Engineering Geology 253 (2019) 137–148

Contents lists available at ScienceDirect

Engineering Geology
journal homepage: www.elsevier.com/locate/enggeo

Tensile strength of clayey soil and the strain analysis based on image T
processing techniques
Hao-Da Li, Chao-Sheng Tang , Qing Cheng, Sheng-Jie Li, Xue-Peng Gong, Bin Shi
⁎

School of Earth Sciences and Engineering, Nanjing University, 163 Xianlin Avenue, Nanjing 210023, China

ARTICLE INFO ABSTRACT

Keywords: Tensile strength is one of the most important mechanical parameters controlling the development of cracks in
Tensile strength soil. However, it is frequently neglected in conventional geotechnical practice, because its magnitude is small
Tensile stress-strain and difficult to measure relative to other soil strength parameters. In this paper, a newly designed direct tensile
Unsaturated compacted soil test apparatus was employed to measure the tensile strength of an unsaturated clayey soil. A digital image
Microstructure
acquisition and analysis system was developed for tensile strain analysis with the help of Particle Image
Particle image velocimetry
Digital image correlation
Velocimetry (PIV) and Digital Image Correlation (DIC) techniques. Six groups of samples were compacted at a
dry density of 1.7 Mg/m3 and different water contents (6.5%, 8.5%, 10.5%, 12.5%, 16.5% and 20.5%). Test
results show that the tensile strength characteristic curve (tensile strength versus water content) of the com-
pacted unsaturated soil exhibits mono-peak feature. When water content is relatively low, the tensile strength
increases with increasing water content and reaches the maximum value at a critical water content of about
9.3%. Then, it declines with further increase in water content. The evolution of tensile strength with water
content depends on both suction and microstructure. Based on plotted tensile load-displacement curves, the
tensile failure process can be divided into three typical stages which are: stress increasing stage (I), failure
developing stage (II) and post-failure stage (III). It is found that the overall tensile failure process presents
different patterns controlled by water content. Generally, the failure developing stage (II) lasts longer and the
failure ductility is more pronounced when the sample is compacted at higher water content. Using PIV and DIC
techniques, the development of displacement direction and strain concentration during tension can be well
captured for appreciation of the soil failure mechanism. Based on the strain concentration information, the
tensile fracture location and direction can be pre-determined for soil samples.

1. Introduction 2017a). It is well investigated, and it has been practically proven that
the existence of cracks can significantly change the mechanical and
Tensile strength is one of the most important soil strength indices hydraulic properties of soil (Konrad and Ayad, 1997; Morris et al.,
used in practice for description of soil resistance to loading. Compared 1992; Venkataramana et al., 2009). For instance, tensile cracks damage
to other strength indices such as compressive strength and shear the integrity of a soil mass and enables easier disaggregation of soil into
strength, tensile strength is much smaller (Zhu et al., 2008; Li, 2018; particles (Tang et al., 2010). The transfer of stress among soil particles
Yin and Vanapalli, 2018). For this reason, when soil is used as a con- may be compromised. As a result, the strength of a soil with cracks is
struction material in conventional engineering projects, its tensile often much lower than that of an intact soil (Morris et al., 1992). Also,
strength is frequently neglected. This neglect of soil tensile strength is cracks can create preferential fluid flow paths inside soil. Usually, this
somewhat conservative in earth structure design when zero suction in results in increase in soil hydraulic conductivity to several orders of
soil is assumed (Tang and Graham, 2000). magnitude (Albrecht and Benson, 2001; Thusyanthan et al., 2007). The
However, as a typical form of tensile failure, desiccation cracking is presence of cracks in soil can also increase water evaporation rate and
a common phenomenon in drying soils. It appears in various earth intensify soil erosion with the possibility of geological disasters and
structures such as landfills, slopes, dams and embankments (Nahlawi environmental problems (Tang et al., 2011).
et al., 2004; Divya et al., 2014; Tang et al., 2015; Tollenaar et al., In recent years, due to global climate change, extreme drought is

Corresponding author.
⁎

E-mail addresses: [email protected] (H.-D. Li), [email protected] (C.-S. Tang), [email protected] (Q. Cheng),
[email protected] (S.-J. Li), [email protected] (X.-P. Gong), [email protected] (B. Shi).

https://doi.org/10.1016/j.enggeo.2019.03.017
Received 3 April 2018; Received in revised form 21 February 2019; Accepted 24 March 2019
Available online 25 March 2019
0013-7952/ © 2019 Elsevier B.V. All rights reserved.
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

occurring with increasing frequency and intensity (Wang et al., 2016). accurately obtained. Common image processing techniques include PIV
More and more researchers have paid attention to soil desiccation (Particle Image Velocimetry) and DIC (Digital Image Correlation)
cracking and its consequence in a variety of geological and geo-en- techniques. The PIV technique was developed in fluid mechanics at the
vironmental engineering fields (Tang et al., 2008; Peron et al., 2009; beginning of the 1980s as a kind of flow measurement method
Costa et al., 2012; Tollenaar et al., 2017a; Ledesma et al., 2018). It is (Hesselink, 1988; Adrian, 1999). Its principle is to observe the con-
assumed that desiccation cracking or tensile failure occurs when the tinuous movement of tracer particles in the image, determine the de-
applied tensile stress or strain exceeds soil tensile strength or failure formation information by comparing particles, and find the same one in
tensile strain (Corte and Higashi, 1960; Nahlawi et al., 2004; Tang different images. DIC technique was independently proposed by
et al., 2011). In order to figure out the cracking mechanism of soil and Yamaguchi (1981), Ranson and Peters (1982) initially. Its basic prin-
improve the stability and safety of earth structures, it is of great sig- ciple is to divide the digital image into small regions, then compare the
nificance to have a comprehensive understanding of the tensile regions from different images by identifying the similarity of gray
strength-strain characteristics of the particular type of soil targeted for characteristic value. Based on the deformation and movement in-
use. formation of the pixels, the strain can be calculated. Due to the ad-
For soil tensile strength, there are generally two categories of vantages of non-contact, and high measurement accuracy and resolu-
measurement methods: namely indirect and direct methods. The in- tion, PIV and DIC techniques have been applied in many disciplines,
direct methods are based on correlations between tensile strength and including geotechnical engineering (White et al., 2003; Higo et al.,
various parameters. With these methods, tension is usually replaced by 2013; Divya et al., 2014; Stanier et al., 2015; Wang et al., 2017). For
compressive force or bending force, applied to soil sample. Then, tensile instance, Tollenaar et al. (2017b) used the PIV technique to assess the
strength can be calculated using some theoretical equations based on generated strain on sample surfaces during direct tensile test. Wang
various hypotheses. Commonly, there are three indirect tensile test et al. (2018) applied the DIC technique to quantify the surface strain
methods that are widely used: the split tensile test (Brazilian test) field and its temporal and spatial evolution characteristics as the soil is
(Frydman, 1964; Ismaiel, 2006; Sobhan and Mashnad, 2002; Festugato subjected to continuous drying. They found that the sites of desiccation
et al., 2018), the beam bending test (Thusyanthan et al., 2007; Narvaez cracks could be reliably predicted based on surface strain concentration
et al., 2015) and the axial fracturing test (double-punch test) (Fang and information.
Chen, 1972; Kim et al., 2007; Liang et al., 2014). In principle, these A large number of experimental and theoretical investigations have
methods are suitable mostly for brittle and elastic materials such as been conducted on soil tensile strength during the past decades (Kim
cement, rock and chemical stabilized soil. As for soft soil or other and Hwang, 2003; Yin and Vanapalli, 2018). Particularly, the link be-
ductile materials, in order to improve the reliability of test results, some tween soil tensile strength and some key textural and load imposition
additional correction coefficients are necessary (Frydman, 1964). mechanisms has been well established. For a given soil, it is found that
The direct test methods are designed to apply tensile force directly the determination of tensile strength not only depends on testing
to both ends of the sample tested (Stirling et al., 2015; Varsei et al., method (indirect or direct way), but also on the adopted theoretical
2016; Trabelsi et al., 2018). The uniaxial tensile test is the most com- assumptions (Barzegar et al., 1995). Lu et al. (2007) postulated that the
monly used method as the tensile stress and strength can be directly tensile strength of unsaturated sands generally originates from the in-
obtained. Meanwhile, both the tensile load and displacement can be terparticle capillary force, and its contribution to the tensile strength
directly controlled and recorded by the test device. However, it has was quite different in different saturation regimes. Furthermore, a
been found that direct tensile test is difficult to perform due to problems theory based on generalized effective stress of unsaturated soil was
in sample preparation and fixing. Initially, cylindrical samples were proposed. This concept can accurately describe tensile strength of moist
used and both ends of each sample were glued or clamped directly to sand (Lu et al., 2009). Trabelsi et al. (2012) introduced a formula for
the loading system (Tang and Graham, 2000). However, these tests characterization of the tensile strength of unsaturated soils that treated
were often unsuccessful as the integrity of the glued or clamped con- its influence of porosity separately. Lakshmikantha et al. (2012) found
nection between the loading system and sample could not be guaran- that soil tensile strength is a function of water content and degree of
teed. Tensile stress concentration or tensile failure usually occurred saturation. Tang et al. (2015) pointed out that the tensile strength of
near the connection (Mesbah et al., 2004). Tapered samples each with a compacted clayey soil not only depends on the degree of saturation, but
reduced central section area, were used to solve this problem (Nahlawi also highly on microstructure. They proposed a model with two sepa-
et al., 2004; Rodríguez et al., 2007). Still, it is very difficult to compress rate expressions to describe the tensile strength characteristic curves of
materials into a tapered mould uniformly. To overcome these short- compacted clayey soil. Tollenaar et al. (2017b) evaluated the combined
comings, standard compacted rectangular soil blocks were used by effects of its pull rate and water content on the tensile response of a
Mesbah et al. (2004). The blocks were sawed at mid-height to create a clay. They found that the effect of pull rate on the tensile strength was
weakened cross section where the tensile failure was expected to occur. insignificant as compared to the effect of the water content. In the study
It should be noted that after sawing tensile stress is no longer uniform of Trabelsi et al. (2018), the evolution of tensile strength during drying
throughout the sample. Significant stress concentration is likely to of remoulded and compacted clays was well linked to soil fabric and
propagate and trigger early cracking. Following the techniques in- water retention. A model for tensile strength was therefore, proposed
troduced in previous literature, Tang et al. (2015, 2016) developed a based on a series of experimental test data on soil suction and pore size
direct tensile test apparatus equipped with a specially designed wedge- distribution. Yin and Vanapalli (2018) proposed a semi-empirical
shaped compaction mould, piston and tensile mould, which can sig- model that is capable of predicting the variation of tensile strength of
nificantly simplify the test procedures and improve test repeatability. unsaturated cohesionless soils arising from matric suction and surface
In regard to tensile strain, most researchers use dial indicator or tension. These previous investigations highlight the fact that the tensile
LVDT to measure the deformation of samples. Then, the strain is cal- strength of soil depends significantly on the test method while it is also
culated theoretically. This method is only suitable for homogeneous soil strongly linked to soil water content, degree of saturation, void ratio/
under even load. As for eccentric load, torsion and other special loading porosity, suction, surface tension, microstructure and so on. Notwith-
conditions, strain concentration and premature failure cannot be de- standing the many studies that have focused on the tensile strength,
tected by conventional ways. In recent decades, the development of only a few investigations have been carried out to estimate the overall
image processing techniques have significantly promoted the revolution tensile stress-displacement/strain characteristics and tensile failure
of strain monitoring and measurement. By calculating the value of process due to the difficulties in accurate measurement of tensile dis-
geometrical parameters recorded in digital images, the displacement placement/strain in the “region of interest” (ROI, a selected region from
field and the strain field information on the surface of sample can be the sample for particular consideration) by conventional techniques.

138
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

Fig. 1. The designed direct tensile test apparatus.

Moreover, the effect of water content on soil tensile failure process and 72 × 72 dpi) was vertically mounted above the sample and manually
the related mechanism have remained unclear. focused on the upper surface of the sample. The necked section was set
To simplify and make soil tensile strength testing more efficient, a as the region of interest (ROI), and this section was adjusted to the
new displacement-controlled direct tensile test apparatus with image center of the captured picture. These procedures minimize possible
capture system was designed. It was used to investigate the tensile re- distortions of image. Otherwise, corrections for camera lens would be
sponse of soil in this study. Both PIV and DIC techniques were employed made. Pictures were taken at certain intervals to record the tensile
to analyze tensile displacement/strain field in ROI. In order to examine loading and deformation process. The pictures were then used for PIV
the effects of water content on tensile strength and the overall tensile and DIC analysis.
failure process, a series of direct tensile tests were conducted on sam-
ples compacted at dry density of 1.7 Mg/m3 but with different water 3. Materials and methods
contents (i.e. 6.5%, 8.5%, 10.5%, 12.5%, 16.5% and 20.5%) by refer-
ring to Tang et al. (2015). The obtained tensile test results were ana- 3.1. Materials
lyzed and discussed with the combination of PIV and DIC images.
The soil used was collected in Nanjing, southeastern China. It is
2. Test apparatus widely found in the middle and lower reaches of the Yangtze River
region and commonly used in different construction applications, in-
The newly designed direct tensile test apparatus with image capture cluding liners of landfills and embankments. Before the test, the soil
system is shown as Fig. 1. This apparatus consists of three essential was air-dried (w = 4.3%), crushed and sieved to pass 2 mm. The soil
systems which are its tensile testing system, its data collection system used in this investigation is the same as that used by Tang et al. (2015).
and the image capture system. Its physical characteristics are given in Table 1.
For the tensile testing system, an ‘8’ shaped split tensile mould with
necked center section was specially designed and built. It comprises 3.2. Sample preparation and installation
four pieces (Fig. 2(a), (b) and (c)). The sample can be directly placed in
this mould. The necked center section was designed to reduce its The crushed air-dried soil powder was mixed with the pre-calcu-
sample cross-sectional area so as to pre-determine the location and lated amount of distilled water to the desired water content. Six groups
direction of tensile failure. During the test, the mould that holds the of mixtures were prepared at different water contents (i.e. 6.5%, 8.5%,
sample was placed on a platform which comprises two parts (Fig. 1(a)). 10.5%, 12.5%, 16.5% and 20.5%). These mixtures were stored and
One part is movable, and the other one is fixed. Half of the mould is sealed in plastic bags for 2 days to allow water to distribute uniformly
fixed on the movable platform and the other half is fixed on the im- within the samples.
mobile platform (Fig. 1(b)). The imposed tension is applied by a motor After that, the required quantity of soil mixture was put into the
with a given displacement rate (ranging from 0 to 10 mm/min). The specially designed ‘8’ shaped compaction mould (Fig. 2(d)) and stati-
tension data is measured by a load cell (with a capacity of 160 N, re- cally compacted to a dry density of 1.7 Mg/m3 by an ‘8’ shaped piston
solution of 0.001 N and accuracy of 0.03%FS) positioned between the (Fig. 2(e) and (f)). The sample was compacted in three layers. The final
moving platform and the motor. To reduce friction, a linear sliding length, width and height of the sample were 200, 100 and 50 mm, re-
roller is placed between the movable platform and base plate. The spectively. The width of the center necked section was 60 mm. In order
movable platform is pulled forward in the horizontal direction until the to provide tracer particles and artificial texture for the subsequent PIV
soil sample fails. All these parts are made with stiff steel and combined and DIC analysis, white paint was sprayed on the upper surface of the
tightly. Therefore, it is reasonably assumed that the sample deformation sample to obtain a random speckle pattern for image correlation ana-
performs synergistically with the loading system. The tensile displace- lyses, as shown in Fig. 3. The size of speckles is on the order of mi-
ment is measured by a linear variable differential transducer (LVDT) crometers. They are sparsely distributed on the sample's surface. The
(with a capacity of 100 mm resolution of 0.001 mm and accuracy of influence of speckles on the mechanical behavior of the sample can be
0.1%FS) positioned on the side of the mould. Both the load cell and neglected (Wang et al., 2018).
LVDT are connected to the data collection system. The sample was extruded from the mould and placed into the ‘8’
A camera with high sensitivity and image quality (the resolution is shaped split tensile mould (Fig. 2) which is fixed onto the platform

139
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

Fig. 2. Four pieces of the split “8” shaped tensile mould (a, b and c), “8” shaped compaction mould (d) and compaction piston (e and f).

Table 1 corresponds to the maximum tensile force that counterbalances the

Physical properties of soil. whole interparticle contact force along the failure plane. It is defined as
Soil properties Value
the tensile strength, σt of the sample. For each group of tests, three
parallel samples were prepared to check the repeatability of the pro-
Specific gravity 2.73 cedures.
Consistency limit
Liquid limit (%) 37
Plastic limit (%) 20 3.4. Image processing
Plasticity index (%) 17
USCS classification CL
During the test, the upper surface deformation process of the sample
Compaction study
Optimum moisture content (%) 16.5 was monitored by a camera (Fig. 1). Pictures were taken at 5 s intervals.
Maximum dry density (Mg/m3) 1.7 After the test, all the pictures were converted to gray scale images, and
Grain size analysis the ROI with 60 × 60 mm around the neck of the sample (marked in
Sand (%) 2 Fig. 3) was selected and imported into the computer program ImageJ
Silt (%) 76
Clay (%) 22
and CorreliQ4 for PIV and DIC analysis (Mathieu et al., 2011; Schneider
et al., 2012). By comparing the subsequent pictures with the reference
picture taken before tension was applied, deformation information in
(Fig. 1). The necked section was exactly aligned to the gap between the ROI at different tensile stages could be obtained. It should be noted that
movable and the unmovable platform. after the calculation and visualization steps, the PIV and DIC results
were output as images with strain scale. In order to better show the
tensile failure process, the scale was adjusted to show as many details as
3.3. Test procedures
possible with consideration of tensile strain levels. Finally, different
colors representing different strain levels were illustrated in DIC
The direct tensile test was performed at a constant displacement
images. The length and direction of vectors that represented the de-
rate of 0.2 mm/min in the horizontal direction. The movable platform
formation were illustrated in the PIV images.
together with half of the sample and split tensile mould, moved towards
the motor and applied tensile force to the sample. During the test,
tensile force and displacement were measured by the load cell and 3.5. Suction measurement
LVDT, respectively. The measured tensile force, T consists of the force
acting on the tensile failure plane and the force overcoming the friction Following the same procedures mentioned above, another series of
between the moveable platform and base plate. The calibration result samples were prepared for suction measurements. Suction at each
indicated that the system friction was quite small and could be ne- compaction water content (i.e. 6.5%, 8.5%, 10.5%, 12.5%, 16.5% and
glected. Therefore, the tensile stress, σ was calculated by dividing the 20.5%) of the sample with a dry density of 1.7 Mg/m3 was measured
tensile force, T by the cross-sectional area A (60 × 50 mm) at the neck using a WP4 dew point hygrometer as introduced by Wang et al.
of the sample: (2015). The soil water retention curve (SWRC) was then determined.
The WP4 device utilizes the chilled mirror dew point method and
= T /A (1)
measures suction by determining the relative humidity of the air above
In the process of loading, the tensile stress increased continually at a soil sample. More details about the principle of WP4 and suction
the beginning until a peak value was attained. This peak value measurement procedures can be found in Mantri and Bulut (2014).

140
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

Fig. 4. Tensile curves of samples with water content of 16.5%.

force of soil particles (Lambe and Whitman, 1979). There are four kinds
of physical mechanisms that contribute to soil cohesion: (1) van der
Waals attraction at or near particle contacts; (2) electrical double-layer
repulsion between platy clay particles and attraction near face-to-edge
contacts; (3) cementation due to solute precipitation; and (4) capillary
attraction due to the existence of water bridges or bodies between
particles (Lu et al., 2007; Tang et al., 2015). With the existing research
methods, it is difficult to study the first three types of mechanisms
mentioned above. Determination methods of their magnitudes are not
well established (Lu et al., 2007). In recent decades, most of the re-
searchers have focused on quantification of its relationship between
capillary force and tensile strength of unsaturated soil by examining
water retention mechanisms like suction or mechanical interactions
between two particles (Helmar, 1975; Orr et al., 1975; Lian et al., 1993;
Lu and Likos, 2006; Lu et al., 2007, Lu et al., 2009; Romero et al., 2011;
Yin et al., 2018). Considering that the soil tested in this study is
dominated by silt (76%, Table 1), the role of van der Waals attraction
and electrical double-layer repulsion between clay particles as the de-
terminant of soil tensile characteristics is less significant as compared
with overall water retention behavior. Therefore, in the following sec-
tions, the obtained tensile results will be analyzed and discussed mainly
from the perspective of soil suction in tandem with soil microstructure.

Fig. 3. The prepared “8” shaped sample and the sprayed white spots on the
upper surface. 4.2. Effects of water content on tensile strength

For completely dry or saturated soils, tensile strength is often con-

WP4 measures total suction.
sidered as a material constant or part of shear strength. For unsaturated
soils, tensile strength is a function of suction controlled by water con-
4. Results and discussion tent (Lu et al., 2007; Villar et al., 2009; Stirling et al., 2015; Tang et al.,
2015; Trabelsi et al., 2018). Generally, as soil is subjected to drying,
4.1. Overall tensile characteristics decrease in water content or degree of saturation gives rise to an in-
crease in suction as well as cohesion between soil particles, which at the
In this experiment configuration, the tensile stress is supposed to same time, produces an increase in the tensile strength (Trabelsi et al.,
distribute uniformly on the failure plane. Therefore, the tensile curves 2012). In this research, according to the measured tensile strength, στ of
(tensile stress-displacement) can be plotted based on the measured data samples at different water contents w, a fitted tensile strength char-
and Eq. (1). Fig. 4 shows the tensile deformation curves of the samples acteristic curve (TSCC, στ versus w) was shown in Fig. 5. In addition, the
with water content of 16.5%. It can be seen that the tensile deformation variation of suction with water content namely the SWRC was also
patterns of the parallel samples are similar, indicating good consistency presented in Fig. 5. It can be seen that TSCC is a distinct unimodal curve
of sample preparation and test procedure. In the initial stage of loading, which is consistent with the results obtained by Tang et al. (2015). The
the stress increases monotonically with increase in displacement until result shows that στ is significantly influenced by w. A critical water
the peak value is reached. Then, the tensile stress begins to decrease, content, wc (about 9.3%) corresponding to the maximum tensile
and tensile failure occurs at the neck position. The maximum tensile strength στmax (63 kPa), can be determined from the curve. When w is
stress on the curve is defined as the tensile strength of the sample. The less than wc (dry side), στ increases quickly with increase of w. How-
average tensile strength is therefore, calculated for to the parallel ever, when w exceeds wc (wet side), στ begins to decrease gradually
samples. with increasing w. As shown in Fig. 5, the increase rate on the dry side
Generally, the tensile strength of soil is derived from the cohesive is apparently higher than the decrease rate on the wet side. This

141
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

increase the initial water content. Generally, the sprayed water is first
wrapped in the surface layer of the aggregate and will quickly form
liquid-bridges with surrounding aggregates at contact points. Due to the
presence of liquid water, soil particle hydration may occur and sig-
nificantly improve local contact conditions among aggregates. Thus, a
force chain is established between aggregates and contributes to macro-
mechanical properties of the soil such as tensile and shear strength. The
excess water will gradually migrate to the inner aggregate due to suc-
tion gradient. In the subsequent compaction process, the contact zones
of some aggregates become denser due to high pressure, and the local
pore water may be extruded from aggregates in ways that further
strengthen the particle liquid-bridge effect. Increase in soil water con-
tent may create more liquid-bridges and improve the resistance to
tensile failure. This can explain observations on the dry side.
As the water content reaches wc, the liquid-bridges appear at all the
contact points, the maximum tensile strength is obtained (Fig. 5).
When the soil is compacted at relatively high water content, it de-
velops a dispersed structure with little or no obvious inter-aggregate
pores. Further increase in water content gives rise to the loss of liquid-
Fig. 5. Tensile strength characteristic curve (TSCC), soil water retention curve
bridges and the reduction of capillary bonding force among soil parti-
and sketch drawing of microstructure features (modified from Tang et al., 2015)
cles or aggregates. Meanwhile, soil suction decreases with increasing
at different water contents.
degree of saturation. These are the main factors responsible for the
observed decrease in tensile strength on the wet side (Fig. 5).
contrast indicates that tensile strength of soil at high degree of sa- When water content is close to saturated water content, the ag-
turation is not as sensitive to the variation of water content as at re- gregates are dispersed into small clay particles and the inter-aggregate
latively low degree of saturation. It should be clarified that the wc de- pores disappear completely (Delage et al., 2006). The soil shows a re-
termined from TSCC is much lower than the optimum water content wop latively homogenous structure. In this state, the suction and capillary
(16.5%, Table 1) determined from the compaction curve. Due to the bonding forces are very low and cannot contribute much to tensile
difference in the definition of these two parameters, the physical strength. The tensile strength is mainly dominated by cohesion between
meaning of the herein, termed “wet side” or “dry side” related to wc is clay particles like van der Waal attractions, electrical double-layer
completely different from that related to wop. Apparently, the range of forces, etc. Therefore, the tensile strength reduces to a very low level. In
“wet side” moisture content determined by wc is much larger than that order to quantitatively describe the role of microstructure and suction
determined by wop. on soil tensile strength, Trabelsi et al. (2018) developed a simplified
In past few decades, the effect of water content on soil tensile expression by introducing two microstructure parameters linked to
strength has been investigated by several researchers (Towner, 1987; micro- and macroscopic degree of saturation respectively.
Morris et al., 1992; Tang and Graham, 2000; Ávila, 2004; Nahlawi In addition to water content, soil tensile strength also depends sig-
et al., 2004; Wang et al., 2007; Trabelsi et al., 2012; Tollenaar et al., nificantly on porosity and dry density. It is found that tensile strength
2017b). Their results show that tensile strength decreases mono- generally increases with decreasing void ratio (Tang et al., 2015), but
tonically with increase in water content or decrease in suction. This Rodríguez (2006) indicated that the influence of void ratio on tensile
observation applies only to the tendency of the wet side shown in Fig. 5 strength becomes insignificant when the soil is almost fully saturated.
which is completely opposite to the dry side. For example, the de- Recently, Trabelsi et al. (2012) proposed a model which linked the
termined tensile strength increases with a decrease in suction on the dry tensile strength to soil porosity as an independent variable. The model
side. It therefore, cannot be explained solely by unsaturated soil me- has been used to simulate the desiccation cracking behavior of soil.
chanics or suction theory. Evolution of microstructure with changes in Trabelsi et al. (2018) investigated the tensile strength of remoulded and
water content should be taken into account (Tang et al., 2015; Trabelsi compacted clay during drying, and found that the increase in dry
et al., 2018). density on drying, dominated the tensile response in all the cases.
Recently, some advanced test methods such as scanning electron
microscopy (SEM) and mercury intrusion porosimetry (MIP) have been 4.3. Effects of water content on stress and strain characteristics
used to study soil microstructure (Romero and Simms, 2008). It is found
that soil develops typical aggregate-structure when it is compacted at Fig. 6(a) shows the typical tensile deformation curves of samples
low water content. The most common simplification of the soil micro- with different water contents. It can be seen that in addition to tensile
structure considers two categories of pores in the soil: inter-aggregate strength, there is a clear difference in the overall tensile behavior of the
pores and intra-aggregate pores (Romero et al., 1999; Alonso et al., samples with different water contents. Generally, with increase in water
2010; Romero, 2013). In this configuration, tensile strength of soil is content, a decrease in tensile stiffness can be observed. This is reflected
mainly determined by the connections in the form of liquid-bridges by the decreased slope of the tensile deformation curve in the initial
between aggregates. It must be pointed out that most of the water is stage when the sample was compacted at relatively higher water con-
stored inside aggregates at low water content state (Alonso et al., 1990; tent. After the peak tensile stress is reached, the post-peak stress de-
Gens and Alonso, 1992). In this case, it is difficult to form enough li- creases rapidly to zero if w ≤ 10.5%, while the decrease rate is much
quid-bridges between aggregates in order to levels that would con- lower at the beginning and then followed by a quick drop of tensile
tribute to the tensile strength of the soil. For instance, for the crushed stress if w > 10.5%. These observations indicate that the samples with
soil powder in the air-dried state, it is very hard to form precast samples higher water content present higher ductility at failure. Based on these
by compaction because the connection between aggregates cannot be observations, the tensile deformation curve can be divided into three
effectively established. Its mechanical behavior is generally equivalent stages to better describe the tensile process: stress increasing stage (I),
to that of granular soil such as sand. In this study, the air-dried water failure developing stage (II) and post-failure stage (III) as illustrated in
content of the tested soil is 4.3%. During the sample preparation pro- Fig. 6(b). In stage I, tensile stress, σ increases almost linearly with the
cess, additional water was sprayed into the crushed soil powder to increase in displacement, s until the peak is reached (i.e., tensile

142
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

Fig. 6. (a) Typical tensile curves of samples with different water contents, and (b) typical three stages of the tensile curve (w = 20.5%).

(a) 30
(2) (3)
S1-1
25

20
Tensile stress (σ ), kPa

15
Stage I Stage III
Stage II
10

5
(4)
(1)
0

0.0 0.2 0.4 0.6 0.8

Tensile displacement (s), mm
(b)

Fig. 7. (a) Typical tensile curve, and (b) the images of ROI, DIC and PIV related to different tensile stages for the sample with water content of 6.5%.

strength στ). In stage II, σ begins to decrease slightly. It means that local tensile deformation curve. Finally, in stage III, σ decreases sharply to
tensile failure would occur and develop gradually in the sample, but the zero in a short period of time, indicating that overall failure occurs. Past
failure surface does not penetrate the whole neck section and the research led up the assumption that the failure process develops at stage
sample is still able to bear tensile force. For samples at relatively low I after yielding (i.e., the soil stiffness starts to decrease). However, the
water content, this stage is very short and difficult to identify from the results in Fig. 6(a) show that in stage I, the stress increases almost

143
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

(a) (2)
50
S3-1

40

Tensile stress (σ ), kPa

Stage I Stage III
30 Stage II

20

10 (3)
(1)

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
Tensile displacement (s), mm

(b)

Fig. 8. (a) Typical tensile curve, and (b) the images of ROI, DIC and PIV related to different tensile stages for the sample with water content of 10.5%.

linearly with displacement. It is difficult to accurately determine the was observed on the right side of the central section (DIC image), in-
position of the yield point. For this reason, the starting point of failure dicating the onset of fracturing. Point (3) was also selected from stage
phenomenon in this study is defined as being just after the peak stress. II, just a little behind the point (2). As expected, the tension zone in-
This can also be assessed using the subsequent image processing results. creased in size with a strain magnitude of up to 0.032, indicating the
It is clear that the tensile behavior of a soil depends on water con- propagation of fracture during loading. The fracture could not be
tent. For further analyses of the effects of water content on the overall clearly distinguished yet on the sample surface. From the PIV image, it
tensile failure process, the ROI images during tensile deformation tests can be seen that the tensile strain concentration in this stage is mainly
were compared with the initial one, taken just before the tensile load contributed by the movement of soil on the upper right section of the
was applied. With the application of image processing techniques (PIV fracture. Point (4) selected from stage III, refers to a tensile displace-
and DIC), the evolution of axial deformation and strain distribution ment of about 0.68 where overall tensile failure has occurred. At this
patterns was characterized. The typical results are presented in Figs. 8, point, the fracture was already visible at the location of the central
9 and 10, and they are related to the samples compacted at relatively section where the corresponding tensile strain could be higher than
low water content (i.e. S1–1, w = 6.5%), around wc (i.e. S3–1, 0.13. Based on these observations, it is further confirmed that the
w = 10.5%), and relatively high water content (i.e. S6–1, w = 20.5%) tensile stress reaches its maximum value at the moment just before the
respectively. Some points in different tensile stages were selected from onset of fracturing. The subsequent decrease in tensile stress in stage II
the tensile deformation curve to evaluate the strain and displacement is attributed to the generation and propagation of fractures. Stage II
that took place in the ROI. These points were then related to their lasted for a very short time, implying that the tension fracture devel-
corresponding DIC and PIV images. Note that the positive strain values oped very rapidly and the sample presented noticeable failure brittle-
in the colorbar of DIC images represent extension while the negative ness.
values represent compression. The size of PIV images was optimized by For the sample compacted at water content of 10.5%, three points
selecting the zone just around the neck section of the sample, trying to were selected from the tensile deformation curve corresponding to
have a more clear observation of the displacement vectors (as marked different tensile stages (Fig. 8(a)). The related DIC and PIV images are
in Figs. 8(b), 9(b) and 10(b)). shown in Fig. 8(b). In stage I, the tensile stress increases linearly with
For samples compacted at water content of 6.5%, four points were increase in displacement. There is no obvious mutation or strain con-
selected from the tensile curve (Fig. 7(a)). Point (1) was selected from centration in ROI. This means that no local damage occurred. After the
stage I at displacement about 0.1 mm. At this initial stage, the tensile tensile stress reached the maximum value, tensile strain concentration
strain and displacement cannot be well identified from the DIC and PIV starts to develop at the left side of the central section. This is reflected
images (Fig. 7(b)). Point (2) was selected from stage II referring to the by the DIC image of point (2) in which the magnitude of tensile strain is
maximum tensile stress. A tensile strain concentration of up to 0.004 up to about 0.008, and the fracture was going to appear (Fig. 8(b). The

144
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

(a) (3)
20 S6-1

(2)
(4)

Tensile stress (σ ), kPa

15

Stage I Stage II Stage III

10

5
(5)
(1)
0

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Tensile displacement (s), mm

(b)

Fig. 9. (a) Typical tensile curve, and (b) the images of ROI, DIC and PIV related to different tensile stages for the sample with water content of 20.5%.

point (3) refers to a state well into the overall failure stage. The fracture decreased gradually. From the displacement vectors shown in the PIV
was distinguished on the sample surface and penetrated the whole neck images, it can be seen that the direction of movement of the soil around
section of the sample, showing a tensile strain magnitude of up to 0.45. the fracture was not completely perpendicular to the direction of
Similar to the observation in Fig. 7(a), the stage II in Fig. 8(a) also took fracturing. The upper part moved to the upper left corner, while the
a very short time, indicating that the sample compacted at water con- lower part moved to the lower left corner. Also, the soil on the right side
tent of 10.5% still presented significant failure brittleness. had a tendency to move to the left side. This is probably due to the
For the sample compacted at water content of 20.5%, the typical uneven distribution of tensile stress across the failure area. One of the
tensile curve is plotted in Fig. 9(a). A total of five points were selected factors contributing to this problem, could be the boundary effect in-
from the tensile curve for DIC and PIV analysis, the related strain and duced by the shape of the tensile test mould or the preparation of the
displacement patterns are shown in Fig. 9(b). Point (1) was selected sample. Similar observations were indicated by Tollenaar et al.
from stage I which refers to a tensile displacement of about 0.2 mm. At (2017b). Point (5) was selected from stage III where the tensile stress
this early tensile stage, no obvious tensile strain and displacement was was close to zero. The fracture was clear and distinct. It crossed the
observed from the corresponding DIC and PIV image. Once the tensile neck section of the sample and a tensile strain magnitude of up to 0.2.
stress reached the maximum value (e.g. point (2) at stage II), a tensile Differently from Figs. 8(a) and 9(a), the stage II in Fig. 9(a) was much
strain concentration zone with a magnitude of around 0.011 was de- longer, indicating that the failure process was more ductile for the
tected at the left central section where the fracture was going to appear. sample compacted at higher water content. Specifically, this phenom-
Points (3) and (4) were also selected from stage II just a little behind enon is more pronounced when w ≥ 12.5% (Fig. 6(a)). Based on the
point (2). They show similar deformation patterns but the tensile strain above analysis, it is herein confirmed that the tensile stress-strain be-
concentration zone and the related strain magnitude increased with havior and the failure process of soil are strongly linked to water con-
increasing tensile displacement (Fig. 9(b)). The tensile strain con- tent.
centration expanded from left to right which controlled the propagation
direction of the fractures. During this process, the tensile stress

145
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

Fig. 10. The cracking tracks and fracture surfaces of samples at different water contents, (a) S1–1 (w = 6.5%), (b) S2–1 (w = 8.5%), (c) S3–1 (w = 10.5%), (d) S4–1
(w = 12.5%), (e) S5–1 (w = 16.5%), (f) S6–1 (w = 20.5%).

4.4. Effects of water content on failure surface characteristics content on tensile strength characteristics were studied. The evolution
of deformation and strain field during tension was analyzed using PIV
The geometrical characteristics of the tensile failure surface at dif- and DIC digital image processing techniques. The following conclusions
ferent water contents are shown in Fig. 10. For the samples at the water can be drawn:
content around wc 8.5, 10.5 and 12.5%, the failure plane is straight, and
the fracture surface is smooth. Relatively, at lower or higher water (1) The designed apparatus with “8” shaped tensile mould, compaction
content (i.e. w = 6.5, 16.5 and 20.5%), the tensile plane is tortuous and mould and piston significantly facilitates the performance of the
the fracture surface is rough. It is believed that the failure surface tensile test procedures. It provides an effective tool for investigating
characteristics at different water contents are related to microstructural soil tensile behavior and for accurately measuring soil tensile
features and the state of stress at failure. Lu et al. (2007) examined the strength.
tensile failure surface of unsaturated sand under three different (2) The tensile deformation curve of compacted soil (tensile stress
moisture states: pendular, funicular and capillary. Within the pendular versus displacement) generally presents three typical stages: stress
regime, a smooth flat failure surface was observed. Within the funicular increasing stage (I), failure developing stage (II) and post-failure
regime, rough failure surface was observed. Within the capillary re- stage (III).
gime, the failure surface was characterized by a major rough plane (3) The overall tensile behavior of the clayey soil tested significantly
intersected by other small planes, perpendicular to the major failure depends on water contents. With increasing water content, the
plane. They pointed out that the smooth and flat failure surface in- tensile stiffness and failure brittleness of samples are reduced while
dicates tensile failure under one-dimensional tensional loading condi- the ductility is improved. The failure developing stage (II) of sample
tion where the tensile stress distributes evenly along the failure plane. lasts longer at relatively higher water content condition.
The rough failure surface indicates a tensile failure under a three-di- (4) There is a critical water content, wc related to the maximum tensile
mensional tensional stress condition where the tensile stress distributes strength that divides the soil tensile strength characteristic curve
unevenly along the failure plane. into two distinct phases. On the dry side of the wc, tensile strength
increases with increasing water content. While at the wet side of the
5. Conclusions wc, tensile strength decreases with increasing water content. It
highlights that the tensile strength characteristics of compacted
A new direct tensile test apparatus was designed, built and used to clayey soil cannot be explained solely by unsaturated soil me-
investigate soil tensile deformation behavior. The effects of water chanics or suction theory. Evolution of microstructure with changes

146
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

in water content should also be taken into account. Ledesma, A., Lakshmikantha, M.R., Prat, P.C., 2018. Boundary effects in the desiccation
(5) Digital image processing techniques like PIV and DIC are suitable of soil layers with controlled environmental conditions. Geotech. Test. J. 41 (4).
https://doi.org/10.1520/GTJ20170018.
for use in analyzing the tensile deformation and strain concentra- Li, Y., 2018. A review of shear and tensile strengths of the Malan Loess in China. Eng.
tion during the whole test. That can provide important information Geol. 236, 4–10.
on various stages of tensile deformation of soils. Generally, the lo- Lian, G., Thornton, C., Adams, M.J., 1993. A theoretical study of the liquid bridge forces
between two rigid spherical bodies. J. Colloid Interface Sci. 161 (1), 138–147.
cation of tensile fracture and the direction of propagation of major Liang, Q., Wu, X., Li, C., Wang, L., 2014. Mechanical analysis using the unconfined pe-
cracks can be pre-determined through the use of obtained strain netration test on the tensile strength of Q3 loess around Lanzhou City, China. Eng.
concentration information, which reflects the tensile stress dis- Geol. 183, 324–329.
Lu, N., Likos, W.J., 2006. Suction stress characteristic curve for unsaturated soil. J.
tribution characteristics of the soil. The tensile stress reaches its Geotech. Geoenviron. 132 (2), 131–142.
maximum at the moment just before the onset of fracturing. Lu, N., Wu, B., Tan, C.P., 2007. Tensile strength characteristics of unsaturated sands. J.
Geotech. Geoenviron. Eng. 133 (2), 144–154.
Lu, N., Kim, T.H., Sture, S., Likos, W.J., 2009. Tensile strength of unsaturated sand. J.
Acknowledgments
Eng. Mech. 135 (12), 1410–1419.
Mantri, S., Bulut, R., 2014. Evaluating performance of a chilled mirror device for soil total
This work was gratefully supported by the National Natural Science suction measurements. In: Advances in Transportation Geotechnics and Materials for
Foundation of China (Grant No. 41572246, 41772280), Natural Science Sustainable Infrastructure, pp. 56–64.
Mathieu, F., Hild, F., Roux, S., 2011. Fatigue crack propagation law measured from in-
Foundation of Jiangsu Province (BK20171228, BK20170394), and the tegrated digital image correlation: the example of Ti35 thin sheets. Procedia Eng. 10
Fundamental Research Funds for the Central Universities. The authors (4), 1091–1096.
would like to thank the two anonymous reviewers for their valuable Mesbah, A., Morel, J.C., Walker, P., Ghavami, K., 2004. Development of a direct tensile
test for compacted earth blocks reinforced with natural fibers. J. Mater. Civ. Eng. 16
comments and suggestions. Dr. M.V. Villar, the guest editor of this (1), 95–98.
paper, is acknowledged for her valuable comments and help to correct Morris, P.H., Graham, J., Williams, D.J., 1992. Cracking in drying soils. Can. Geotech. J.
the language. Special thanks to Prof. H.I. Inyang for the great effort to 29 (2), 263–277.
Nahlawi, H., Chakrabarti, S., Kodikara, J., 2004. A direct tensile strength testing method
improve the quality and language of this paper. Also, the authors ex- for unsaturated geomaterials. Geotech. Test. J. 27 (4), 356–361.
tend their thanks to Mr. Zhang Qi, Mr. Wang De-Yin, and Mr. Xu Qi- Narvaez, B., Aubertin, M., Saleh-Mbemba, F., 2015. Determination of the tensile strength
Liang at Nanjing University for their contributions to the work. of unsaturated tailings using bending tests. Can. Geotech. J. 52 (11), 1874–1885.
Orr, F.M., Scriven, L.E., Rivas, A.P., 1975. Pendular rings between solids: meniscus
properties and capillary force. J. Fluid Mech. 67 (2), 723–742.
References Peron, H., Hueckel, T., Laloui, L., Hu, L.B., 2009. Fundamentals of desiccation cracking of
fine grained soils: experimental characterisation and mechanisms identification. Can.
Geotech. J. 46, 1177–1201.
Adrian, R.J., 1999. Particle imaging techniques for experimental fluid mechanics. Annu.
Ranson, W.F., Peters, W.H., 1982. Digital image techniques in experimental stress ana-
Rev. Fluid Mech. 23 (1), 261–304.
lysis. Opt. Eng. 21 (3), 427–431.
Albrecht, B.A., Benson, C.H., 2001. Effect of desiccation on compacted natural clay. J.
Rodríguez, R.R., 2006. Hydrogeotechnical characterization of a metallurgical waste. Can.
Geotech. Geoenviron. Eng. 127 (1), 67–75.
Geotech. J. 43 (10), 1042–1060.
Alonso, E.E., Gens, A., Josa, A., 1990. A constitutive model for partially saturated soils.
Rodríguez, R., Sanchez, M., Ledesma, A., Lloret, A., 2007. Experimental and numerical
Géotechnique 40 (3), 405–430.
analysis of desiccation of a mining waste. Can. Geotech. J. 44 (6), 644–658.
Alonso, E.E., Pereira, J.M., Vaunat, J., Olivella, S., 2010. A microstructurally based ef-
Romero, E., 2013. A microstructural insight into compacted clayey soils and their hy-
fective stress for unsaturated soils. Géotechnique 60 (12), 913–925.
draulic properties. Eng. Geol. 165, 3–19.
Ávila, G., 2004. Study on the Cracking Behaviour of Clay. Application to Bogota Clay.
Romero, E., Simms, P.H., 2008. Microstructure investigation in unsaturated soils: a re-
Ph.D. thesis. Technical Univ. of Catalonia, Barcelona, Spain.
view with special attention to contribution of mercury intrusion porosimetry and
Barzegar, A.R., Oades, J.M., Rengasamy, P., Murry, R.S., 1995. Tensile strength of dry,
environmental scanning electron microscopy. Geotech. Geol. Eng. 26 (6), 705–727.
remoulded soils as affected by properties of the clay fraction. Geoderma 65, 93–108.
Romero, E., Gens, A., Lloret, A., 1999. Water permeability, water retention and micro-
Corte, A., Higashi, A., 1960. Experimental research on desiccation cracks in soil. In:
structure of unsaturated compacted Boom clay. Eng. Geol. 54 (1–2), 117–127.
Research Report 66. U.S. Army Snow Ice and Permafrost Research Estabilishment,
Romero, E., Della Vecchia, G., Jommi, C., 2011. An insight into the water retention
Wilmette, Illinois.
properties of compacted clayey soils. Géotechnique 61 (4), 313–328.
Costa, S., Kodikara, J., Shannon, B., 2012. Salient factors controlling desiccation cracking
Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH image to ImageJ: 25 years of
of clay in laboratory experiments. Géotechnique 63 (1), 18–29.
image analysis. Nat. Methods 9 (7), 671.
Delage, P., Marcial, D., Cui, Y.J., Ruiz, X., 2006. Ageing effects in a compacted bentonite:
Sobhan, K., Mashnad, M., 2002. Tensile strength and toughness of soil-cement-fly-ash
a microstructure approach. Géotechnique 56 (5), 291–304.
composite reinforced with recycled high-density polyethylene strips. J. Mater. Civ.
Divya, P.V., Viswanadham, B.V.S., Gourc, J.P., 2014. Evaluation of tensile strength-strain
Eng. 14 (2), 177–184.
characteristics of fiber reinforced soil through laboratory tests. J. Mater. Civ. Eng. 26
Stanier, S.A., Blaber, J., Take, W.A., White, D., 2015. Improved image-based deformation
(1), 14–23.
measurement for geotechnical applications. Can. Geotech. J. 53 (5), 727–739.
Fang, H.Y., Chen, W.F., 1972. Further study of double-punch test for tensile strength of
Stirling, R.A., Hughes, P., Davie, C.T., Glendinning, S., 2015. Tensile behaviour of un-
soils. In: Proceedings of the 3rd Southeast Asian Conference on Soil Engineering,
saturated compacted clay soils – a direct assessment method. Appl. Clay Sci.
Hong Kong, China.
112–113.
Festugato, L., da Silva, A.P., Diambra, A., Consoli, N.C., Ibraim, E., 2018. Modelling
Tang, G.X., Graham, J., 2000. A method for testing tensile strength in unsaturated soils.
tensile/compressive strength ratio of fibre reinforced cemented soils. Geotext.
Geotech. Test. J. 23 (3), 377–381.
Geomembr. 46 (2), 155–165.
Tang, C.S., Shi, B., Liu, C., Zhao, L., Wang, B.J., 2008. Influencing factors of geometrical
Frydman, S., 1964. The application of the Brazilian (indirect tension) test to soils. Aust. J.
structure of surface shrinkage cracks in clayey soils. Eng. Geol. 101, 204–217.
Appl. Sci. 15, 335–343.
Tang, C.S., Cui, Y.J., Tang, A.-M., Shi, B., 2010. Experiment evidence on the temperature
Gens, A., Alonso, E.E., 1992. A framework for the behaviour of unsaturated expansive
dependence of desiccation cracking behavior of clayey soils. Eng. Geol. 114 (34),
clays. Can. Geotech. J. 29 (6), 1013–1032.
261–266.
Helmar, S., 1975. Tensile strength of agglomerates. Powder Technol. 11 (2), 107–119.
Tang, C.S., Shi, B., Liu, C., Suo, W.B., Gao, L., 2011. Experimental characterization of
Hesselink, L., 1988. Digital image processing in flow visualization. Annu. Rev. Fluid
shrinkage and desiccation cracking in thin clay layer. Appl. Clay Sci. 52 (1–2), 69–77.
Mech. 20 (1), 421–486.
Tang, C.S., Pei, X.J., Wang, D.Y., Shi, B., Li, J., 2015. Tensile strength of compacted clayey
Higo, Y., Oka, F., Sato, T., Matsushima, Y., Kimoto, S., 2013. Investigation of localized
soil. J. Geotech. Geoenviron. 141 (4), 1–8.
deformation in partially saturated sand under triaxial compression using microfocus
Tang, C.S., Wang, D.Y., Cui, Y.J., Shi, B., Li, J., 2016. Tensile strength of fiber-reinforced
X-ray CT with digital image correlation. Soils Found. 53 (2), 181–198.
soil. J. Mater. Civ. Eng. 28 (7), 04016031.
Ismaiel, H.A.H., 2006. Treatment and Improvement of the Geotechnical Properties of
Thusyanthan, N.I., Tanke, W.A., Madabhushi, S.P.G., Bolton, M.D., 2007. Crack initiation
Different Soft Fine-Grained Soils Using Chemical Stabilization. Ph.D. thesis. Martin-
in clay observed in beam bending. Géotechnique 57 (7), 581–594.
Luther University of Halle-Wittenberg, Saxony-Anhalt, Germany.
Tollenaar, R.N., van Paassen, L.A., Jommi, C., 2017a. Observations on the desiccation and
Kim, T.H., Hwang, C., 2003. Modeling of tensile strength on moist granular earth material
cracking of clay layers. Eng. Geol. 230, 23–31.
at low water content. Eng. Geol. 69, 233–244.
Tollenaar, R.N., van Paassen, L.A., Jommi, C., 2017b. Experimental evaluation of the
Kim, T.H., Kim, C.K., Jung, S.J., Lee, J.H., 2007. Tensile strength characteristics of con-
effects of pull rate on the tensile behavior of a clay. Appl. Clay Sci. 144, 131–140.
taminated and compacted sand-bentonite mixtures. Environ. Geol. 52 (4), 653–661.
Towner, G.D., 1987. The mechanics of cracking of drying clay. Span. J. Agric. Res. 36 (2),
Konrad, J.M., Ayad, R., 1997. An idealized framework for the analysis of cohesive soils
115–124.
undergoing desiccation. Can. Geotech. J. 34 (4), 477–488.
Trabelsi, H., Jamei, M., Zenzri, H., Olivella, S., 2012. Crack patterns in clayey soils: ex-
Lakshmikantha, M.R., Prat, P.C., Ledesma, A., 2012. Experimental evidence of size effect
periments and modeling. Int. J. Numer. Anal. Methods Geomech. 36 (11),
in soil cracking. Can. Geotech. J. 49 (3), 264–284.
1410–1433.
Lambe, T., Whitman, R., 1979. Soil Mechanics. Wiley, New York.
Trabelsi, H., Romero, E., Jamei, M., 2018. Tensile strength during drying of remoulded

147
H.-D. Li, et al. Engineering Geology 253 (2019) 137–148

and compacted clay: the role of fabric and water retention. Appl. Clay Sci. 162, Wang, D.Y., Tang, C.S., Shi, B., Li, J., 2016. Studying the effect of drying on soil hydro-
57–68. mechanical properties using micro-penetration method. Environ. Earth Sci. 75 (12),
Varsei, M., Miller, G.A., Hassanikhah, A., 2016. Novel approach to measuring tensile 1009.
strength of compacted clayey soil during desiccation. Int. J. Geomech. 16 (6), Wang, L.L., Zhang, G.Q., Hallais, S., Tanguy, A., Yang, D.S., 2017. Swelling of shales:
D4016011. amultiscale experimental investigation. Energy Fuel 31 (10), 10442–10451.
Venkataramana, K., Rao, B.H., Singh, D.N., 2009. A critical review of the methodologies Wang, L.L., Tang, C.S., Shi, B., Cui, Y.J., Zhang, G.Q., Hilary, I., 2018. Nucleation and
employed for determination of tensile strength of fine-grained soils. J. Test. Eval. 37 propagation mechanisms of soil desiccation cracks. Eng. Geol. 238, 27–35.
(2), 532. White, D.J., Take, W.A., Bolton, M.D., 2003. Soil deformation measurement using particle
Villar, L.F.S., de Campos, T.M.P., Azevedo, R.F., Zornberg, J.G., 2009. Tensile strength image velocimetry (PIV) and photogrammetry. Géotechnique 53 (7), 619–632.
changes under drying and its correlations with total and Matric Suctions. In: Yamaguchi, I., 1981. A laser-speckle strain gauge. J. Phys. E Sci. Inst. 14 (11),
Proceedings of the 17th International Conference of Soil Mechanics and Geotechnical 1270–1273.
Engineering, Alexandria, Egypt, pp. 793–796. Yin, P., Vanapalli, S.K., 2018. Model for predicting tensile strength of unsaturated co-
Wang, J.J., Zhu, J.G., Chiu, C.F., Zhang, H., 2007. Experimental study on fracture hesionless soils. Can. Geotech. J. 55 (9), 1313–1333.
toughness and tensile strength of a clay. Eng. Geol. 94 (1–2), 65–75. Zhu, C.H., Liu, J.M., Yan, B.W., Ju, J.L., 2008. Experimental study on relationship be-
Wang, Y., Cui, Y.J., Tang, A.M., Tang, C.S., Benahmed, N., 2015. Effects of aggregate size tween tensile and shear strength of unsaturation clay earth material. Chin. J. Rock
on water retention capacity and microstructure of lime-treated silty soil. Géotech. Mech. Eng. 27 (Supp.2), 3453–3458 (in Chinese).
Lett. 5 (4), 269–274.

148