εν,air and εν,s0 may
Fig. 24. Unloading curves of isotropic consolidation tests.
be regarded as the motions bridging contact by the self-
compression of pore air and elastic self-rebounding of the
particles. In another words, εν,air is the motion inducing loss of
effective stress (i.e. the motion to trigger the liquefaction) and
εν,air and εν,s0 are the motion required to recover the effective
stress (i.e. the motion required to resist liquefaction). The
percentages of effective stress recovery induced by εν,air and
εν,s0 are related to the compressibility of the pore air and the
stiffness of the soil skeleton. Thus, use of the single parameter
ε*ν,air as an index, which is the maximum value of εν,air under
the test conditions employed, may be insufficient to represent
the response characteristics of different soils or the same soil
under different test conditions. As another index, in this paper,
the volumetric strain ratio (Rv ¼ εν,air/εν,s0 ) is proposed to
correlate LRR with the volumetric expansion of the specimens
due to a reduction in confining pressure. Since εν,s0 also exists
ð1Þ in the saturated condition, Rv can be regarded as the weight of
εν,air in motions resisting liquefaction.
To estimate εν,s0 the three types of tested soils, isotropic
consolidation tests (with loading and unloading processes)
were conducted on the saturated specimens under the test
conditions shown in Table 2. After consolidation up to a
specified initial effective confining pressures (s00 ), the effective
confining pressure (s0 ) was decreased step by step to simulate
the reduction process of s0 during undrained cyclic loading. It
should be noted that the degree of saturation affects the
volumetric strains due to the reduction in confining pressure.
In this study, however, we used the measurements of the
saturated specimen for simplicity.
ð2Þ
The relationship between the ratio s0 /s00 and εν,s0 during the
unloading process is shown in Fig. 24. Clearly, this relation-
ship differs considerably for different materials. Toyoura sand
specimens under both density conditions (i.e. loose and dense)
exhibit relatively small volume expansion compared with IOF-
B. Inagi sand specimen shows a similar magnitude of εν,s0 to
IOF-B with a Dc of 75%. The magnitude of εν,s0 of IOF-B
increases as the density decreases, while the effect of s00
(50 kPa and 100 kPa) seems small. The results also show that
 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
the volume of the specimens may expand significantly when s’
reduces to a relatively low value, e.g. less than 0.1s00 . In
addition, non-uniform deformation was observed for some
specimens at very low confining pressure. In this study,
εν,0.9s00 (i.e. εν,s0 induced by 90% reduction of s00 ) is used to
represent εν,s0 in the calculation of Rv. The obtained values of
εν,0.9s00 are listed in Table 2. Note that Dr values of Toyoura
sand specimens in Table 2 slightly deviate from those of the
medium dense specimens or extremely loose specimens of
Toyoura sand indicated in Table 1. However, since εν,0.9s00
values of Toyoura sand specimens are only marginally affected
by the density, εν,0.9s00 values of the specimens with a Dr of
76% and 4% are conveniently used for medium dense speci-
mens and extremely loose specimens, respectively, in the
calculation of Rv.
Fig. 25(a) shows the relationship between Rv, in which εν,air
is represented by ε*ν,air, and LRRDA ¼ 5% of the three types of
granular media. Compared with ε*ν,air (Fig. 23), Rv is better
correlated with LRRDA ¼ 5% (i.e. the effect of soil types is
minimized). However, some data are still dispersed in the case
of IOF-B.
Since εν,0.9s00 is used in the calculation of Rv, it is more
reasonable to use εν,0.9air, which is εν,air when pore pressure
equals 90% of s00 , instead of ε*ν,air for calculating Rv.
Accordingly, using LRRΔu ¼ 0.9s00 is more appropriate than
LRRDA ¼ 5%. The relationship between the new Rv and
LRRΔu ¼ 0.9s00 is shown in Fig. 25(b). It can be seen that the
uniqueness of the relationship is further improved compared
with Fig. 25(a).
Due to the limitation of experimental data available from
this study, it is difficult to conclude that εν,0.9s00 ’ and εν,0.9air
are the most appropriate parameters for evaluating Rv, while
the large difference in εν,s0 between the three granular media
(Fig. 24) suggests that it is necessary to consider εν,s0 based on
the existing theoretical expression (Eq. (2)) to study the
liquefaction properties of unsaturated granular media. It must
also be stressed that the effect of suction on resistance to
liquefaction and the evaluation of εν,s0 was not discussed in
this study for the purpose of simplicity. Suction is considered
another important factor for further study.
Fig. 25. Relationship between volumetric strain ratio (Rv) and LRR (a) Rv ¼ε*ν,air/εν,0.9s0 vs LRRDA ¼ 5% and (b) Rv ¼εν,0.9air/εv,0.9s’ vs LRRu ¼ 0.9s00 .
743
5. Conclusions
In this paper, the results of undrained cyclic triaxial shear tests
on saturated and unsaturated iron ore fines (i.e. IOF-B) and two
ordinary sandy soils (Inagi sand, Toyoura sand) are presented.
The liquefaction behaviors of IOF-B under several density
conditions are evaluated in the light of recent concern regarding
the liquefaction of solid bulk cargo during maritime transportation
and the accidents attributed to this phenomenon. In the search for
a better parameter correlating the liquefaction resistance of
unsaturated granular media, including those with considerable
fines content, two existing parameters (Sr and ε*ν,air) are
examined and a new index (Rv) is proposed. The following
conclusions can be drawn from this study:
1. The behaviors of IOF-B are similar to those of ordinary
sandy soils, for example, the cyclic mobility behavior for
dense specimens and flow type liquefaction behavior for
loose specimens are observed under the saturated condition;
not all specimens reach the zero effective stress state under
the unsaturated condition; the flow type liquefaction beha-
vior evolves to cyclic mobility behavior as the saturated
condition changes to the unsaturated condition of the
specimen; and the density and liquefaction criterion affect
the liquefaction resistance curves. It is worth mentioning
that the effect of confining pressures (50 kPa and 100 kPa
used in this study) seems to be negligible on the liquefac-
tion resistance of saturated IOF-B with a Dc of 93%
and 81%.
2. It is confirmed that the liquefaction resistance of granular
media under the unsaturated condition defined by either the
DA ¼ 5% criterion or the Δu¼ 0.9s00 criterion is higher than
that under the saturated condition. However, two existing
parameters, Sr and the potential volumetric strain, show
non-unique relationships with the liquefaction resistance
ratio (LRR) especially when considering materials with a
considerable fines content.
3. To consider the effects of pore air compressibility and the
volume expansion of specimens during cyclic loading on the
liquefaction resistance of the unsaturated media, a new index,
744 H. Wang et al. / Soils and Foundations 56 (2016) 732–744
volumetric strain ratio (Rv), is proposed to correlate LRR.
Based on the experimental data obtained in this study, it is
demonstrated that Rv can minimize the effect of types of
granular media on liquefaction resistance and exhibits a much
better correlation with LRR than the degree of saturation and
potential volumetric strain, the two existing parameters.
Acknowledgment
The first author would like to express his gratitude to the
Ministry of Education, Culture, Sports, Science and Technol-
ogy of Japan for its financial support during the doctoral
course. This work was supported by JSPS Grants-in-Aid for
Scientific Research, Grant numbers 26630219 and 15H04036.
References
Ampadu, S.K., Tatsuoka, F., 1993. Effect of setting method on the behavior of
clays in triaxial compression from saturated to undrained shear. Soils
Found. 33 (2), 14–34.
Arab, A., Shahrour, I., Lancelot, L., 2011. A laboratory study of liquefaction of
partially saturated sand. J. Iber. Geol. 37 (1), 29–36.
Castro, G., 1975. Liquefaction and cyclic mobility of saturated sands. In:
Proceeding of ASCE, vol. 101, GT6, pp. 551–569.
Castro, G., Poulos, S.J., 1977. Factors affecting liquefaction and cyclic
mobility. In: Proceeding of ASCE, vol. 103, GT6, pp. 502–516.
IMO, 2013. Early implementation of draft amendments to the IMSBC Code
related to the carriage and testing of iron ore fines. International Maritime
Organization, DSC.1/Circ.71.
Gard., 2012. Brazil-liquefaction of bauxite cargoes. Gard. 〈http://www.gard.no/
ikbViewer/Content/20651235/Gard%20Alert%20Brazil%20Liquefaction%
20of%20bauxite%20cargoes.pdf〉 (accessed on 26.03.2015).
Goto, S., Shamoto, Y., 2002. Estimation method for the liquefaction strength
of unsaturated sandy soil (part 2). In: Proc. 37th Jpn. Nat. Conf. Geotech.
Engrg., pp. 1987-1988 (in Japanese).
Huang, Y., Tsuchiya, H., Ishihara, K., 1999. Estimation of partial saturation
effect on liquefaction resistance of sand using P-wave velocity. In: Proc.
JGS Symp., 113, 431-434 (in Japanese).
Isacson, C., 2010a. Re: India-safe shipment of iron ore fines from Indian ports.
Int. Group Member Circular No.16. 〈http://www.gard.no/ikbViewer/Con
tent/10129118/Member%20Circular%2016%202010%20India%20Safe%
20shipment%20of%20iron%20ore%20fines%20from%20Indianports.pdf〉
(accessed on 26.03.2015).
Isacson, C., 2010b. Re: Indonesia and the Philippines-Safe Carriage of Nickel
Ore Cargoes. Int. Group Member Circular No. 23. 〈http://www.gard.no/
ikbViewer/Content/12143419/Member%20Circular%2023%202010%
20Indonesia%20and%20the%20Philippines%20%BF%20Safe%20Carriage
%20of%20Nickel%20Ore%20Cargoes.pdf〉 (accessed on 26.03.2015).
Ishihara, K., 1993. Liquefaction and flow failure during earthquakes. Geo-
technique 43 (3), 351–415.
Ishihara, K., Tsuchiya, H., H., Huang, Y., Kamada, K., 2001. Recent studies
on liquefaction resistance of sand- effect of saturation. In: Proc. 4th Int.
Conf. Recent Advance in Geotech. Earthquake Engrg. and Soil Dynamics,
pp. 1–7.
Koseki, J., Yoshida, T., Sato, T., 2005. Liquefaction properties of Toyoura
sand in cyclic torsional shear tests under low confining stress. Soils Found.
45 (5), 103–113.
LPC, 2011. Cargo liquefaction problems – sinter feed from Brazil. Loss
prevention circular, No. 06-11. 〈http://www.gard.no/ikbViewer/Content/
15758052/06-11%20Cargo%20liquefaction%20problems%20-%20sinter%
20feed%20from%20Brazil.pdf〉 (accessed on 26.03.2015).
Mulilis, J.P., Seed, H.B., Chan, C.K., Mitcholl, J.K., Arulanandan, K., 1977.
Effects of sample preparation on sand liquefaction. In: Proceeding of
ASCE, vol. 103, GT2, pp. 91–108.
NILIM, 2011. Report of continuous monitoring of an air injection ground:
regarding the method of the micro-bubble water injection as a counter-
measure of liquefaction of residential land (in Japanese). 〈http://www.nilim.
go.jp/lab/jbg/takuti/ekijyoka/20110513sato.pdf〉 (accessed on 26.03.2015).
NILIM, 2013. Examination and investigation on the ground dewatering
method applied on liquefied areas (in Japanese). 〈http://www.mlit.go.jp/
common/000986855.pdf〉 (accessed on 26.03.2015).
Nishimura, T., Koseki, J., Fredlund, D.G., Rahardjo, H., 2012. Micro-porous
membrane technology for measurement of soil-water characteristic curve.
Geotech. Test. J 35, 201–208.
Okamura, M., Ishihara, M., Tamura, K., 2006. Degree of saturation and
liquefaction resistances of sand improved with sand compaction pile. J.
Geotech. Geoenviron. Eng. 132 (2), 258–264.
Okamura, M., Soga, Y., 2006. Effects of pore fluid compressibility on
liquefaction resistance of partially saturated sand. Soils Found. 46 (5),
695–700.
Seed, H.B., 1979. Soil liquefaction and cyclic mobility evaluation for level
ground during earthquakes. In: Proceeding of ASCE, vol. 105, GT2,
pp. 201–255.
Seed, H.B., Idriss, M., 1971. Simplified procedure for evaluating soil liquefac-
tion potential. In: Proceeding of ASCE, vol. 97, SM9, pp. 1249–1273.
Sherif, M.A., Ishibashi, I., Tsuchiya, C., 1977. Saturation effects on initial soil
liquefaction. In: Proceeding of ASCE, vol. 103, GT8, pp. 914–917.
Sze, H.Y., Yang, J., 2014. Failure modes of sand in undrained cyclic loading
impact of sample preparation. J. Geotech. Geoenviron. Eng. 140 (1),
152–169.
Tatsuoka, F., Muramatsu, M., Sasaki, T., 1982. Cyclic undrained stress–strain
behavior of dense and by torsional simple shear test. Soils Found. 22 (2),
55–70.
Tatsuoka, F., Ochi, K., Fujii, S., Okamoto, M., 1986. Cyclic undrained triaxial
and torsional shear strength of sands for different sample preparation
methods. Soils Found. 26 (3), 23–41.
Towhata, I., Maruyama, S., Kasuda, K., Koseki, J., Wakamatsu, K., Kiku, H.,
Kiyota, T., Yasuda, S., Taguchi, Y., Aoyama, S., Hayashida, T., 2014.
Liquefaction in the kanto region during the 2011 off the pacific coast of
Tohoku earthquake. Soils Found. 54 (4), 859–873.
Tsukamoto, Y., Ishihara, K., Nakazawa, H., Kamada, K., Huang, Y., 2002.
Resistance of partly saturated sand to liquefaction with reference to
longitudinal and shear wave velocities. Soils Found. 42 (6), 93–104.
Unno, T., Kazama, M., Uzuoka, R., Sento, N., 2008. Liquefaction of
unsaturated sand considering the pore air pressure and volume compres-
sibility of the soil particle skeleton. Soils Found. 48 (1), 87–99.
Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., Clifton, A.W., 1996. Model for
the prediction of shear strength with respect to soil suction. Can. Geotech.
J. 33, 379–392.
Wang, H., Koseki, J., Nishimura, T., 2014. SWCC measurement of two types
of iron ores. In: Proc. 6th Int. Conf. on unsaturated soils, Unsat2014,
Sydney, pp. 973–979.
Wang, H., Koseki, J., Sato, Tan Tian, J., 2015. Experimental evaluation of
liquefaction resistance of unsaturated sandy soils. In: Proc. 6th Int. Symp. on
Deformation Characteristics of Geomaterials, Buenos Aires, pp. 299–306.
doi:10.3233/978-1-61499-601-9-299.
Yang, J., 2002. Liquefaction resistance of sand in relation to P-wave velocity.
Geotechnique 5 (4), 295–298.
Yang, J., Savidis, S., Roemer, M., 2004. Evaluating liquefaction strength of
partially saturated sand. J. Geotech. Geoenviron. Eng. 130 (9), 975–979.
Yamaguchi, T., Mori, T., Kazama, M., Yoshida, N., 2012. Liquefaction in
Tohoku district during the 2011 off the pacific coast of Tohoku earthquake.
Soils Found. 52 (5), 811–829.
Yasuda, S., Harada, K., Ishikawa, K., Kanemaru, Y., 2012. Characteristics of
liquefaction in Tokyo Bay area by the 2001 Great East Japan Earthquake.
Soils Found. 52 (5), 793–810.
Yasuda, S., Kobayashi, T., Fukushima, Y., Kohari, M., Simazaki, T., 1999.
Effect of degree of saturation on the liquefaction strength of Masa. In: Proc.
34th Jpn. Nat. Conf. Geotech. Engrg., pp. 2071–2072 (in Japanese).
Yoshimi, Y., Tanaka, K., Tokimatsu, K., 1989. Liquefaction resistance of a
partially saturated sand. Soils Found. 29 (3), 157–162.