Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274

Contents lists available at ScienceDirect

Journal of Rock Mechanics and

Geotechnical Engineering
journal homepage: www.rockgeotech.org

Review

Review of collapse triggering mechanism of collapsible soils due to

wetting
Ping Li a, b, Sai Vanapalli a, *, Tonglu Li b
a
Department of Civil Engineering, University of Ottawa, Ottawa, ON, K1N6N5, Canada
b
School of Geological Engineering and Geomatics, Chang’an University, Xi’an, Shaanxi, 710054, China

a r t i c l e i n f o a b s t r a c t

Article history: Loess soil deposits are widely distributed in arid and semi-arid regions and constitute about 10% of land
Received 4 August 2015 area of the world. These soils typically have a loose honeycomb-type meta-stable structure that is
Received in revised form susceptible to a large reduction in total volume or collapse upon wetting. Collapse characteristics
7 December 2015
contribute to various problems to infrastructures that are constructed on loess soils. For this reason,
Accepted 9 December 2015
Available online 18 January 2016
collapse triggering mechanism for loess soils has been of significant interest for researchers and prac-
titioners all over the world. This paper aims at providing a state-of-the-art review on collapse mecha-
nism with special reference to loess soil deposits. The collapse mechanism studies are summarized under
Keywords:
Collapse mechanism
three different categories, i.e. traditional approaches, microstructure approach, and soil mechanics-based
Microstructure approaches. The traditional and microstructure approaches for interpreting the collapse behavior are
Constitutive relationships comprehensively summarized and critically reviewed based on the experimental results from the
Compacted soils literature. The soil mechanics-based approaches proposed based on the experimental results of both
Natural loess soils compacted soils and natural loess soils are reviewed highlighting their strengths and limitations for
Elastoplastic models estimating the collapse behavior. Simpler soil mechanics-based approaches with less parameters or
Yield surface parameters that are easy-to-determine from conventional tests are suggested for future research to
Structural strength
better understand the collapse behavior of natural loess soils. Such studies would be more valuable for
use in conventional geotechnical engineering practice applications.
Ó 2016 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Production and hosting by
Elsevier B.V. All rights reserved.

1. Introduction collapsible fabric provided there is an open, potentially meta-

Loess soils are widely distributed and constitute about 10% of
the total land area of the world. Several countries including China,
Russia, United States, France, Germany, New Zealand and
Argentina, have a large area of loess soil deposits (Phien-wej et al.,
1992; Rogers et al., 1994; Rogers, 1995; Al-Rawas, 2000; Nouaouria
et al., 2008; Ryashchenko et al., 2008; Gaaver, 2012). These soils are
typically formed with a loose honeycomb-type meta-stable struc-
ture and are susceptible to a sudden decrease in total volume or
collapse upon wetting (Feda, 1988; Houston et al., 1988; Lommler
and Bandini, 2015). Different types of natural soils may develop a
* Corresponding author. Tel.: þ1 613 562 5800.
E-mail address: vanapall@eng.uottawa.ca (S. Vanapalli).
Peer review under responsibility of Institute of Rock and Soil Mechanics,
Chinese Academy of Sciences.
1674-7755 Ó 2016 Institute of Rock and Soil Mechanics, Chinese Academy of
Sciences. Production and hosting by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jrmge.2015.12.002
stable, partly saturated structure, and a high enough applied
stress (Barden et al., 1973; Lawton et al., 1989). In addition, any type
of soil compacted at dry of optimum condition is collapsible in
nature (Fredlund and Gan, 1995; Kato and Kawai, 2000; Pereira and
Fredlund, 2000). Collapse and other collapse associated problems,
such as differential settlement, earth cracks, landslides and falls,
have contributed to serious damages to the infrastructures that are
constructed on loess soils, including loss of human lives in certain
scenarios (Derbyshire, 2001; Houston et al., 2001; Peng et al.,
2006). During the period of 1974e1975, in China, it was reported
that a total of 1505 buildings were damaged and 80 km-long un-
derground pipeline ruptured due to collapse of loess soils (Sun
et al., 2013). Fu (2005) reported that typically a quarter of time is
required with respect to the entire construction period to treat the
collapsible soils prior to placing foundation. Also, the cost associ-
ated with collapsible ground preparation is typically about one
third of the total cost of the infrastructure. With urban develop-
ment on the rise in various regions of the world with collapsible soil
deposits, there will be more access to water for these soils. As a
result, there will be more wetting associated collapse problems. For
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
this reason, it is important to understand the collapse mechanism
of these soils.
During the last six decades, many researchers have focused on
studying the collapse mechanism of various collapsible soils upon
wetting. The discussions on this topic are summarized under three
categories, i.e. traditional approaches, microstructure approach,
and soil mechanics-based approaches. Among the traditional ap-
proaches, collapsibility was always interpreted considering one
sole factor. For example, collapse was attributed to loss of capillary
tension or solution of soluble salts (Guo, 1958; Dudley, 1970). In
addition, many researchers studied the influence of soil properties,
such as the density, clay content, Atterberg limits and grain size
distribution, on the collapsibility of loess soils (Sun, 1957; Liu, 1994;
Zhao and Chen, 1994; Zhang, 2002; Fan and Guo, 2003; Song and
Wang, 2004; Chen et al., 2006; Yuan, 2009), and a large number
of empirical equations have been proposed in the literature relating
the collapsibility to conventional soil properties (Clevenger, 1958;
Feda, 1964, 1966, 1995; Handy, 1973; Basma and Tuncer, 1992;
Fan and Guo, 2003; Song and Wang, 2004; Ayadat and Hanna,
2007; Zorlu and Kasapoglu, 2009; Noor et al., 2013). However,
most of these empirical equations can be extended only for local
soils for which they have been developed. In other words, the
proposed empirical equations are not universally valid for use in
conventional geotechnical engineering practice (Sun, 1957; Gao,
1979, 1990; Yang, 1988). This is due to soils exhibiting different
forms of microstructure and hence presenting different mechanical
behaviors in spite of having similar physical properties. For
example, a loess soil with few large pores may have the same void
ratio or density as another loess soil with many small pores;
however, their collapse behaviors will be significantly different.
The collapse nature of loess and other soils is interpreted using
the information of soil microstructure as a tool (Derbyshire and
Mellors, 1988; Delage et al., 1996; Romero and Simms, 2008).
Loess soil microstructure can be analyzed in terms of four key
factors based on the comprehensive studies on microstructure of
loess soils, i.e. particle pattern, contact relation, pore form, and
bonding material. These four factors are dependent on each other;
however, the pore form and bonding material are suggested as the
two dominant factors that have more influence on the collapse
behavior (Gao, 1980a, b; Lei, 1983, 1987; Yang, 1988; Zhao and
Chen, 1994). Many researchers have been involved in classifying
soil pores and distinguishing the water stability of various possible
bonding materials, as well as exploring their respective influence
on the collapsibility of loess (Lei, 1983, 1987; Derbyshire and
Mellors, 1988; Yang, 1988; Zhao and Chen, 1994; Osipov and
Sokolov, 1995; Assallay et al., 1997; Smalley et al., 2001; Jiang
et al., 2014a, b; Lommler and Bandini, 2015). It is widely
accepted that microstructure plays a key role in controlling the
collapse behavior; however, it lacks a simple quantitative
descriptor for estimating collapse deformations (Alonso et al.,
1993). This limitation has been addressed and alleviated to a
great extent in recent years by extending image processing tech-
niques which facilitate the quantitative analysis of soil micro-
structure (e.g. Chen and Sha, 2009a, b; Gu et al., 2011; Fang et al.,
2013a, b). These image processing programs are mostly based on
the principle of binary grey segmentation to provide reasonably
satisfactory results, particularly, for coarse soils. However, their
application to fine-grained silts and clays has not been well vali-
dated. Gu et al. (2011) indicated that it is difficult to clearly
distinguish the grains and pore sizes from digital images using the
presently available programs.
The third category approaches use the concepts of soil me-
chanics for explaining the collapse behavior. Collapsible soils are
257
typically unsaturated and significant collapse generally occurs prior
to reaching fully saturated condition (El-Ehwany and Houston,
1989). For this reason, the concepts of mechanics for unsaturated
soils are more rational for interpreting the collapse behavior
(Tadepalli and Fredlund, 1991; Fredlund and Rahardjo, 1993;
Fredlund and Gan, 1995; Habibagahi and Mokhberi, 1998; Chen,
1999; Pereira and Fredlund, 2000). The two stress state variables
approach proposed by Fredlund and Morgenstern (1977) for
describing the mechanical behavior of unsaturated soils has been
extended to studies of volume change behavior due to loading and
wetting of collapsible soils. Collapse is attributed to the loss of
strength associated with suction decrease as a result of wetting
(Fredlund and Gan, 1995). For this reason, collapse behavior has
been widely investigated using suction-controlled wetting tests
(e.g. Chen, 1999; Chen et al., 1999; Sun et al., 2004, 2007a, b;
Jotisankasa, 2005), based on which, various models have been
proposed for modeling the collapse behavior with respect to
varying stress state variable (i.e. matric suction) (Tadepalli et al.,
1992; Fredlund and Gan, 1995; Pereira and Fredlund, 1997, 2000).
During the last quarter century, a number of elastoplastic models
have been developed for modeling the behavior of unsaturated
soils (Alonso et al., 1990; Josa et al., 1992; Gens and Alonso, 1992;
Wheeler and Sivakumar, 1995; Cui and Delage, 1996; Wheeler,
1996; Wheeler et al., 2003). These models have been extended or
modified to interpret the volume change behavior of collapsible
soils (e.g. Chen et al., 1999; Kato and Kawai, 2000; Sun et al., 2004,
2007a, b). In these models, collapse is explained as a part of
deformation when the stress path crosses the elastic region or yield
surfaces. For quantitative analysis, elastoplastic models provide a
more precise way to estimate the collapse deformations with
respect to varying stress state variables. However, they are too
complex as a large number of parameters are required to be
determined from cumbersome suction-controlled tests. On the
other hand, for addressing some scenarios of collapse, only a few
model parameters are required. However, the parameters will be
different for the same soils when their initial conditions, such as the
initial water content, and the stress state, are different. In other
words, when using these models, for each soil sample with a
different initial condition, one suction-controlled test should be
conducted, from which the model parameters can be determined.
In addition, most of these models were proposed based on the
experimental results of compacted soils, which may show different
collapse behavior from the natural loess soils. For this reason, soil
mechanics-based approaches proposed especially for natural loess
soils are reviewed in this paper. Loss of structural strength is
considered to be a key factor for natural loess soil collapse. The
structural strength is mainly influenced by the factors associated
with loess soil deposition and initial water content (Hu et al., 2000,
2004). Two models based on the concepts of breakage mechanics
proposed by Shen and his group (Shen, 1993, 2003; Shen and Deng,
2003; Shen and Hu, 2003) for modeling the collapse behavior of
natural loess soils due to loading and wetting were reviewed. These
models have the same limitation as the models described earlier;
they are complex, and cumbersome tests are required for deter-
mining the parameters, which restricts the application of these
models in conventional geotechnical engineering practice. For
these reasons, relatively simple models with less parameters or
parameters that are easy-to-determine from conventional tests are
suggested for advancing research related to collapse behavior of
natural loess soils.
The comprehensive summary provided in this paper can be
useful for addressing problems associated with collapsible soils and
for proposing more efficient approaches in the future.
258 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
2. Traditional approaches for interpreting the wetting-
induced collapse mechanism
2.1. Loss of capillary tension
Unsaturated soils exhibit apparent strength due to the capillary
tension acting along the tangent of the menisci (Terzaghi, 1943).
Water and clays suspending in the water retreat towards silt con-
tacts as the soil desiccates, which produces considerably high
capillary tension to maintain the soil structure under unsaturated
condition (Fig. 1). However, addition of water destroys the capillary
tension, which likely contributes to reduction of soil strength and
triggers collapse. Capillary tension concept has been used to
explain the collapsibility of loess soils (Alfi, 1984), which however
was questioned because of limitations in explaining the phenom-
enon of identical loess soil samples showing different collapse
behaviors when they are treated with different liquids, such as
water or saline solutions (Guo, 1958).
2.2. Solution of soluble salts
Crystal or microcrystalline soluble salts are widely considered to
act as bonds in unsaturated loess soils. These salts are expected to
dissolve as soil water content increases, contributing to a decrease
in bonding strength and leading to collapse (Guan, 1986; Houston
et al., 1988). This approach however was questioned because sol-
uble salts are found dissolved in pore water even when the water
content is as low as 8% (Tan, 1988). In addition, the soluble salts take
up typically less than 0.1% in loess soils. Table 1 summarizes various
soluble salts contents in loess soils in China (Yang, 1988). Guan
(1989) suggested that self-weight collapsible loess soils have a
natural water content of 4%e5%, while non-self-weight collapsible
loess soils have a value of 21%e24%, which indicates that wetting-
induced collapse could not be solely attributed to the solution of
soluble salts, especially for non-self-weight collapsible loess soils.
2.3. Shortage of clays
Several researchers described collapse arises in some soils as a
result of shortage of clay size fraction (for example, with a clay
content less than 10%) (Gao, 1979; Rogers et al., 1994). If a soil has
enough clays, it may expand instead of collapse on soaking. Kraev
(1969) suggested that limited montmorillonite presence would
swell and loosen the primary loose soil structure on soaking, which
contributes to collapse. However, Guan (1989) suggested that self-
Fig. 1. Capillary analogy applied to loess condition.
weight collapsible loess soils have a clay content of 23%e39%, and
non-self-weight collapsible loess soils have a value of 30%e42%.
These loess soils show serious collapsibility upon wetting in spite of
their high clay content. In addition, some researchers reported that
a monotonic relationship does not exist between the collapsibility
of loess soil and its clay content. The maximum collapse occurs at
an intermediate rather than the minimum clay content (Hu et al.,
1999; Li et al., 2004; Song and Wang, 2004). For this reason, clay
content should not be considered as the key factor that triggers the
wetting-induced collapse for loess soils.
2.4. Under compaction
Under compaction attributes the collapse nature to the charac-
teristic of porous fabric of loess (Houston et al., 1988; Tadepalli and
Fredlund, 1991). However, pores in loess soils having different sizes
and forms would have completely different influences on the
collapsibility. The size and form of pores are controlled by the
factors associated with loess soil deposition. Various classifications
of loess soil pores taking account of their size and form and their
respective effect on the collapsibility are discussed in detail in later
sections.
The traditional or conventional approaches listed above were
found to be unsatisfactory for universally explaining the collapse
behavior of loess soils. However, these approaches have signifi-
cantly contributed towards better understanding the collapsible
loess soils.
3. Microstructure approach for interpreting wetting-induced
collapse behavior
Numerous researchers have reported that any soil compacted at
dry of optimum condition typically exhibits collapse behavior upon
wetting, while soils compacted at wet of optimum condition show
no collapsibility (Barden, 1965; Lawton et al., 1989, 1992; Tadepalli
and Fredlund, 1991). Collapse is therefore attributed to the meta-
stable nature of the open flocculated structure associated with
dry of optimum water content condition. This understanding
contributed to interpreting loess soil collapsibility from the
microstructure information. Recent advances in techniques and
methods, including scanning electronic microscope (SEM), X-ray
diffraction and mercury intrusion porosimetry (MIP), have been
found to be valuable for understanding the microstructure and its
significance on soil mechanical behavior. Several investigators
provided reviews on the techniques and methods specifically
aimed at characterizing the microstructure of unsaturated soils,
among which SEM and MIP have been frequently used (Al-Mukhtar
et al., 1996; Simms and Yanful, 2001, 2002; Romero and Simms,
2008). In addition, several image processing techniques have
been developed, such as the Scion (Fang et al., 2013a, b), Videolab
(Shi et al., 1995) and Mirage (Sha and Chen, 2006; Chen and Sha,
2009a, b), and simple commercial image processing programs
such as MATLAB (Wang et al., 2011; Mao et al., 2004) and Photo-
shop (Zhang et al., 2009). These programs extract quantitative in-
formation about soil pores or grains from the digital images,
providing strong support for interpreting collapse behavior from
microstructure information.
The microstructure of loess soils is analyzed in terms of four
factors, i.e. (i) particle pattern, (ii) contact relation, (iii) pore form,
and (iv) bonding material.
3.1. Particle pattern
It is widely agreed that loess soils are composed of particles
typically in the range of 10e60 mm, of which particles larger than
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274 259

Table 1
Soluble salts contents in loess soils in China (Yang, 1988).

Location NaHCO3 Na2SO4$10H2O KCl NaCl MgSO4$7H2O MgCl2 Total amount (%)

Zanhuang, Hebei 0.001 0.005 0.001 0.001 0 0 0.008

Taiyuan, Shanxi 0 0.036 0.001 0 0.004 0.003 0.044
Linfen, Taiyuan 0 0.068 0.002 0.002 0 0 0.072
Luochuan, Shaanxi 0.012 0.027 0.001 0.001 0 0 0.041
Tianshui, Gansu 0.003 0.038 0.001 0.004 0 0 0.046
Dingxi, Gansu 0 0.073 0.002 0 0.018 0.003 0.096

20 mm account for about 75% of the total solid constituents (Tan,
1988; Dijkstra et al., 1994). During the loess deposition, primary
grains may be coated and cemented by clay platelets or calcium
carbonate to form aggregates or coagulum (i.e. collection of ag-
gregates). Together with primary grains, they all were regarded as
skeleton particles of loess soil structure. Fig. 2 shows the devel-
opment of loess soil structure from single grain to an element of
loess soil structure (Gao, 1980a, b; 1981). All these particle pat-
terns are also illustrated using SEM images (see Fig. 3) which are
from natural Malan loess and Lishi loess, respectively, from
Lanzhou, China. Both primary grains and aggregates are generally
found in a loess soil. Primary grain dominant soils are more likely
to have higher collapse possibility since they have a more open
structure compared to aggregate dominant soils. In latter type
soils, interspaces are well filled with fine grains, these soils
therefore have a more stable structure than the former ones (Gao,
1979, 1980a). In general, the more the aggregates, the weaker the
collapsibility of loess would be. The particle pattern in loess soils
varies with climate factors (Liu, 1978; Beckwith and Hansen, 1982;
Lin and Liang, 1982; Gao, 1984). Typically, the effects of weath-
ering, clayization and calcium carbonate microcrystallization in
wet and warm environment are more pronounced compared to
those in dry and cold environment. Due to this reason, more small
size grains, clays and microcrystalline calcium carbonates are

Fig. 2. Particle patterns in loess soils: (a) primary grain; (b) primary grain coated by clays or calcium carbonate; (c) aggregate; (d) coagulum (collection of aggregates); (e)
microstructure element (modified after Gao, 1980a).

Fig. 3. The SEM photographs of (a) natural Malan loess and (b) natural Lishi loess in Lanzhou, China (modified after Dijkstra et al., 1994).
260 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
prone to forming in wet and warm areas, which are favorable for
aggregates development. In addition, the longer time is available
for loess deposition, the formation of more aggregates is likely
(see Fig. 3).
On the other hand, particles in loess soils could be classified
based on their size. Lei (1983, 1987) suggested grains ranging from
5 mm to 250 mm as skeleton particles, particles larger than 10 mm as
coarse grains and those smaller than 10 mm as fine grains. Yang
(1988) suggested particles larger than 10 mm, which account for
more than 70% of the total solid mass, form the skeleton of loess
soils, particles smaller than 2 mm act as bonding materials, those
between 2 mm and 10 mm are typically filling materials. Yuan et al.
(2007) divided loess soil particles into sand (60 mm < d < 2000 mm),
silt (2 mm < d < 60 mm) and clay (d < 2 mm). A loess soil with higher
percentage of silts is more favorable to form large pores and shows
high collapsibility (Derbyshire and Mellors, 1988; Assallay et al.,
1997; Grabowska-Olszewska, 1975, 1988). Handy (1973) stated
that collapsibility is inversely proportional to clay content for loess
soils. However, several researchers stated that there is an optimum
clay content corresponding to the maximum collapsibility for loess
soils (Hu et al., 1999; Li et al., 2004; Song and Wang, 2004). This
value varies depending on loess soil type and its initial water
content.
Present understanding of particle pattern in loess soils is mostly
based on particle size and origin; however, classifications were also
proposed based on particle shape (including sphere, flake, tube,
needle and floccules) (Krinsley and Smalley, 1973). Studies on
particle shape and the influence of particle shape on soil structure
stability are more meaningful for granular soils. It is however
difficult to identify the shape of aggregates or coagulum in loess
soils. Particle shape is also a significant indicator of the mechanical
behavior of clays which are mostly fillings or bonds in loess soils;
however, loess particle shape has not been comprehensively
investigated (Dijkstra et al., 1995).
3.2. Contact relation
For most loess soils, significant collapse starts to occur at a
critical water content, which varies depending on the initial water
content and vertical stress under which the soil is wetted. As the
soil water content increases beyond the critical value, bonding
strength fails to support the soil fabric continuously (Zheng and
Zhang, 1989). Subsequently, soil structure fails and a fundamental
particle rearrangement takes place, transforming an initial stable
open structure into a remolded close structure, which is considered
as the collapse process (Dijkstra et al., 1995). Therefore, contact
relation is an important indicator for determining whether or not it
is easy for particles to slip over each other. Particles typically have
two main contact relations in loess soils depending on the amount
of bonds at the contact, i.e. point-contact and face-cementation
relation, as illustrated in Fig. 3 (Gao, 1980a, 1981). Particles in
point-contact relation are mostly naked and there are only a few
bonds at the contact. On the contrary, particles in face-cementation
relation are typically thickly coated, and there are large amount of
bonds at the contact, making it look like face-to-face contact
relation.
Conditions are favorable for maintenance of point-contact
relation in dry cold areas like northwest China for several rea-
sons: (i) weak weathering makes it unfavorable for grains to be
compressed; (ii) a little precipitation makes it difficult for calcium
carbonate microcrystallization and clayization. Conversely, in wet
and warm areas, frequent circles of precipitation and evaporation
as well as active biological chemical action result in sufficient
microcrystalline calcium carbonate, clays and small size grains.
Generally, two contact relations coexist in loess with different
proportions, the variation in terms of contact relation is consistent
with climate variation (Gao, 1980a, 1981; Guan, 1989). In addition,
point-contact relation is more likely to be in grain dominant soils,
while aggregate dominant soils have more face-cementation
relation.
Contact relation in loess soils was also classified into overhead,
interlocking and dispersed contact (Lei, 1983, 1987). These kinds of
contact relations were defined taking into account the pore form
and bonds state. Overhead contact is for surrounding particles of
overhead pores; interlocking contact is a result of interlocking
particle arrangement, which is commonly observed in any kind of
soil; dispersed contact is the scenario that soil particles are finely
divided by bonds.
A kind of contact relation “clay-bridge” was found in loess soils
(Derbyshire and Mellors, 1988). The “ideal bridge” was found to
comprise clay particles peeling away from the surface of one grain
linking up with similarly arranged clays on the surface of an adja-
cent grain. Clay-bridge was found to vary in thickness from site to
site depending on the degree of clayization and climate condition.
3.3. Pore form
Different particle patterns and contact relations result in
different pore forms. Pores in loess soils were divided into macro-
pores, spaced pores, intergranular pores and intragranular pores
from the pore size with respect to surrounding particles by Gao
(1980a, 1981), as shown in the sketches in Fig. 4. All forms in
loess soil except for macropores are illustrated in a SEM image
modified after Derbyshire (2001) (Fig. 5), which is from natural
Malan loess in Lanzhou region in China. Macropores usually are
root holes, wormholes and holes made by rodents’ action. These
pores are commonly found at a relatively shallow depth in
collapsible loess soils (i.e. within 1.0 m approximately, see in Lei,
1983), and their walls are heavily covered by reborn calcium car-
bonates in most cases. For these reasons, macropores are believed
to contribute little to overall collapse or particle rearrangement
because of their low proportion in spite of individual macropore
having a large volume (Gao, 1980a; Lei, 1983).
Spaced pores are typically associated with spaced arrangement
of aeolian deposits. They are also characterized by the size larger
than surrounding particles which are poorly cemented and more
likely in point-contact relation (Yang, 1988). Spaced pores
contribute to favorable spatial conditions for collapse to occur. In
addition, spaced pores contribute to a critical state condition as the
bonding strength equals the gravitational potential of overhanging
soil particles. Under such a scenario, when the soil gets wet and
bonding strength weakens, overhanging particles fall into spaced
pores and soil collapses. In many scenarios, intergranular pores
formed by interlocking particle arrangement are observed in any
kind of soil, including expansive soils (Popescu, 1985; Gens and
Alonso, 1992; Saba et al., 2014). Intergranular pores have been
experimentally proved to contribute slightly to loess soil collaps-
ibility (Reznik, 2007). Typically, all kinds of pores coexist in a loess
soil except for macropores. In arid environment, spaced pores
develop since the poorly-coated grains are prone to having higher
hardness for transferring forces, which effectively prevents the soil
from being compressed under dead-weight (Gao, 1980a, 1981).
However, loess soils in humid regions that lie deep beneath the
surface predominantly have intergranular and intragranular pores.
Compared with Gao’s (1980a, 1981) classification, Lei (1983, 1987)
classified loess soil pores into original pores and secondary pores.
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274 261

Fig. 4. The classification for loess soil pores: (a) spaced pores; (b) intergranular pores; (c) intragranular pores and (d) macropore (modified after Gao, 1980a, 1981).

Original pores include intergranular pores (i.e. overhead pores and

Fig. 5. The SEM photograph of natural Malan loess from Lanzhou, China (modified
after Derbyshire, 2001).
interlocking pores) and intragranular pores (i.e. pores within the
aggregates). Overhead pores, interlocking pores and intragranular
pores correspond to three different soil structures, i.e. overhead
structure, interlocking structure and matrix structure, as shown in
Figs. 5 and 6. The secondary pores include root holes, wormholes,
rodent holes, joints, fissures and Karst caves. The characteristics of
all kinds of pores, as well as their influences on the collapse
behavior are summarized in Table 2.
Yang (1988) found that pores smaller than 54 mm account for
97% of the total void volume from the microstructure information
of loess soils deposited in different locations and depths. These
small pores (<54 mm) contribute approximately 80% to collapse.
Although pores larger than 54 mm have a high failure rate, they
were found to make little contribution to collapse or particle
rearrangement because of their low proportion. Fig. 7 presents the
pore size distribution in terms of both cumulative pore volume and
pore size density of a loess soil in Shaanxi, China, from MIP tests
(Jiang et al., 2014a, b). From the pore size distribution (PSD), it can
be observed that loess soils consist predominantly of pores smaller
than 20 mm. The pore size density function (PSDF) shows two peaks

Fig. 6. Original pores in loess soils: (a) overhead structure and overhead pores; (b) interlocking structure and interlocking pores; (c) matrix structure and intragranular pores (the
closed spaces represent soil particles, modified after Lei, 1987).

Table 2
Characteristics of all kinds of pores in loess soils (modified after Lei, 1987).

Type Radius (mm) Characteristics Influence on collapsibility

Secondary Root holes >16 (large pores) Dense walls Depend on whether the pore walls are cemented or
not
Wormholes Loose walls
Mouse holes Large, fragile
Joints and fissures Closed, with rough edges
Karst caves Closed, radial, irregular, with smooth
walls
Original Intergranular Overhead pores 416 (mediate Spaced pores as per Gao (1980a) Significant
pores pores)
Interlocking 14 (small pores) Intergranular pores as per Gao (1980a) Insignificant
pores
Intragranular <1 (micropores) Numerous, irregular distribution Extremely insignificant
pores
262 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Fig. 7. The MIP results of natural loess soil in Shaanxi, China (modified after Jiang et al.,
2014a).
at pore diameters of about 7 mm and 0.025 mm, respectively. This
dual characteristic of the PSDF indicates that there are two main
pore groups in the soil, i.e. inter-aggregate type and intra-aggregate
type (similar to intergranular and intragranular pores defined by
several investigators (Gao, 1980a; Al-Mukhtar et al., 1996)). MIP
was also used for the microstructure investigation of both natural
and compacted loess soils during the stress path tests by Jiang et al.
(2012a, 2014a). Changes in the PSDs due to stress path tests show
that loading leads to a significant change in inter-aggregate pores
while a slight change in intra-aggregate pores, for both natural and
compacted loess soils. Recently, Lommler and Bandini (2015)
studied a collapsible soil in New Mexico, USA that has “pin holes”
of tubular structure of 0.3 mm diameter visible to the naked eye.
These pin holes were recognized as indicator of the collapsibility of
this soil. However, increasing magnification revealed the presence
of honeycomb structure around the pin holes. The honeycomb
structure and highly porous matrix provide a more reasonable
explanation for the high collapsibility of local soil. It is also indi-
cated that macropores contribute little to collapse. Zhao and Chen
(1994) and Zhao et al. (1997) stated that early classifications for
loess soil pores mislead spaced pores with overhead pores. Over-
head pores are different from spaced pores not only in the size but
also in the influence on collapsibility. Both spaced and overhead
pores are formed by abnormal arrangement of coarse particles with
point-contact relation. Spaced pores are typically larger than sur-
rounding particles (i.e. >20 mm). Spaced pores ranging from 20 mm
to 80 mm contribute to heavy collapse, while overhead pores
ranging from 8 mm to 20 mm contribute to medium or weak
collapse. As soil collapses, spaced pores may convert into overhead
pores. Due to this reason, loess soils have been found to collapse
more than once. This is consistent with the results reported by Yang
(1988). In his study, overhead pores and most spaced pores fall in
the group of pores smaller than 54 mm. Table 3 shows the charac-
teristics of each kind of pores and the corresponding collapsibility
level (modified after Zhao and Chen, 1994). Assallay et al. (1997)
divided soil pores into small, medium, large and giant pores,
among which only large pores make a contribution to wetting-
induced collapse. Classifications of soil pores as summarized in
this paper are more fundamental in nature and well defined for
rational explanation of collapse behavior of loess soil deposits.
These classifications suggest that pores with different sizes and
forms have different influences on the collapsibility. For this reason,
void ratio, which is not able to represent pore size and form suffi-
ciently, cannot be used as an index for estimating the collapsibility
of loess soils. For example, the same soils when compacted at dry
Table 3
Combinations of pores and corresponding collapsibility levels (modified after Zhao
and Chen, 1994).
Collapsibility Dominant pore Collapsibility Changes in pores after collapse
level type coefficient
Non-collapse Overhead and <0.015 Not obvious
interlocking
pores
Weak Overhead pores 0.015e0.04 Not obvious
collapse
Medium Overhead pores 0.04e0.07 Obvious decrease in overhead
collapse pores and spaced pores
Strong Overhead pores 0.07e0.1 Obvious decrease in spaced
collapse and spaced pores pores and part of overhead
pores
Extremely Spaced pores >0.1 Obvious decrease in spaced
strong pores
collapse
and wet of optimum conditions have the same dry density but
different collapse behaviors. Extending the same argument, two
soils having the same dry density or void ratio may have different
structures, such as flocculated and dispersed structures (Vanapalli
et al., 1999). This behavior also explains why the relationship be-
tween collapsibility parameters and void ratio derived for collaps-
ible loess soil in one region provides unsatisfactory results when
used for prediction of the collapsibility of loess soil in a different
region.
From the above discussions it is clear that large pores or inter-
aggregate pores (i.e. spaced pores and overhead pores) have a
major influence on collapse behavior of loess soils. Information of
changes in PSDs due to loading and wetting provides better evi-
dence that different kinds of pores contribute differently to collapse
(Yang, 1988; Jiang et al., 2012a, 2014a, b). More studies in this di-
rection would be useful for better understanding of the loess soil
microstructure and influence of soil pores on collapsibility.
3.4. Bonding materials
Bonding strength in loess soils can be attributed to several
sources, such as capillary tension, surface friction and bonding
materials. The influence of bonding materials, which include cal-
cium carbonate and clays in most cases, is described in this section.
The other bonding strength sources, such as capillary tension and
friction, are dependent on the minerals and modes of formation of a
soil (i.e. natural deposition or compaction). Calcium carbonate is
regarded as one of the key bonding materials in loess soils since it is
found not only as film coating on grains but also because of its
concentration at grain contacts. More importantly, loess soils are
always characterized by high carbonate content. Recently, loess
soils were mixed with calcium oxide (i.e. CaO) of different per-
centages, and then water and carbon dioxide (i.e. CO2) were added,
in order to prepare artificial structural loess soil samples for study
of natural loess soil mechanical behavior (Hu et al., 2000, 2004;
Jiang et al., 2012b). Table 4 shows the information of carbonate
content for loess soils from different regions in China (Derbyshire
and Mellors, 1988).
Gao (1980a, 1990, 1994) stated that different calcium carbonate
states (i.e. granular, reborn crystalline and microcrystalline calcium
carbonate) have different effects. However, only microcrystalline
calcium carbonate contributes to bonding strength and affects the
collapsibility of loess soils. For this reason, the calcium carbonate
content cannot be regarded as an index to estimate collapse
behavior. Cabrera and Smalley (1973) proposed the concepts of
long- and short-range bonding. Calcium carbonate bonding may be
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274 263

Table 4 1988; Derbyshire, 2001). The clay coatings are commonly

Carbonate content of loess samples in China (Derbyshire and Mellors, 1988). observed lying parallel to the surface of silts. These coatings were
Site Loess soil Carbonate content (%) neither continuous nor of uniform thicknesses. Clay buttresses
Jiuzhoutai, Gansu Malan 11.19e11.59
comprising silt-size clay aggregations were observed between
Lishi 11.67e15.37 quartz grains (Derbyshire and Mellors, 1988). Similarly, Barden
Wucheng 8.76e16.03 et al. (1973) found that the form of clays in collapsible loess soils
Pingliang, Gansu 16.5 varies depending on the geologic origins and stress history expe-
Qin’an, Gansu 16.8
rienced by the soil. If the clays originally suspend in pore water,
Lanzhou, Gansu 15.43
Luochuan, Shaanxi L4 25.8 then gradual evaporation would cause the retreat of clay plates
L9 23.81 with pore water to the contacts, giving a buttress type of support to
Yulin, Shaanxi 9.14 silts. There are many possible forms of clay materials in loess soils,
Yan’an, Shaanxi 11.3
the nature of clay bonding is complex, and it is not clear how much
Mengxian, Henan Malan 9.25e9.34
is due to electro-chemical effect and due to capillary effect,
respectively (Barden et al., 1973).
Yang (1988) divided bonding materials into water-stable and
non-water-stable materials. The former includes calcium car-
either long- or short-range bonding depending on the amount of
bonate, calcium sulfate, iron oxide, iron hydroxide, water-stable
calcium carbonate bonds at contact. It is the short-range bonding
secondary mica and secondary zeolite, and the latter includes
that has a significant effect on the collapsibility of loess. Guo (1958)
soluble salts, clay materials and secondary mica. Since contents of
found that loess soils wetted by hot water, NaCl and CaCl2 solution
both soluble salts and secondary mica are typically less than 0.1%
respectively show a gradually decreasing collapsibility. It was
in loess soils, clay bonding has a significant contribution to
suggested in his study that salt dissolution weakens bonding
collapse upon wetting. However, if clay grade minerals are the
strength and induces collapse upon wetting by regarding calcium
only bonding materials in loess soils, collapse deformation upon
carbonate as soluble and Ca2þ as the main ion in pore water. It is
wetting should exhibit a degree of plasticity instead of a sudden
unreasonable to regard calcium carbonate as soluble (Smalley,
failure. In summary, calcium carbonate and clay minerals should
1971; Guo et al., 2004, 2005, 2008) and to attribute collapse to
be realized as the key bonding materials that control the
calcium carbonate leaching on soaking. Calcium carbonate leaching
collapsibility of loess soils.
is certainly a slow process that requires many cycles of wetting and
As per the microstructure information summarized, collapse
drying, which cannot explain the collapse phenomenon of sudden
behavior of loess soils is influenced by all four structural factors
decrease in volume upon wetting. This reasoning implies that there
(i.e. particle pattern, contact relation, pore form, and bonding
are other non-water-stable materials contributing to bonding
materials). All the four factors are dependent on each other; for
strength in unsaturated condition except for calcium carbonate.
example, spaced pores and point-contact relation always rein-
Clay minerals (i.e. < 2 mm) have been regarded as dominant
force each other. The pore form and bonding material could be
bonding agents because of their electrically charged characteristic
suggested as the two main important factors that have more in-
and flocculated or flowing-gelatinous form (Barden et al., 1973;
fluence on the collapse behavior of loess soils compared to the
Osipov and Sokolov, 1995; Assallay et al., 1997; Smalley et al.,
other two factors as per the discussions summarized in the paper.
2001). Capillary tension and electrostatic attraction can also be
However, the comprehensive behavior of a collapsible loess soil
attributed to the presence of clay minerals. Considerable amount of
cannot be fundamentally characterized considering one single
clays is typically found in loess soils from China (Derbyshire and
factor. Further investigations on all the four structural factors are
Mellors, 1988) (Fig. 8). The clay grade minerals were also found to
required for rational interpretation of collapse behavior of loess
have three major forms in loess soils, i.e. (i) coatings, (ii) clay bridge
soils.
as described above, and (iii) buttresses (Derbyshire and Mellors,
4. Soil mechanics-based approaches for interpreting the
collapse behavior

Microstructure is widely acknowledged as important infor-

Fig. 8. Grain size distributions of loess soil samples collected from different regions in
China (modified after Derbyshire and Mellors, 1988).
mation required in explaining the collapse mechanism. How-
ever, as discussed earlier, it lacks a simple quantitative
descriptor, which makes the analysis more subjective (Alonso
et al., 1993; Chen et al., 2006; Tian et al., 2011). This has been
addressed by the recently developed image processing pro-
grams, which facilitate the quantitative analysis of soil micro-
structure (Gu et al., 2011). The image processing programs are
mostly based on the principle of binary grey segmentation and
may provide satisfactory results for granular soils. Their appli-
cation for fine-grained soils such as silts and clays has not been
well validated.
The measurable mechanical behaviors which include shear
strength and volume change can be described in terms of the
stress state in the soil for both saturated and unsaturated soils.
The effective stress equation proposed by Terzaghi (1943) is
widely used for interpretation of saturated soils. The same
equation has been extended by Bishop and Blight (1959),
Bishop (1960) for unsaturated soils. Fredlund and Morgenstern
(1977) introduced two stress state variables, i.e. net normal
264 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
stress and matric suction, for rational interpretation of unsat-
urated soils. The volume change of collapsible soils has been
attributed to changes in both stress state variables (Alonso
et al., 1990; Chen, 1999; Chen et al., 1999; Sun et al., 2004,
2007a, b).
It is always believed that collapsible soils experience the
maximum settlement as their degree of saturation approaches
100%. For this reason, the collapse settlement at saturated condition
of soil is of concern and interest for geotechnical engineers. Several
testing methods have been suggested in the literature for esti-
mating collapsibility such as single- and double-oedometer tests
(Jennings and Knight, 1957; Knight, 1963). However, in general,
collapsible soils undergo significant volume change prior to
reaching saturated state. El-Ehwany and Houston (1989) performed
a one-dimensional (1D) infiltration test by applying water with a
small head at one end of the sample. Their test results suggest that
for degrees of saturation above 65%e70%, essentially full collapse
occurs for the tested soil. The relationship between degree of
saturation and ratio of partial collapse to full collapse is shown in
Fig. 9. This result is reasonable because even the soils within the
wetted zone do not reach fully saturated condition (El-Ehwany and
Houston, 1989).
4.1. Understanding collapse behavior from the two stress state
variables approach
Several early researchers extended effective stress principle
to interpret the mechanical behavior of unsaturated soils
(Aitchison, 1960; Bishop, 1960; Jennings, 1961; Newland, 1965;
Richards, 1966; Allam and Sridharan, 1987; Khalili and
Khabbaz, 1998). However, studies, notable as Jennings and
Burland (1962), have shown that a single-valued effective
stress is not capable of completely describing unsaturated
volumetric behavior. The two stress state variables approach
proposed by Fredlund and Morgenstern (1977) provides a more
meaningful description because it is applicable to both shear
strength and volume change behavior of unsaturated soils and
does not use other empirical properties or parameters in the
description of the soil stress state. From the perspective of two
Fig. 9. Partial collapse due to partial wetting (modified after El-Ehwany and Houston,
1989).
stress state variables approach, the triggering mechanism for
collapse is attributed to the loss of strength due to reduction in
matric suction as a result of wetting. In other words, collapse
occurs when there is a change in stress state of the soil as it goes
from unsaturated condition towards a saturated condition
(Fredlund and Gan, 1995).
The stress and deformation state variables can be combined
using suitable constitutive relations proposed by Fredlund and
Morgenstern (1977) for the soil structure, air phase and water
phase. Generally, the constitutive relations for the soil structure and
water phase are used in volume change behavior analysis. The
proposed constitutive relations can be expressed in a compress-
ibility form, which is consistent with the soil mechanics principles
(i.e. Eqs. (1) and (2)). They can also be presented graphically by
plotting the deformation state variables with respect to two inde-
pendent stress state variables in the form of constitutive surfaces
(i.e. the three-dimensional (3D) constitutive surfaces for an un-
saturated soil, as shown in Fig. 10). The constitutive surfaces pro-
vide rational explanation of volume change behavior of
unsaturated soils in terms of two stress state variables, which is
consistent with the continuum mechanics principles. In addition,
they also provide a way to estimate the volume change due to
varying stress state variables. For an isotropic, linear elastic mate-
rial, the coefficients of volume change (i.e. ms1 and ms2 , mw w
1 and m2 )
under various loading conditions could be quickly calculated using
the elasticity modulus of the material. However, the estimation of
these coefficients is much more complex as the soil yields (Alonso
et al., 1990). In addition, these test results also are influenced by soil
type, initial soil properties and the soil stress state (Fredlund and
Gan, 1995). Therefore, simple procedures for estimating these co-
efficients from conventional tests would be valuable for using such
constitutive relationships in modeling the collapse behavior.
d3 v ¼ ms1 dðsmean  ua Þ þ ms2 dðua  uw Þ (1)
dðqÞ ¼ mw
1 dðsmean  ua Þ þ m2 dðua  uw Þ
w
(2)
where 3 v is the volumetric strain; q is the volumetric water content;
ms1 and ms2 are the coefficients of total volume change with respect
to changes in net normal stress and matric suction, respectively;
mw w
1 and m2 are the coefficients of water volume change with
respect to changes in net normal stress and matric suction,
respectively; smeanua is the net mean normal stress; uauw is the
matric suction; smean is the mean normal stress; ua is the pore-air
pressure; uw is the pore-water pressure.
Tadepalli and Fredlund (1991), Tadepalli et al. (1992), Fredlund
and Gan (1995) conducted a series of tests on a statically com-
pacted silt using a modified oedometer apparatus. Their results
show a one-to-one relationship between matric suction and total
volume change due to collapse, as shown in Fig. 11. S1M, S2M, S3M
and S4M represent four specimens that were tested. Since volume
changes are predominantly confined within the wetted zone, only
the coefficient of volume change with respect to changes in matric
suction ms2 , is required. Constant ms2 was used for specimens S1M
and S2M and linearly decreasing ms2 was used for specimens S3M
and S4M in numerical simulations to achieve the best agreement
between measured and predicted total volumes with respect to
changes in matric suction.
Pereira and Fredlund (1997, 2000) conducted a series of suction-
controlled tests on a compacted residual soil. The results show that
at a given net mean stress, the soil goes through three distinct
phases in terms of total deformation following a wetting stress path
(Fig. 12). The first phase (pre-collapse phase) occurs when the soil is
subjected to high values of matric suction. In this phase, the soil
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Fig. 10. 3D constitutive surfaces for an unsaturated soil: (a) soil structure constitutive surface; (b) water phase constitutive surface (modified after Fredlund and Rahardjo, 1993).
undergoes small volumetric deformations in response to matric
suction decrease, no slippage occurs between the grains and the
soil structure remains intact. The second phase (collapse phase)
occurs as the soil experiences intermediate values of matric suc-
tion. In this phase, the soil suffers a significant volume decrease in
response to matric suction decrease. The soil structure is altered
due to bonding break. The third phase (post-collapse phase) occurs
as the unsaturated soil approaches full saturation. In this phase, the
soil does not undergo further volume decrease in response to
matric suction decrease. A similar collapse behavior was found in
many studies on both compacted soils and natural collapsible soils
(Kato and Kawai, 2000; Sun et al., 2007a; Garakani et al., 2015).
Based on the experimental observations, Pereira and Fredlund
(1997, 2000) proposed a model for predicting the volume change
due to collapse at a given net mean stress. Six curve-fitting pa-
rameters including d1, d2, ei, ef, (ua  uw)c, (ua  uw)f were involved
(see Fig. 12), where d1 is the slope of volumetric deformation in the
pre-collapse phase, d2 is the slope of volumetric deformation in the
collapse phase, ei is the initial void ratio (i.e. before the wetting), ef
is the final void ratio (i.e. after the complete saturation), (uauw)c is
the critical matric suction below which the soil structure starts
collapsing, (uauw)f is the final matric suction below which the soil
structure stops collapsing, and qs is the saturated volumetric water
content. Although Pereira and Fredlund’s (1997, 2000) model pro-
vides a smooth void ratio state surface with respect to matric
Fig. 11. Variation of total volume with respect to matric suction for four different soils
by Fredlund and Gan (1995).
265
suction, the relationships between parameters of the model and net
mean stress were mostly derived using the fitting method. This
method may not be suitable for higher net mean stress since some
assumptions are only valid for net mean stress values lower than
200 kPa. However, an insight into the shape of Pereira and Fred-
lund’s (2000) model indicates that the collapse behavior in terms of
void ratio with respect to changes in matric suction is reverse to the
SWCC, especially the wetting curve (see Fig. 12). The three phases of
collapse, i.e. pre-collapse phase, collapse phase and post-collapse
phase would correspond to the three zones on the wetting SWCC
(i.e. residual zone, transition zone and capillary saturation zone)
(after Vanapalli et al., 1996, 1999). For this reason, it is very likely
that relationships can be found between the parameters in Pereira
and Fredlund’s (2000) model and special values on the wetting
SWCC. Such a modified model may have wider application in en-
gineering practice.
4.2. Understanding collapse behavior from the elastoplastic
modeling
The widely used model that was firstly presented in a quanti-
tative form by Alonso et al. (1990) is referred to as the Barcelona
basic model (BBM) in the literature. Since then, several modified
forms of the BBM were proposed (Gens and Alonso, 1992; Josa et al.,
Fig. 12. Volume change behavior of collapsible soils during the wetting process
(modified after Vanapalli et al., 1996 and Pereira and Fredlund, 2000).
266 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
1992; Wheeler and Sivakumar, 1995; Sun et al., 2003). More
recently, some models were proposed in which the SWCC was
incorporated into the stress-strain constitutive relationships to
simulate the nonlinear variation of unsaturated soil properties
(Gallipoli et al., 2003; Wheeler et al., 2003; Sun et al., 2007c; Thu
et al., 2007a, b). The validity of all these elastoplastic models was
tested using several suction-controlled compression tests con-
ducted on unsaturated soils (Cui and Delage, 1996; Chen, 1999;
Chen et al., 1999; Rampino et al., 2000; Sun et al., 2004, 2007a,
b). The extensions or modifications of the BBM make it more
convenient to apply the BBM for general problems associated with
unsaturated soils (Sheng et al., 2008). In the BBM, a yield curve
representing the locus of yield points in the (p, s) space was
formulated which was labeled as LC (after loading-collapse) yield
curve, where p stands for either net normal or net mean stress
depending on the type of test considered, and s stands for suction.
When the stress path crosses the yield curve, irreversible volu-
metric strain or collapse occurs. The irreversible volumetric strains
may be due to suction decrease or to load increase; both suction
decrease and load increase will have a similar effect on the moving
LC yield curve. Such a behavior is consistent with the observations
of Matyas and Radhakrishna (1968) who suggested that wetting
has a two-fold effect on soil structure, i.e. a reduction in inter-
granular stress and a reduction in the rigidity. Collapse happens
when the volume decrease due to rigidity decrease exceeds the
volume increase due to intergranular stress decrease; however, the
reverse situation results in swelling. The yield locus with respect to
changes in suction is referred to as SI (after suction increase) yield
curve. Therefore, the complete framework for describing the
volumetric behavior of an unsaturated soil should be represented
by two yield curves LC and SI bounding the elastic region (see
Fig. 13). The irreversible strains or collapse deformations due to
crossing the SI or LC surface would indicate a change in the soil
structure that affects the position of the LC or SI curves. Alonso
et al.’s (1990) model is versatile and capable of estimating volume
change during wetting (swell or collapse depending on the
magnitude of initial suction and net mean stress); however, a large
number of parameters are required from suction-controlled tests.
Due to this reason, such a model has limitations for wider use in
conventional geotechnical engineering practice.
Chen (1999) and Chen et al. (1999) conducted a series of wetting
tests on an unsaturated, compacted loess soil from China. The shape
of the LC curve obtained from the isotropic compression tests
(Fig. 14) supports the concept of the LC model in Alonso et al.
(1990). The compressibility indices associated with isotropic
compression tests with respect to soil structure are relatively
Fig. 13. Loading-collapse (LC) and suction increase (SI) yield behavior of an unsatu-
rated soil (Gens and Alonso, 1992).
Fig. 14. Loading-collapse yield curve for compacted loess soils (Chen et al., 1999).
constant with increasing suction. However, the compressibility
indices corresponding to water volume change vary significantly in
the low suction range. As per the test results, yield due to increasing
suction was found to be dependent on the initial soil density and
net mean stress. If the initial density is low, then yield may occur at
a low suction during drying. However, higher suction will be
needed to cause yield by drying for soil specimens having high
initial density. Similarly, for wetting case, it is easier for collapsible
loess soils with lower initial density to collapse than that with
higher initial density. In other words, a larger matric suction
decrease is required for loess soils with higher initial density to
collapse by wetting. In addition, it is found that if the compress-
ibility of soil structure decreases significantly due to LC yielding, the
soil will be insensitive to subsequent increase in suction and may
exhibit elastic behavior. In other words, a much higher yield suction
is required, which could be attributed to the expansion of yield
surface. Consistently, for wetting case, if the soil structure is com-
pressed significantly due to loading, then the soil becomes denser
and the pores smaller, making it insensitive to suction decrease due
to wetting. For this reason, it can be postulated that suction in-
crease and suction decrease (i.e. drying and wetting) have similar
effect on the structure of loess soils.
Kato and Kawai (2000) were among the first to study the
collapse behavior using suction-controlled triaxial tests. They
found the dependence of collapse deformation on stress path could
be interpreted using the expansion of the yield surface due to
wetting or loading. The loading-wetting sequence may influence
the collapse deformation (collapse here is not limited to wetting-
induced collapse) because of the difference between both elastic
and plastic stiffness parameters for changes in net mean stress and
that for changes in matric suction. For example, soil specimen that
is loaded first and then wetted may yield due to wetting, while the
soil specimen wetted first and then loaded may yield due to
loading. The difference between stiffness parameters results in the
difference between collapse deformations induced under two
stress paths. In general, the elastic stiffness parameters associated
with changes in net mean stress and in matric suction are assumed
to be independent of matric suction and net mean stress, respec-
tively (Alonso et al., 1990; Wheeler and Sivakumar, 1995). Due to
this reason, collapse deformation may not be affected by the
loading-wetting sequence if the stress state is confined to elastic
region. However, many researchers have experimentally proved
that the volume change is independent of stress path for natural
loess soils, where volume change is not limited to that induced by
wetting (Zheng and Zhang, 1989; Jiang et al., 1999).
While Kato and Kawai (2000) tests were limited to single den-
sity, Sun et al. (2004, 2007a) performed a series of wetting tests on
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
a compacted clay with different initial densities. They found that
the wetting-induced collapse deformations mainly depend on the
initial void ratio and net mean stress under which the soil is wetted,
irrespective of the matric suction that was initially imposed on the
soil. The collapse behavior of compacted clay can be explained
using the BBM. Later, Sun et al. (2007b) investigated the effects of
initial density and stress states on the wetting-induced collapse of
compacted clay. The collapse behavior with respect to initial den-
sity from isotropic compression tests is shown in Fig. 15. The
volumetric strain and shear strain due to collapse with respect to
initial density in triaxial tests are shown in Fig. 16. From Sun et al.
(2007b) tests, it is also found that the volume contraction due to
wetting is more sensitive to changes in the degree of saturation
than changes in the suction. These results suggest the possibility to
predict the wetting-induced collapse behavior from the degree of
saturation, using the SWCC as a tool.
4.3. Studies conducted on natural loess soils for interpreting their
collapse mechanism
The soil mechanics-based approaches for interpreting the
collapse behavior discussed above were mostly proposed based on
the experimental results of compacted soils, which may show
different collapse behaviors from natural loess soils. Hu et al.
(2000) highlighted the difference between the volume change
behavior of natural and compacted loess soils from the results of a
set of oedometer tests on both natural and compacted loess soil
specimens. Their test results are shown in Fig. 17, in which the after-
collapse curves of initially unsaturated loess soils coincide with the
compression line of saturated natural loess soil. These results
suggest that wetting-induced collapse deformation is independent
of the loading-wetting sequence. This means that both single- and
double-oedometer tests would yield similar results about the
collapsibility of loess soils. This also illustrates the difference be-
tween the mechanical behavior of natural and compacted loess
soils. Similar conclusion was arrived at by Jiang et al. (2012a, b)
from different stress path tests on artificial loess soil specimens
(including single- and double-oedometer tests, strain- and stress-
controlled triaxial tests, etc.). They attributed this difference to
the different sources of bonding strength for natural loess soils and
compacted or remolded loess soils. For this reason, it is implied that
models proposed for compacted soils may not be reliable to explain
the collapse behavior of natural loess soils although most of these
Fig. 15. Collapse behavior of compacted soil samples with different initial void ratios
under different isotropic stresses (Sun et al., 2007b, e0 means the initial void ratio).
267
Fig. 16. Collapse behavior of compacted soil samples with different initial void ratios
under different stress ratios from triaxial tests (the net mean stress is 196 kPa,
modified after Sun et al., 2007b; R means the stress ratio).
models have been validated for collapsible soils. On the other hand,
because of the difficulties associated with obtaining “identical”
natural loess soil specimens and the variability in natural soil
properties, experimental studies on compacted or remolded soils
are valuable for understanding the mechanical behavior of natural
loess soils. In addition, artificial loess soil samples which have been
experimentally proved to behave similar to natural loess soils
compensate this limitation (Hu et al., 2000; Jiang et al., 2012a, b).
Many researchers who have been involved in the studies on
natural loess soils proposed the concept of structural strength,
which is associated with bonding and mainly influenced by initial
water content. Some researchers suggested the structural strength
of natural loess soils as the preconsolidation stress (i.e. yield stress)
(e.g. Hu et al., 2000). Fig. 18a shows the variation of structural
strength with respect to initial water content for natural loess soils
in Shaanxi in China modified after Hu et al. (2000). They attributed
collapse to the loss of structural strength and bonding failure as a
result of wetting. Hu et al. (2004) proposed a constitutive rela-
tionship for relating collapse deformation to water content after
wetting based on experimental results, see Eq. (3), in which the two
parameters are independent of the loading condition (i.e. confining
pressure and stress level) and are required to be determined from
triaxial wetting tests. However, most researchers defined the
structural strength of natural loess soils as the difference in peak
Fig. 17. Comparison between the compression lines of natural and compacted loess
soils at natural water content condition (modified after Hu et al., 2000).
268 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274

Fig. 18. The structural strength with respect to initial water content of natural loess soils from China: (a) determined from the first method (Hu et al., 2000), (b) determined from the
second method (Zhang et al., 1994; Dang and Hao, 1998; Liu et al., 2008).

strength between natural and compacted loess soils. Fig. 18b shows
the variation of structural strength with respect to initial water
content for natural loess soils from different regions in China
(Zhang et al., 1994; Dang and Hao, 1998; Liu et al., 2008). The
structural strength has been found to be influenced not only by the
initial water content but also by the stress applied on the soil, grain
size distribution, chemical composition, environmental and climate
conditions for loess soil deposition. Fig. 18 highlights that the
structural strength of natural loess soils determined from the first
method (i.e. preconsolidation stress) is much larger than that
determined from the second method (i.e. difference in peak
strength between natural and compacted loess soils). Recently,
several studies were undertaken to study the relation between
structural strength and yield stress or shear strength for natural
loess soils (e.g. Dang and Li, 2001; Chen, 2008; Tian et al., 2011; Luo
et al., 2014).
w  w0
D3 v ¼ av D3 vf
D3 vf ðws  w0 Þ þ av ðw  w0 Þ
where ws and w0 are the saturated and natural water contents,
respectively; D3 vf is the final collapse deformation; av is the initial
slope of the D3 v with respect to Dw curve.
Shao et al. (2006, 2010, 2014) put forward a structural param-
eter, so-called the “structural index”, for structural soils, especially
for natural loess soils. They stated that the natural soil structure can
change due to disturbance, loading and wetting. Shear strength,
unconfined compressive strength or deformation between natural
and compacted soil specimens could indicate the bonding effect
and particle interlocking arrangement in natural soil structure,
while the properties between natural and saturated natural soil
specimens could indicate the effect of soluble salt bonding and
capillary tension in natural soil structure. The structural index is
defined as the product of strength reduction due to disturbance and
that due to wetting:
the relationships between the structural index and conventional
properties of natural loess soils based on test results of natural,
compacted and saturated loess soil specimens. They found that soil
properties associated with grading, density and humidity (such as
Atterberg limits, water content, dry density and void ratio) have
significant influence on the structural index (see Fig. 19), while soil
with the same structural index may have different values of such
conventional properties. For this reason, the structural index
together with conventional properties would be more reasonable
and accurate for estimating the mechanical behavior of a structural
soil. Their work is valuable for interpretation of mechanical
behavior, especially strength behavior, of natural loess soils. Jiang
et al. (2012b) proposed the concepts of structural damage stress
(i.e. SDS, the structural yield stress of saturated soil) and general-
ized structural damage stress (i.e. GSDS, the structural yield stress
of unsaturated soil), both of which could be determined from
confined compression tests. Wetting-induced deformation is rela-
tively small when the vertical pressure is less than SDS, and it in-
(3)
creases rapidly once the vertical pressure exceeds SDS and
decreases after it reaches the peak value of GSDS.
Shen (1993) and his group were the earliest investigators to
interpret the mechanical behavior of natural loess soils using the
concepts of elastoplastic breakage mechanics as a tool. They stated
that the strength decrease and volume change of loess soils are the
results of transition from natural structural soils to remolded soils;

ðs1  s3 Þy ðs1  s3 Þy
ms ¼ (4)
ðs1  s3 Þr ðs1  s3 Þs

where ms is the structural index; s1s3 is the maximum deviatoric

stress during shearing process; subscripts y, r and s represent
natural, compacted and saturated states, respectively. The first
component on the right side of the equation is also called load-
disturbance sensitivity and the second component is called Fig. 19. Relationship between the structural index and composite physical index
water-immersion sensitivity. Shao et al. (2006, 2010, 2014) studied (modified after Shao et al., 2014).
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
during this process, natural structure fails and secondary structure
develops. Based on the concepts of breakage mechanics, the binary
medium model and the block structural model were proposed
particularly for natural loess soils.
As per the studies of Shen (2003), Shen and Deng (2003) and
Shen and Hu (2003), natural loess soil structure is regarded as being
composed of two elements in the binary medium model, i.e.
cemented block and weak zone. The former element is actually the
aggregate in natural loess soils that are firmly cemented by bonding
materials (i.e. clay platelets and calcium carbonate) and the latter
element is the zone among aggregates that has pore space (i.e.
inter-aggregate pores) and separate particles (i.e. with no or very
small amount of bonds at the contacts). For this reason, when a
stress is applied on the soil, both elements are responsible for
bearing and transferring the stress (see Eq. (5)), and deformations
occur in both elements in response to changes in stress state in the
soil. However, since the aggregates are well cemented and the pore
space within the aggregates is relatively small that could be
neglected, the volume change of natural loess soils due to loading
or wetting is predominantly associated with the plastic de-
formations that occur in weak zones (see Eq. (6)). In the binary
medium model, nine parameters are required for modeling the
mechanical behavior of natural loess soils due to loading or wet-
ting, which could be determined from compression tests, triaxial
tests, etc. on both natural and remolded loess soil samples. The
model has been found to be reasonable for estimation of the me-
chanical behavior (i.e. shear strength and volume change) for nat-
ural loess soils. A good agreement between the measured and
predicted mechanical behavior using the binary medium model for
natural loess soils in Shaanxi in China is presented in Fig. 20.
Although it is easy to understand the process of natural loess soil
structure failure due to loading and wetting using this model, a
large number of tests are required to determine the parameters of
the model.
n o
fsg ¼ ð1  bÞfsi g þ b sf
sm ð1 þ kÞ
3v ¼ cc ln
s0 m0
where {s} is the total stress applied on the soil; {si} is the stress
born by cemented blocks; {sf} is the stress born by weak zones; b is
the damage parameter; 3 v is the volumetric strain; cc is the
parameter associated with the compression index of natural loess
soils; sm is the mean stress; s0 m0 ¼ sm0 =Snr , sm0 is the reference
Fig. 20. Predicted stress-strain behavior and volume change behavior of natural loess
soils in Shaanxi in China using binary medium model (modified after Shen and Hu,
2003).
269
stress at zero volumetric strain, Sr is the degree of saturation, n is
the parameter of the model; k is the parameter of the model.
The block structural model treats the natural loess soil structure
as an elastic block, in which bonding agents among particles or
within aggregates are randomly distributed (Hu, 2000; Hu et al.,
2005). When a stress is applied on the soil structure, contacts
which have few bonds fail firstly as the stress exceeds the bonding
strength at these contacts, resulting in breakage of primary large
soil block into several small blocks. The block theory that was
originally proposed for rock mass was extended to model the vol-
ume change process of natural loess soils. The weakly bonded
contacts in loess soils were considered as elements similar to
structural surfaces in rock mass. However, almost all the contacts
including the ones that are strongly bonded or within the aggre-
gates in loess soils would fail as stress increases to considerably
high level. The failure of weak bonding and breakage of the primary
soil block were assumed to be elastic, while the subsequent slip of
small blocks over each other and further breakage of small blocks
were assumed to contribute to plastic deformations. For such a
scenario, the stress-strain behavior could be represented using the
following equation:
   
vf vg
fD3 g ¼ ½CfDs0 g þ A1 Df þ A Dg (7)
vs0 2
vs*
where {3 } is the total strain due to loading and wetting; [C] is the
elastic flexibility matrix; {s0 } and {s*} are the effective stress and
the net stress, respectively; f and g are the yield and damage
functions, respectively, in which matric suction was involved; A1
and A2 are the plastic parameters corresponding to yielding and
damage, respectively.
On the right hand side of Eq. (7), the first part is due to the elastic
deformations, the other two are associated with the plastic de-
formations due to the slip of small blocks over each other and their
breakage, respectively. The model involves six parameters which
(5)
are influenced by the damage parameter or loading condition. In
general, both confined and unconfined compression tests, wetting
tests and triaxial undrained tests on natural loess soil specimens are
(6) required for determining these parameters. The model was vali-
dated for natural loess soil in Shaanxi in China, and a comparison
between the measured and predicted volume change behavior is
shown in Fig. 21. It is shown that the model is capable of modeling
the volume change behavior of natural loess soils under relatively
higher confining pressures (i.e. 200 kPa and 300 kPa), while the
deformations are overestimated under lower confining pressures
(i.e. 50 kPa and 100 kPa). Both the binary medium model and block
structural model provide reasonable predictions of the mechanical
behavior associated with loading and wetting for natural loess soils;
however, they were rarely used in conventional geotechnical en-
gineering practice since cumbersome tests (such as the oedometer
tests and triaxial tests) are required for determining the various
parameters. For this reason, much simpler models with less pa-
rameters or parameters that are easy-to-determine from conven-
tional tests are suggested for future researches for interpreting and
predicting the mechanical behavior of collapsible loess soils.
Collapse surface has been proposed particularly for natural loess
soils during the last two decades (Zhang and Zheng, 1990; Shen,
1993; Hu et al., 2000, 2004; Jiang et al., 2012a, b), which is similar
to the yield surface in elastoplastic models for judging whether
collapse or irreversible deformations will occur. Zhang and Zheng
(1990) proposed the concept of moistening deformation, suggest-
ing that significant volume contraction would occur when there is a
soil water content increase rather than the soil getting fully satu-
rated. In such a scenario, the deformation is controlled by two
270 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Fig. 21. Predicted volume change behavior of natural loess soils in Shaanxi in China using block structural model (modified after Hu, 2000).
external factors, i.e. water and pressure. Therefore, the collapse
surface for natural loess soils would be presented using a 3D plot
with two abscissas representing the water content and volumetric
strain, and the ordinate representing the vertical pressure under
which the soil is wetted, as shown in Fig. 22a. This figure clearly
Fig. 22. Collapse surface for natural loess soils: (a) proposed by Zhang and Zheng
(1990); (b) proposed by Jiang et al. (2012a).
shows the deforming development due to loading and wetting. Shen
(1993) stated that there is an identifiable collapse surface for natural
loess soils. When the stress path is within the collapse surface, de-
formations are almost elastic that could be modeled using elastic
models. When the stress path crosses the collapse surface, the soil
would show high compressibility, and soil collapses. In Hu et al.
(2000, 2004) tests, since all the compression lines after collapse
due to free access to water coincide with the compression line of
initially saturated loess soil, they presumed that there would be a
collapse surface for natural loess soils. Such a collapse surface could
be defined from the compression line of natural saturated loess soils.
However, as per Jiang et al. (2012a, b) study results of triaxial wetting
tests on artificial loess soil specimens, it can be concluded that there
exists a particular collapse surface, which is denoted as the dashed
curve as shown in Fig. 22b. The vectors of total wetting-induced
strains under different loading conditions are shown in p-q space
in Fig. 22b (p is the mean stress; q is the deviatoric stress; Rs is the
stress level, which is the ratio of the deviatoric stress to its maximum
value or the value when axial strain reaches 10%; 3 1f is the axial
strain; Tw1 to Tw9 represent different stress paths for triaxial tests,
as summarized in the inset table in Fig. 22b). The horizontal and
vertical components of the vector represent the total wetting-
induced volumetric strain D3 tv , and deviatoric strain, D3 ts , respec-
tively. The proportion of vertical to horizontal components increases
with stress level and the surface perpendicular to these vectors is
approximately oval in shape. The wetting-induced strain is negli-
gible when a specimen is wetted at a stress point inside the surface,
while it becomes significant when the soil is wetted beyond the
surface. Such a collapse surface could be defined from the results of
wetting tests under different stress levels.
5. Summary and conclusions
The collapse triggering mechanism associated with wetting
received much concern from various soils researchers during the
past six decades all over the world. The discussions and contribu-
tions on collapse behavior from the literature are summarized in
this paper under three categories, i.e. traditional, microstructure
and soil mechanics-based approaches, with special reference to
loess soil deposits.
Traditional approaches have been found to be unsatisfactory to
universally explain the collapse behavior for all loess soils. These
approaches however were valuable to better understand collaps-
ibility of local loess soils from simple tests. Soil microstructure is
widely acknowledged to play an important role in controlling the
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
mechanical behavior of loess soils. The microstructure of loess soils
can be analyzed in terms of four factors, i.e. particle pattern, contact
relation, pore form and bonding material. Among these factors,
pore form and bonding material are suggested as the two dominant
factors that have more influence on the collapse behavior. During
the last two decades, a large number of researchers have devoted to
classifying soil pores and exploring the water stability of possible
bonding materials in order to predict the collapsibility of a loess soil
from its microstructure. However, the microstructure approach
lacks basic quantitative descriptor, in spite of significant advances
have been made in image processing techniques during the last
decade for quantitative analysis of the soil microstructure. Ap-
proaches extending the mechanics of unsaturated soils, not only
explain the collapse as a result of changes in stress state variables,
but provide a precise way to predict volume change due to collapse
by suitable constitutive relations. Elastoplastic models define the
yield surface for unsaturated soils, which clearly divide elastic and
plastic deformations of an unsaturated soil. These models can also
be used as tools to explain the collapse phenomenon as soil yields
or the stress path crosses the yield surface due to either loading or
wetting. Approaches based on the concepts of elastoplastic
breakage mechanics (i.e. the binary medium model and the block
structural model) have also been validated to provide reasonable
prediction of collapse behavior for natural loess soils. However,
there are some limitations for the application of these models in
conventional practice mainly because of the difficulty in deter-
mining the parameters required for the models from time- and
resource-consuming experimental tests. For these reasons, much
simpler models with less parameters or parameters that are easy-
to-determine from conventional tests are suggested for future
research studies to better understand the mechanical behavior of
natural loess soils. Such studies would be valuable for conventional
geotechnical engineering practice applications.
Conflict of interest
We wish to confirm that there are no known conflicts of interest
associated with this publication and there has been no significant
financial support for this work that could have influenced its
outcome.
Acknowledgments
The first author gratefully acknowledges her appreciation to the
Chinese Scholarship Council, which funded her Joint PhD research
program. The second author thanks the support from Natural
Sciences and Engineering Research Council of Canada (NSERC) for
his research programs. The third author thanks the Chinese Min-
istry of Science and Technology for supporting his research pro-
gram (grant No. 2014CB744701).
References
Aitchison GD. Relationships of moisture stress and effective stress functions in
unsaturated soils. In: Proceedings of conference of the British national society
of the international society for soil mechanics and foundation engineering,
London, UK; 1960.
Alfi AAS. Mechanical and electron optical properties of a stabilized collapsible soil
in Tucson, Arizona. PhD Thesis. Arizona, US: The University of Arizona; 1984.
Alonso EE, Gens A, Josa A. A constitutive model for partially saturated soils. Géo-
technique 1990;40(3):405e30.
Alonso EE, Gens A, Lloret A. The landslide of Cortes de Pallas, Spain. Géotechnique
1993;43(4):507e21.
Allam MM, Sridharan A. Stresses present in unsaturated soils. Journal of Geotech-
nical Engineering 1987;113(11):1395e9.
Al-Rawas AA. State-of-the-art review of collapsible soils. Science and Technology
Review 2000:115e35.
271
Al-Mukhtar M, Belanteur N, Tessier D, Vanapalli SK. The fabric of a clay soil under
controlled mechanical and hydraulic stress states. Applied Clay Science
1996;11:99e115.
Assallay AM, Rogers CDF, Smalley IJ. Formation and collapse of meta-stable particle
packings and open structures in loess deposits. Engineering Geology
1997;48(1e2):101e15.
Ayadat T, Hanna A. Prediction of collapse behavior in soil. Revue Europeenne de
Genie Civil 2007;11(5):603e19.
Barden L. Consolidation of compacted and unsaturated clays. Géotechnique
1965;15(3):267e86.
Barden L, McGown A, Collins K. The collapse mechanism in partly saturated soil.
Engineering Geology 1973;7(1):49e60.
Basma AA, Tuncer ER. Evaluation and control of collapsible soils. Journal of
Geotechnical Engineering 1992;118(10):1491e504.
Beckwith GH, Hansen LA. Calcareous soils of the southwestern United States. West
Conshohocken, USA: ASTM International; 1982.
Bishop AW, Blight GE. Some aspects of effective stress in saturated and partly
saturated soils. Géotechnique 1959;13(3):177e97.
Bishop AW. The principle of effective stress. Teknisk Ukeblad 1960;106(39):859e63.
Cabrera JG, Smalley IJ. Quickclays as products of glacial action: a new approach to
their nature, geology distribution and geotechnical properties. Engineering
Geology 1973;7(2):115e33.
Chen CL, Gao P, Hu ZQ. Moistening deformation characteristic of loess and its
relation to structure. Chinese Journal of Rock Mechanics and Engineering
2006;25(7):1352e60 (in Chinese).
Chen KS, Sha AM. Microstructural characteristics of compacted loess under
different water contents. Highway 2009a;11:103e6 (in Chinese).
Chen KS, Sha AM. Research on microstructure characteristics of compacted loess
based on digital image processing technique. Highway 2009b;26(4):152e7 (in
Chinese).
Chen L. Relationship between structural strength and structural yield pressure of
loess. MS Thesis. Xi’an, China: Northwest A & F University; 2008 (in Chinese).
Chen ZH, Fredlund DG, Gan JKM. Overall volume change, water volume change, and
yield associated with an unsaturated compacted loess. Canadian Geotechnical
Journal 1999;36:321e9.
Chen ZH. Deformation, strength, yield and moisture change of a remolded unsat-
urated loess. Chinese Journal of Geotechnical Engineering 1999;21(1):82e90 (in
Chinese).
Clevenger WA. Experiences with loess as foundation material. Transactions of the
American Society of Civil Engineering 1958;123(1):151e69.
Cui YJ, Delage P. Yielding and plastic behavior of an unsaturated compacted silt.
Géotechnique 1996;46(2):291e311.
Dang JQ, Hao YQ. Effect of water content on the structure strength of loess. Water
Resources and Water Engineering 1998;9(2):15e9 (in Chinese).
Dang JQ, Li J. The structural strength and shear strength of unsaturated loess.
Journal of Hydraulic Engineering 2001;7:79e83 (in Chinese).
Delage P, Audiguier M, Cui YJ, Howat MD. Microstructure of a compacted silt. Ca-
nadian Geotechnical Journal 1996;33(1):150e8.
Derbyshire E, Mellors TW. Geological and geotechnical characteristics of some loess
and loessic soils from China and Britain: a comparison. Engineering Geology
1988;25(2e4):135e75.
Derbyshire E. Geological hazards in loess terrain, with particular reference to the
loess regions of China. Earth-Science Review 2001;54(1e3):231e60.
Dijkstra TA, Rogers CDF, Smalley IJ, Derbyshire E, Li YJ, Meng XM. The loess of north-
central China: geotechnical properties and their relation to slope stability. En-
gineering Geology 1994;36(3e4):153e71.
Dijkstra TA, Smalley IJ, Rogers CDF. Particle packing in loess deposits and the
problem of structure collapse and hydroconsolidation. Engineering Geology
1995;40(1e2):49e64.
Dudley JH. Review of collapsing soils. Journal of Soil Mechanics and Foundations
Division 1970;97(SM3):925e47.
El-Ehwany M, Houston SL. Settlement and moisture movement in collapsible soils.
Journal of Geotechnical Engineering 1989;116(10):1521e35.
Fan HR, Guo R. Influencing factors of water-collapsible loess of Guanzhong area.
Journal of Xi’an University of Science and Technology 2003;23(2):160e3 (in
Chinese).
Fang XW, Shen CN, Li CH, Wang L, Liu HJ. Quantitative analysis of microstructure
characteristics of Pucheng loess in Shaanxi Province. Chinese Journal of Rock
Mechanics and Engineering 2013a;34(5):1319e24 (in Chinese).
Fang XW, Shen CN, Wang L. Research on microstructure of Q2 loess before and after
wetting. Rock and Soil Mechanics 2013b;34(5):1319e24 (in Chinese).
Feda J. Colloidal activity, shrinking and swelling of some clays. In: Proceedings of
the Soil Mechanics Seminar, Illinois, United States; 1964. p. 531e64.
Feda J. Structural stability of subsident loess soil from Praha-Dejvice. Engineering
Geology 1966;1(3):201e19.
Feda J. Collapse of loess upon wetting. Engineering Geology 1988;25(2e4):263e9.
Feda J. Mechanisms of collapse of soil structure. In: Genesis of Properties of
Collapsible Soils. NATO ASI Series, vol. 468. Dordrecht, the Netherlands: Kluwer
Academic Publishers; 1995. p. 149e72.
Fredlund DG, Morgenstern NR. Stress state variables for unsaturated soils. Journal of
the Geotechnical Engineering Division 1977;103(5):447e66.
Fredlund DG, Rahardjo H. Soil mechanics for unsaturated soils. New York, USA: John
Wiley and Sons; 1993.
272 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Fredlund DG, Gan JKM. The collapse mechanism of a soil subjected to one-
dimensional loading and wetting. In: Genesis of Properties of Collapsible
Soils. NATO ASI Series, vol. 468. Dordrecht, the Netherlands: Kluwer Academic
Publishers; 1995. p. 173e205.
Fu H. Treatment of foundation of wet sinking loess. City Bridge and Flood 2005;1:
97e8 (in Chinese).
Gaaver KE. Geotechnical properties of Egyptian collapsible soils. Alexandria Engi-
neering Journal 2012;51(3):205e10.
Gallipoli D, Gens A, Sharma R, Vaunat J. An elasto-plastic model for unsaturated soil
incorporating the effects of suction and degree of saturation on mechanical
behavior. Géotechnique 2003;53(1):123e35.
Gao GR. Study of the microstructures and the collapse mechanism in loess soil from
Lanzhou. Journal of Lanzhou University 1979;6:123e34 (in Chinese).
Gao GR. The microstructures of loess in China. Chinese Science Bulletin 1980a;20:
945e8 (in Chinese).
Gao GR. Classification for microstructure of loess and its collapsibility. Chinese
Science Bulletin 1980b;12:1203e12 (in Chinese).
Gao GR. Classification of microstructure of loess in China and their collapsibility.
Scientia Sinica 1981;7:962e70.
Gao GR. Microstructure of loess soil in China relative to geographic and geologic
environment. Acta Geological Sinica 1984;3:265e74 (in Chinese).
Gao GR. A structure theory for collapsing deformation of loess soils. Chinese Journal
of Geotechnical Engineering 1990;12(4):1e10 (in Chinese).
Gao GR. The formation of collapsibility of loess soils in China. Journal of Nanjing
Architectural and Civil Engineering Institute 1994;2:1e8 (in Chinese).
Garakani AA, Haeri SM, Khosravi A, Habibagahi G. Hydro-mechanical behavior of
undisturbed collapsible loessial soils under different stress state conditions.
Engineering Geology 2015;195:28e41.
Gens A, Alonso EE. A framework for the behavior of unsaturated expansive clays.
Canadian Geotechnical Journal 1992;29(6):1013e32.
Grabowska-Olszewska B. SEM analysis of microstructures of loess deposits.
Bulletin of the International Association of Engineering Geology 1975;11(1):
45e8.
Grabowska-Olszewska B. Engineering geological problems of loess in Poland. En-
gineering Geology 1988;25(2e4):177e99.
Guan WZ. On soluble salts and the mechanism of loess collapsibility. Journal of
Guilin College Geology 1986;6(3):271e8 (in Chinese).
Guan WZ. The collapse types and characteristics of loess in Xi’an and Lanzhou.
Journal of Guilin College of Geology 1989;9(1):74e80 (in Chinese).
Guo JY. Study on the reasons for collapse. Hydrogeology and Engineering Geology
1958;4:7e11 (in Chinese).
Guo YW, Kato MP, Song F, Zhang YL, Zeng SW, Wang DK. Composition of loess
aggregate and its relationship with CaCO3 on the Loess Plateau. Acta Pedologica
Sinica 2004;41(3):362e8 (in Chinese).
Guo YW, Song F, Kato MP. EDX of CaCO3 distribution in loess. Chinese Journal of
Geotechnical Engineering 2005;27(9):1004e7 (in Chinese).
Guo YW, Zhang YL, Dang XL, Kato M. Behaviors of CaCO3 in loess collapse caused by
irrigation. Acta Pedologica Sinica 2008;45(6):1034e9 (in Chinese).
Gu TF, Wang JD, Guo L, Wu DL, Li KC. Study of Q3 loess microstructure changes
based on image processing. Chinese Journal of Rock Mechanics and Engineering
2011;30(Suppl. 1):3185e92 (in Chinese).
Handy RL. Collapsible loess in Lowa. Soil Science Society of America Journal
1973;37(2):281e4.
Habibagahi G, Mokhberi M. A hyperbolic model for volume change behavior of
collapsible soils. Canadian Geotechnical Journal 1998;35(2):264e72.
Houston SL, Houston WN, Spadola DJ. Prediction of field collapse of soils due to
wetting. Journal of Geotechnical Engineering 1988;114(1):40e58.
Houston SL, Houston WN, Zapata CE, Lawrence C. Geotechnical engineering practice
for collapsible soils. Geotechnical and Geological Engineering 2001;19(3):333e
55.
Hu RL, Guan GL, Li XQ, Zhang LZ. Microstructure effect on the subsidence of loess.
Journal of Engineering Geology 1999;7(2):161e7 (in Chinese).
Hu ZQ. Inundation deformation test and numerical analysis of loess structural
model and loess canal. PhD Thesis. Xi’an, China: Xi’an University of Technology;
2000 (in Chinese).
Hu ZQ, Shen ZJ, Xie DY. Research on structural behavior of unsaturated loess. Chi-
nese Journal of Rock Mechanics and Engineering 2000;19(6):775e9 (in
Chinese).
Hu ZQ, Shen ZJ, Xie DY. Deformation properties of structural loess. Chinese Journal
of Rock Mechanics and Engineering 2004;23(24):4142e6 (in Chinese).
Hu ZQ, Shen ZJ, Xie DY. Constitutive model of structural loess. Chinese Journal of
Rock Mechanics and Engineering 2005;24(4):565e9 (in Chinese).
Jennings JE, Knight K. The additional settlement of foundations due to collapse of
sandy soils on wetting. In: Proceedings of the 4th international conference on
soil mechanics and foundation engineering, London; 1957. p. 316e9.
Jennings JE. A revised effective stress law for use in the prediction of the behavior of
unsaturated soils. In: Pore pressure and suction in soils. London: Butterworths;
1961. p. 26e30.
Jennings JE, Burland JB. Limitations to the use of effective stresses in partly satu-
rated soils. Géotechnique 1962;12(2):125e44.
Jiang MJ, Shen ZJ, Zhao KZ. Laboratory determination of collapsibility index of
structural loess. Hydro-Science and Engineering 1999;1:65e71 (in Chinese).
Jiang MJ, Hu HJ, Peng JB, Yang QJ. Pore changes of loess before and after stress path
tests and their links with mechanical behavior. Chinese Journal of Geotechnical
Engineering 2012a;34(8):1369e78 (in Chinese).
Jiang MJ, Hu HJ, Liu F. Summary of collapsible behavior of artificially structural loess
in oedometer and triaxial wetting tests. Canadian Geotechnical Journal
2012b;49(10):1147e57.
Jiang MJ, Zhang FG, Hu HJ, Cui YJ, Peng JB. Structural characterization of natural
loess and remoulded loess under triaxial tests. Engineering Geology 2014a;181:
249e60.
Jiang MJ, Li T, Hu HJ, Thornton C. DEM analysis of one-dimensional compression and
collapse behavior of unsaturated structural loess. Engineering Geology
2014b;60:47e60.
Josa A, Balmaceda A, Gens A, Alonso EE. An elastoplastic model for partially
saturated soils exhibiting a maximum of collapse. In: Proceedings of the 3rd
international conference on computational plasticity, Barcelona; 1992.
p. 815e26.
Jotisankasa A. Collapse behavior of a compacted silty clay. PhD Thesis. London, UK:
University of London; 2005.
Kato S, Kawai K. Deformation characteristics of a compacted clay in collapse under
isotropic and triaxial stress state. Soils and Foundations 2000;40(5):75e90.
Khalili N, Khabbaz MH. A unique relationship of c for the determination of the shear
strength of unsaturated soils. Géotechnique 1998;48(5):681e7.
Knight K. The origin and occurrence of collapsing soils. In: Proceedings of the 3rd
regional conference for Africa on soil mechanics and foundation engineering;
1963. p. 127e30.
Kraev VF. On subsidence of loess soils of the Ukraine. In: Proceedings of the Tokyo
Symposium, Tokyo, Japan; 1969.
Krinsley DH, Smalley IJ. Shape and nature of small sedimentary quartz particles.
Science 1973;180:1277e9.
Lawton EC, Fragaszy RJ, Hardcastle JH. Collapse of compacted clayey sand. Journal of
Geotechnical Engineering 1989;115(9):1252e67.
Lawton EC, Fragaszy RJ, Hetherington MD. Review of wetting-induced collapse in
compacted soil. Journal of Geotechnical Engineering 1992;118(9):1376e94.
Lei XY. Type of the loess microtextures in Xi’an district. Journal of Northwest Uni-
versity 1983;4:56e65 (in Chinese).
Lei XY. Pore types of loess soils in China and its collapsibility. Science in China:
Series B 1987;12:1310e6 (in Chinese).
Li P, Xue ZN, Wang ZJ, Cao JQ, Du DJ. Loess engineering geological characteristics of
eastern Gansu Province, China. Journal of Earth Science and Environment
2004;26(2):59e62 (in Chinese).
Lin ZG, Liang WM. Engineering properties and zoning of loess and loess-like soils in
China. Canadian Geotechnical Journal 1982;19(1):76e91.
Liu DS. Geological environment for loess soils in China. Geological Bulletin of China
1978;1:1e10 (in Chinese).
Liu HS, Ni WK, Yan B, Wang CY. Discussion on relationship between structural
strength and collapsibility of loess. Rock and Soil Mechanics 2008;29(3):722e6
(in Chinese).
Liu ZD. Analysis of the factors affecting collapsibility coefficient of loess soils. En-
gineering Survey 1994;5:6e11 (in Chinese).
Lommler JC, Bandini P. Characterization of collapsible soils. In: Proceedings of IFCEE
2015, San Antonio, Texas; 2015.
Luo AZ, Shao SJ, Chen CL, Fang J. Structural yield and strength properties of collapse
loess under moisture, compression and shearing. Chinese Journal of Under-
ground Space and Engineering 2014;10(6):1243e9 (in Chinese).
Mao LT, Xue R, An LQ. Quantitative analysis of SEM image of microstructure with
MATLAB. Journal of Chinese Electron Microscopy Society 2004;23(5):579e83
(in Chinese).
Matyas EL, Radhakrishna HS. Volume change characteristics of partially saturated
soils. Géotechnique 1968;18(4):432e48.
Newland PL. The behavior of soils in terms of the kinds of effective stress. In:
Proceedings of conference on engineering effects of moisture changes in soils.
College Station, USA: Texas A&M University Press; 1965. p. 27e45.
Noor ST, Hanna A, Mashhour I. Numerical modeling of piles in collapsible soil sub-
jected to inundation. International Journal of Geomechanics 2013;13(5):514e26.
Nouaouria MS, Guenfoud M, Lafifi B. Engineering properties of loess in Algeria.
Engineering Geology 2008;99(1e2):85e90.
Osipov VI, Sokolov VN. Factors and mechanism of loess collapsibility. In: Genesis of
properties of collapsible soils. NATO ASI seriesvol. 468. Dordrecht, the
Netherlands: Kluwer Academic Publishers; 1995. p. 49e63.
Peng JB, Sun P, Li X. Ground fissure: the major geological and environmental
problem in the development of Xi’an City, China. Environmental Science and
Technology 2006;2:469e74.
Pereira JHF, Fredlund DG. Constitutive modeling of a meta-stable-structured com-
pacted soil. In: Proceedings of the Symposium on recent developments in soil
and pavement mechanics; 1997. p. 1e10.
Pereira JHF, Fredlund DG. Volume change behavior of collapsible compacted gneiss
soil. Journal of Geotechnical and Geoenvironmental Engineering 2000;126(10):
907e16.
 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Phien-wej N, Pientong T, Balasubramaniam AS. Collapse and strength characteris-
tics of loess in Thailand. Engineering Geology 1992;32(1e2):59e72.
Popescu ME. A comparison between the behavior of swelling and of collapsing soils.
Engineering Geology 1985;23(2):145e63.
Rampino C, Mancuso C, Vinale F. Experimental behavior and modelling of an
unsaturated compacted soil. Canadian Geotechnical Journal 2000;37(4):
748e63.
Reznik YM. Influence of physical properties on deformation characteristics of
collapsible soils. Engineering Geology 2007;92(1e2):27e37.
Richards BG. Measurement of free energy of soil moisture by the psychrometric
technique using thermistors. In: Moisture equilibria and moisture changes in
soils beneath covered areas. Butterworths; 1966. p. 39e46.
Rogers CDF, Dijkstra TA, Smalley IJ. Hydroconsolidation and subsidence of loess:
studies from China, Russia, North America and Europe. Engineering Geology
1994;37(2):83e113.
Rogers CDF. Types and distribution of collapsible soils. In: Genesis of properties of
collapsible soils. NATO ASI seriesvol. 468. Dordrecht, the Netherlands: Kluwer
Academic Publishers; 1995. p. 1e17.
Romero E, Simms PH. Microstructure investigation in unsaturated soils: a review
with special attention to contribution of mercury intrusion porosimetry and
environmental scanning electron microscopy. Journal of Geotechnical and
Geological Engineering 2008;26(6):705e27.
Ryashchenko TG, Akulova VV, Erbaeva MA. Loessial soils of Priangaria, Trans-
baikalia, Mongolia, and northwestern China. Quaternary International
2008;179(1):90e5.
Saba S, Barnichon JD, Cui YJ, Tang AM, Delage P. Microstructure and anisotropic
swelling behavior of compacted bentonite/sand mixture. Journal of Rock Me-
chanics and Geotechnical Engineering 2014;6(2):126e32.
Sha AM, Chen KS. Relationship between collapsibility and microstructure of com-
pacted loess. Journal of Chang’an University: Natural Science 2006;26(4):1e4
(in Chinese).
Shao SJ, Zhou F, Long JY. Structural properties of loess and its quantitative param-
eter. Chinese Journal of Geotechnical Engineering 2006;26(4):531e6 (in
Chinese).
Shao SJ, Zheng W, Wang ZH, Wang S. Structural index of loess and its testing
method. Rock and Soil Mechanics 2010;31(1):15e20 (in Chinese).
Shao SJ, Wang LQ, Tao H, Wang Q, Wang S. Structural index of loess and its relation
with granularity, density and humidity. Chinese Journal of Geotechnical Engi-
neering 2014;36(8):1387e93 (in Chinese).
Shen ZJ. An elasto-plastic damage model for cemented clays. Chinese Journal of
Geotechnical Engineering 1993;15(3):21e8 (in Chinese).
Shen ZJ, Hu ZQ. Binary medium model for loess. Journal of Hydraulic Engineering
2003;7:1e6 (in Chinese).
Shen ZJ. Breakage mechanics for geological materials: an ideal brittle-elasto-plastic
model. Chinese Journal of Geotechnical Engineering 2003;25(3):253e7 (in
Chinese).
Shen ZJ, Deng G. Binary-medium model for over-consolidated clays. Rock and Soil
Mechanics 2003;24(4):495e9 (in Chinese).
Sheng DC, Fredlund DG, Gens A. A new modelling approach for unsaturated soils
using independent stress variables. Canadian Geotechnical Journal 2008;45(4):
511e34.
Shi B, Li SL, Tolkachev M. Quantitative study on microstructure image of clay soil.
Science in China: Series A 1995;25(6):666e72 (in Chinese).
Simms PH, Yanful EK. Measurement and estimation of pore shrinkage and pore
distribution in a clayey till during soil-water characteristic curve tests. Canadian
Geotechnical Journal 2001;38(4):741e54.
Simms PH, Yanful EK. Predicting soil-water characteristic curves of compacted
plastic soils from measured pore-size distributions. Géotechnique 2002;52(4):
269e78.
Smalley IJ. In-situ theories of loess formation and the significance of the calcium
carbonate content of loess. Earth Science Reviews 1971;7(2):67e85.
Smalley IJ, Jefferson IF, Dijkstra TA, Derbyshire E. Some major events in the devel-
opment of the scientific study of loess. Earth Science Review 2001;54(1e3):5e18.
Song YS, Wang XG. Research into the collapsibility of loess in Longdong District.
Soil Engineering and Foundation 2004;18(4):37e40 (in Chinese).
Sun DA, Matsuoka H, Cui HB, Xu YF. Three-dimensional elasto-plastic model for
unsaturated compacted soils with different initial densities. International
Journal for Numerical and Analytical Methods in Geomechanics 2003;27(12):
1079e98.
Sun DA, Matsuoka H, Xu YF. Collapse behavior of compacted clays in suction-
controlled triaxial tests. Geotechnical Testing Journal 2004;27(4):1e9.
Sun DA, Sheng DC, Xu YF. Collapse behavior of unsaturated compacted soil with
different initial densities. Canadian Geotechnical Journal 2007a;44(6):673e86.
Sun DA, Sheng DC, Cui HB, Sloan SW. A density-dependent elastoplastic hydro-
mechanical model for unsaturated compacted soils. International Journal for
Numerical and Analytical Methods in Geomechanics 2007b;31(11):1257e79.
Sun DA, Sheng DC, Sloan SW. Elastoplastic modelling of hydraulic and stress-strain
behavior of unsaturated soils. Mechanics of Materials 2007c;39(3):212e21.
Sun JZ. Collapsibility of loess and the relationship with moisture. Hydrogeology and
Engineering Geology 1957;11:18e21 (in Chinese).
273
Sun PP, Zhang MS, Zhu LF. Typical case study of loess collapse and discussion on
related problems. Geological Bulletin of China 2013;32(6):847e51 (in Chinese).
Tadepalli R, Fredlund DG. The collapse behavior of a compacted soil during inun-
dation. Canadian Geotechnical Journal 1991;28(4):477e88.
Tadepalli R, Rahardjo H, Fredlund DG. Measurement of matric suction and volume
changes during inundation of collapsible soil. Geotechnical Testing Journal
1992;15(2):115e22.
Tan T. Fundamental properties of loess from northwestern China. Engineering Ge-
ology 1988;25(2e4):103e22.
Terzaghi K. Theoretical soil mechanics. New York, USA: John Wiley and Sons; 1943.
Thu TM, Rahardjo H, Leong EH. Elastoplastic model for unsaturated soil with
incorporation of the soil-water characteristic curve. Canadian Geotechnical
Journal 2007a;44(1):67e77.
Thu TM, Rahardjo H, Leong EH. Soil-water characteristic curve and consolidation
behavior for a compacted silt. Canadian Geotechnical Journal 2007b;44(3):
266e75.
Tian ZL, Ma J, Li YH. Discussion on quantitative parameters of loess structure.
Chinese Journal of Rock Mechanics and Engineering 2011;30(Suppl. 1):3179e84
(in Chinese).
Vanapalli SK, Fredlund DG, Pufahl DE, Clifton AW. Model for the prediction of shear
strength with respect to soil suction. Canadian Geotechnical Journal
1996;33(3):379e92.
Vanapalli SK, Fredlund DG, Pufahl DE. The influence of soil structure and stress
history on the soil-water characteristics of a compacted till. Géotechnique
1999;49(2):143e59.
Wang CL, Lin R, Chen DC, Zhang X, Wang K. Collapsible deformation characteristics
and its changes in microstructure of loess in west of Liaoning. Journal of Jilin
University: Earth Science Edition 2011;41(2):471e7 (in Chinese).
Wheeler SJ, Sivakumar V. An elasto-plastic critical state framework for unsaturated
soil. Géotechnique 1995;45(1):35e53.
Wheeler SJ. Inclusion of specific water volume within an elasto-plastic model for
unsaturated soil. Canadian Geotechnical Journal 1996;33(1):42e57.
Wheeler SJ, Sharma RJ, Buisson MSR. Coupling of hydraulic hysteresis and stress-
strain behavior in unsaturated soils. Géotechnique 2003;53(1):41e54.
Yang YL. Study on collapsible mechanism of loess soils. Science in China: Series B
1988;7:756e66 (in Chinese).
Yuan JX. Analysis of the influences for collapsibility of loess soils in China. West-
China Exploration Engineering 2009;10:31e4 (in Chinese).
Yuan ZX, Wang LM, Wang J. Discussion on water collapsibility of loess considering
unsaturated soil and structure characteristics of loess. Northwestern Seismo-
logical Journal 2007;29(1):12e7 (in Chinese).
Zhao JB, Chen Y. Study on pores and collapsibility of loess. Journal of Engineering
Geology 1994;2(2):77e83 (in Chinese).
Zhao JB, Yue YL, Chen Y. Collapsibility of loess and its origin. Journal of Geo-
mechanics 1997;3(4):62e8 (in Chinese).
Zhang BP, Yuan HZ, Wang L. The quantitative analysis of effects of soil moisture
upon the loess structure strength. Acta Universitatis Agriculturae Boreali e
Occidentalis 1994;22(1):54e60 (in Chinese).
Zhang MH. Experimental study on wetting and drying deformation characteristics
of collapsible loess soils. MS Thesis. Xi’an, China: Chang’an University; 2002 (in
Chinese).
Zhang XF, Cai ZX, Hu WX, Li L. Using Adobe photoshop to quantify rock Textures.
Acta Sedmentologica Sinica 2009;27(4):667e73 (in Chinese).
Zhang SM, Zheng JG. The deformation characteristics of collapsible loess during
moistening process. Chinese Journal of Geotechnical Engineering 1990;12(4):
21e31 (in Chinese).
Zheng JG, Zhang SM. Initial collapse pressure and initial collapse water content of
loess soils. Engineering Investigation 1989;2:6e10 (in Chinese).
Zorlu K, Kasapoglu KE. Determination of geomechanical properties and collapse
potential of a caliche by in situ and laboratory tests. Environmental Geology
2009;56:1449e59.
Ping Li received her M.Sc. degree in Geological Engineer-
ing in 2013 from Chang’an University, Xi’an, China. In the
same year, Li started her Ph.D. program in the same uni-
versity. In September 2014, she was awarded a scholarship
under the Chinese State Scholarship Fund for supporting
her to study in Civil Engineering in University of Ottawa,
Canada, for 2 years based on her excellent academic per-
formance. Ping Li’s Ph.D. program focuses on developing a
simple model for estimating the collapse behavior of loess
soils upon wetting based on the concepts of unsaturated
soil mechanics. Through her thesis, she would like to make
a significant contribution to the knowledge of collapsible
loess soils and provide methods that can be applied to
engineering practice. Ping Li has authored or co-authored
6 journal papers and 3 international conference papers to date.
274 P. Li et al. / Journal of Rock Mechanics and Geotechnical Engineering 8 (2016) 256e274
Sai Vanapalli is presently a professor of the Department of
Civil Engineering, University of Ottawa, Ottawa, Canada.
He received his Ph.D. in 1994 from University of Sas-
katchewan, Saskatoon, Canada. Since then, his research
focus has been directed towards the development of
simple techniques to interpret and predict the engineering
behavior of unsaturated soils. These studies are used in the
design of geotechnical structures such as pavements,
shallow and deep foundations, slopes, soil nails and
retaining walls. More recently, Sai Vanapalli has been
developing simple techniques working with his graduate
students to estimate and model (i) the stiffness (i.e. resil-
ient modulus MR, maximum shear modulus Gmax, and
modulus of elasticity E) of unsaturated soils, (ii) the multi-
dimensional heave in expansive soils and its influence on geotechnical structures
including slope stability, (iii) extending the mechanics of unsaturated soils in the
design of pavements, shallow and deep foundations, and (iv) the influence of dynamic
loads and earthquakes on the unsaturated soil behavior. Dr. Sai Vanapalli to-date has
authored or co-authored 220 research articles that are disseminated in several journals
and conferences, and has delivered 13 keynote lectures in international conferences.
He received Stermac Award for service contributions from the Canadian Geotechnical
Society in 2010 and Glinski Award in 2015 for his outstanding research contributions
from the University of Ottawa.
Tonglu Li obtained M.Eng. Degree from Changchun Uni-
versity of Earth Science, Changchun, China, and Ph.D. de-
gree from Xi’an College of Geology, Xi’an, China. He is
presently a professor of Geological Engineering, Chang’an
University, Xi’an, China, where he used to be the Head of
the Department and the Vice-dean of the School of
Geological Engineering and Geomatics. Li has been
involved in geological research, consultation and educa-
tion for more than 25 years, focusing on geo-
environment and geo-disaster in Chinese loess regions.
He is authored and co-authored more than 80 academic
articles and sits on the Editorial Board of Geoenvironmen-
tal Disasters. He is a member of IAEG and the Vice-
President of International Consortium on Geo-disaster
Reduction. He has consulted for many key projects of urban buildings, highways, hy-
dropower stations, oil-gas transfer lines and nuclear electric plants in China. In recent
years, he has been supported by two National Natural Science Funds and one National
Basic Research Program to relate the permeability of loess soils to geo-disasters in loess
regions based on the concepts of unsaturated soil mechanics. The research studies that
he and his collaborators have undertaken are widely used for both interpreting and
estimating the geo-disasters associated with loess soils.