Cold Regions Science and Technology 141 (2017) 1–7

Contents lists available at ScienceDirect

Cold Regions Science and Technology

journal homepage: www.elsevier.com/locate/coldregions

A novel method for estimating the elastic modulus of frozen soil MARK
⁎
Feng Ming , Dongqing Li, Mingyi Zhang, Yu Zhang
State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000,
China

A R T I C L E I N F O A B S T R A C T

Keywords: In order to provide more reliable design parameters for geotechnical engineering, a completely analytical theory,
Frozen soil which based on the assumption of linear elastic deformation in tension, is developed to determine the
Measured deformation compressive elastic modulus and the tensile elastic modulus of materials simultaneously. With the Brazilian test
Compressive and tensile elastic modulus method, the quantity-relationship between the displacement and the elastic modulus was derived. Five Brazilian
Brazilian test
splitting tests are used to check out the operability of the proposed test theory. The results show that the
compressive elastic modulus is larger than the tensile elastic modulus under all loading conditions adopted in the
test. The two elastic moduli in tension and compression change with the elapsed time when the samples are still
in the elastic deformation stage. This is different from the previous consensus that is the elastic modulus is
treated as a constant during the elastic deformation stage. This paper explores the reasons for those differences
from the calculation method. The test results indicate that the proposed test method is feasible and liable, and
can be treated as a convenient way to determine the elastic modulus through the Brazilian test. It was further
thought that a new approach could be provided to estimate the compressive and tensile elastic moduli in
Brazilian tests.

1. Introduction the direct tensile methods were usually carried out on dumbbell
samples (Akagawa and Nishisato, 2009; Christ and Park, 2010).
The tensile strength characteristics of frozen soil have an important However, due to some limitations, such as, difficult sample preparation,
role in the geocryology, both in the fundamental research and the poor clamping condition and eccentricity loading, the direct test
engineering application. On the one hand, tensile strength has a method was usually not adopted. As an alternative, the indirect method
potential influences on ice lens initiation. According to the publicized was present and has several forms, such as four point bending test
literatures (Akagawa and Nishisato, 2009; Azmatch et al., 2011; Ming (Azmatch et al., 2011), the Brazilian test (Shen et al., 1994a; Zhou
et al., 2016), a new ice lens will be formed when the water pressure, ice et al., 2015). Since the Brazilian test was invented in 1940s, it has been
pressure or ice-water interface pressure exceeds the sum of overburden widely used to determine the tensile strength for rock and concrete
pressure and the tensile strength. On the other hand, numerous materials. Hondros (1959) formulated a complete stress solution for the
structures whether they are completed or being under construction in case of a radial load distributed over a finite circular arcs of the disc.
the permafrost regions, the stabilities of the structures are strongly Shloido (1968) through the Brazilian test, investigated the tensile
influenced by the macro mechanical characteristics, such as the tensile strength of frozen sand, sandy loam, clay and loam. Zhou et al.
and the compress behaviors of frozen soil (Christ and Park, 2010; Ma (2015) studied the tensile strength for frozen soil samples under various
and Wang, 2015). Furthermore, the artificial freezing technique has temperatures warmer than − 2 °C through the Brazilian splitting
been widely used in underground engineering (Gianpiero et al., 2015). method. However, concentrated loading will cause the stress concen-
As a very important parameter of frozen soil, the tensile strength tration and result in the break firstly at the loading point. Latter, some
characteristics has a remarkable influence on the stability of the relevant modifications of the traditional Brazilian test were put forward
underground engineering and needs farther investigation. and studied (Cauwellaert and Eckmann, 1994; Yu et al., 2009). As an
Generally, there are two methods to obtain the tensile strength. The improved test method, two flats are induced to the Brazilian disc as
first one is the direct test. The second method is the indirect test. In the loading surfaces for improving the stress states (Wang et al., 2004).
earlier studies, in order to investigate the tensile strength of frozen soils, Furthermore, the International Society for Rock Mechanics (ISRM)

⁎
Corresponding author at: State Key Laboratory of Frozen Soil Engineering, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences,
Lanzhou, Gansu 730000, China.
E-mail address: [email protected] (F. Ming).

http://dx.doi.org/10.1016/j.coldregions.2017.05.005
Received 19 April 2016; Received in revised form 11 April 2017; Accepted 20 May 2017
Available online 22 May 2017
0165-232X/ © 2017 Published by Elsevier B.V.
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

recommended the Brazilian test as a method for determining the tensile

strength of rock materials (ISRM, 1978), the Brazilian test was also
standardized by ASTM (2008).
In recent years, many researchers have paid attention to the
problem that is how to determine the elastic modulus from the
Brazilian test (Cauwellaert and Eckmann, 1994; Wang et al., 2004;
Elghazel et al., 2015). Based on the tensile strain, a formula was given
to calculate the tensile deformation modulus of rock (Liu, 1996). In
order to record the tension strain, two strain gauges were pasted at the
center of the disc. However, the result was affected by the length of
strain gage. As an improvement, the strain gage was glued horizontally
at the central point to record the strain (Ye et al., 2009). Therefore, the
result of tensile elastic modulus is more reliable than that determined
by Liu's method. Based on the analytical theory presented by Ye et al.
(2009), Gong et al. (2010) developed a method to determine the tensile
elastic modulus through monitoring displacements between two points
located on the circular disc. Liu (2010) used the optical digital image
correlation (DIC) technique to obtain the displacement field on the
specimen surface and proposed a scheme for determining the elastic Fig. 1. Loading state of the Brazilian disc.

constants from the measured displacement field and the applied load. In
a conclusion, many researchers have attempted to obtain more 2P ⎡ cos3 θ1 cos3 θ2 ⎤ 2P
σy = ⎢ + ⎥−
information besides the tensile strength of materials from the Brazilian πL ⎣ r1 r2 ⎦ πDL (2)
test.
where σx is the normal stress in x-direction; σy is the normal stress in y-
In the previous literatures, the test material was assumed as a
direction; P is the applied loading; L is the length of the sample; D is the
homogeneous material. Due to the difference of compressive and tensile
diameter of the sample; θ1 and θ2 are the angles; r1 and r2 are the
behavior of materials, the assumption, the tensile elastic modulus is
distances from the point M to the loading contact points A and B
considered the same with the compressive elastic modulus, is inap-
(Fig. 1), respectively.
propriate for geomaterials. As an improvement, more attention has
been paid to the tensile elastic modulus (Liu, 1996; Li and Yin, 1998).
To determine the tensile elastic modulus of rock materials, two strain 2.2. Stress field of the Brazilian test
gauges were pasted at the center of a Brazilian disc, and the tensile
elastic modulus can be calculated according to those related formula According to the geometry relationship (Fig. 1), the following
(Ye et al., 2009). A method to determine the tensile elastic modulus formula can be written,
through the measured displacement between two points located on the 0.5∗D + y 0.5∗D + y
circular disc was present by Gong et al. (2010) and Ye et al. (2012). In cos θ1 = =
r1 0.5∗D + y 2 + x 2 0.5 (3)
order to calculate the tensile elastic modulus through the horizontal
displacement, the difference between the compressive and tensile 0.5∗D − y 0.5∗D − y
cos θ2 = =
modulus is ignored (Ye et al., 2012). Actually, this simplification is r2 0.5∗D − y 2 + x 2 0.5 (4)
inappropriate for orthotropic materials, such as soil and rock materials.
To summarize, research on the tensile strength characteristics has x x
sin θ1 = =−
been focused on the following two aspects: (1) How to determine the r1 ((0.5∗D + y )2 + x 2 )0.5 (5)
tensile strength of soil and rock materials? (2) How can we improve the x x
test method to obtain more accurate test information? On the basis of sin θ2 = =−
r1 0.5∗D − y 2 + x 2 0.5 (6)
the primary research, a completely analytical theory is given out to
determine the elastic modulus in Section 2. In this theory, the Substituting Eqs. (3)–(6) into Eq. (1) and Eq. (2),
differences of tensile and compressive behaviors are adequately con-
sidered. To verify the proposed theory, five Brazilian tests are presented 2P ⎡ 0.5∗D + y∗x 2 0.5∗D − y∗x 2 1⎤
σx = ⎢ + − ⎥
in Section 3. With the help of displacement sensors, both the vertical πL ⎣ 0.5∗D + y 2 + x 2 2 0.5∗D − y 2 + x 2 2 D⎦ (7)
and horizontal displacements have been measured. Based on the
quantitative relationship between the displacement and the elastic 2P ⎡ 0.5∗D + y 3 0.5∗D − y 3 1⎤
σy = ⎢ + − ⎥
modulus, the calculation results of elastic modulus are presented in πL ⎣ 0.5∗D + y + x
2 2 2 2
0.5∗D − y + x 2 2
D⎦ (8)
Section 4. To better understand the proposed theory, some relative
contents are discussed in Section 5. Finally, some conclusions are drawn The stress field of the Brazilian disc based on elasticity mechanics
in Section 6. has been known completely according to Eqs. (7) and (8). It is the
theoretical foundation for measuring the elastic modulus of material
2. Analytical algorithm to calculate the elastic module through the Brazilian test.

2.1. Principles of Brazilian test 2.3. Estimation of elastic modulus

For a traditional Brazilian disc, a pair of diametrically opposite, From the view of physical and mechanical property of the geotech-
symmetric and compressive line loads were applied to a disc (Fig. 1). nical materials, the difference between compressive and tensile elastic
For an isotropic Brazilian disc subjected to concentrated loads, the modulus should be considered in the analytical solution of stress field in
solution for the stress can be described as (Muskhelishvili, 1958), the Brazilian test. It means that the strain is influenced by both the
compressive elastic modulus and the tensile elastic modulus.
2P ⎡ sin2 θ1 cos θ1 sin2 θ2 cos θ2 ⎤ 2P When the point M on the x-axis (y = 0), then Eqs. (7) and (8) can be
σx = ⎢ + ⎥−
πL ⎣ r1 r2 ⎦ πDL (1) simplified as,

2
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

2P ⎡ 16D 2 ∗x 2 ⎤ y
σx = ⎢ 2
πLD ⎣ D + 4x 2 2
− 1⎥
⎦ (9)
σ
uyt = 2 E0 D ∫
c
0
( 4D 2
D 2 − 4y 2 )
− 1 + μ dy

σ0 ⎡ ⎤
σy =
2P ⎡
⎢
4D 4 ⎤
− 1⎥
= D ⎢4 ln
Ec ⎣ (1 + ) + 2 D2
b2
D
b
b
tan−1 D + μ − 1⎥
⎦ (21)
πLD ⎣ D 2 + 16x 2 2 ⎦ (10)
where b is the width of the steel bars.
When the point M on the y-axis (x = 0), then Eqs. (7) and (8) can be Eq. (21) provided an approximate value of the total vertical
simplified as, displacement of the diameter AB, however, the accuracy of this method
was not verified. Based on Cauwelaert's result, Wang et al. (2004)
2P
σx = − through the numerical results modified the expression of the total
πDL (11)
vertical displacement,

2P ⎡ 4D 2 ⎤
σ0 ⎡ ⎛ D2 ⎞ ⎤
σy = ⎢ 2 − 1⎥ uyt = D ⎢4 ln⎜1 + 2 ⎟ + μ − 1⎥
πLD ⎣ D − 4y 2 ⎦ (12) Ec ⎣ ⎝ b ⎠ ⎦ (22)
According to Hooke's law, the vertical strain on diameter CD can be
Integrating Eq. (19) with Eqs. (20) and (22), using the processing
calculated by
methods presented by Cauwellaert and Eckmann (1994), Wang et al.
1 1 (2004). The vertical displacement of the diameter AB can be described
εx = σ
Es x
− Ec
μσy
(13) as,

where Es and Ec are the tensile and compressive elastic modulus of ⎡ 1 ⎛ D2 ⎞ 1 1⎤

uyt = σ0 D ⎢ ln⎜1 + 2 ⎟ − +μ ⎥
the material; μ is Poisson's ratio. ⎣ Ec ⎝ b ⎠ Ec Es ⎦ (23)
Substituting Eqs. (9) and (10) into Eq. (13),
In the proposed analytical algorithm, the difference between the
σ0 ⎡ 16D 2∗x 2 ⎤ σ ⎡ 4D 4 ⎤
compressive and tensile elastic module has been taken into considera-
εx = ⎢ 2 − 1⎥ − μ 0 ⎢ − 1⎥
Es ⎣ D + 4x 2 2 ⎦ Ec ⎣ D + 16x
2 2 2 ⎦ (14) tion. As long as the (vertical and horizontal) displacement and the
applied force are recorded in the Brazilian test, then the elastic modulus
Similarly, the vertical strain on diameter AB can be calculated by
can be determined by Eqs. (20) and (23). Accordingly, the most
Eq. (15)
important problem needed to be solved is to record the force-displace-
σ0 ⎛ 4D 2 ⎞ σ ment curve, especially when the sample is in the elastic state.
εy = ⎜ − 1⎟ + μ 0
Ec ⎝ D 2 − 4y 2 ⎠ Es (15)
3. Experimental schedule
Integrating Eqs. (14) and (15),

⎤ ⎞⎟
x x 3.1. Testing system and test method
σ0 ⎡ 16D 2∗x 2 ⎤ σ0 ⎡ 4D 4
uxt = ∫ εx dx = ∫ ⎢ 2
Es ⎣ D + 4x 2 2
− 1 ⎥
⎦
− μ ⎢
Ec ⎣ D 2 + 16x 2 2
− 1 ⎥ ⎟ dx
⎦⎠
0 0 The configuration of the test system is composed of universal
(16) material testing machine and the Brazilian splitting test apparatus, as
shown in Fig. 2.
y y
⎛ σ ⎛ 4D 2 ⎞ σ ⎞
In the test configuration (Fig. 2), two steel bars are used to apply
uyt = ∫ εy dy = ∫ ⎜ 0⎜ 2 − 1⎟ + μ 0 ⎟dy loading. Due to the dimensions of the steel bars (10 mm) is smaller than
⎝ Ec ⎝ D − 4y 2 ⎠ Es ⎠ (17)
0 0 the diameter of the soil samples (61.8 mm), the contact area between
the steel bars and the sample is very small. As a consequent, the loads
Integrating Eqs. (16) and (17),
applied by the two steel bars can be considered as two concentrated
⎡⎛ 2xD 2 ⎞⎛ 1 1⎞ ⎛ 2x ⎞⎛ 1 1 ⎞⎤ loads applied to the disc oppositely and diametrically along the
uxt = σ0 ⎢⎜ 2 ⎟⎜ + μ ⎟ + ⎜D arctan − x ⎟⎜μ − ⎟⎥ diameter.
⎣⎝ 4x + D ⎠⎝ Es
2 Ec ⎠ ⎝ D ⎠⎝ Ec Es ⎠⎦
(18)

⎛ 2D D + 2y 1 1 ⎞
uyt = σ0 ⎜ ln − y + μ y⎟
⎝ c
E D − 2y Ec Es ⎠ (19)

According to Eq. (18), the displacement at the end of the diameter

CD (x = D/2) in Fig. 1 can be described as,

⎡1⎛ π⎞ μπ ⎤
uxt = σ0 D ⎢ ⎜1 − ⎟ + ⎥
⎣ Es ⎝ 4⎠ 4Ec ⎦ (20)

In Eq. (19), if y = D/2 or y = − D/2, the vertical displacements at

the ends of diameter AB is infinite. This indicates that the displacement
of the contact points should not be used to determine the elastic
modulus of materials by adopting Eq. (19). To solve this problem, the
concentrated load was spread over a certain horizontal finite width
(Cauwellaert and Eckmann, 1994), and the total vertical displacement
can be described by Eq. (21). Details of this processing method can be
found in their original paper (Cauwellaert and Eckmann, 1994). Here,
we only introduce some necessary results. Fig. 2. The configuration of the test system.

3
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

tests under the condition of loading with constant strain rate are
proposed. As a consequent, the vertical displacement increased propor-
tional with elapsed time. It should be noted that the recorded vertical
displacement in Fig. 5 is two times of the real displacement of points A
or point B.
At the beginning of the test, small tensile stress occurred near the
point C and point D, due to the sample was being squeezed. This tensile
stress results in the micro fractures, especially in the region near the
line CD. Consequently, the horizontal displacement at point C or D
increases. With the increasing of applied force, the lateral deformation
increased. As a result, the horizontal displacement increased. When the
sample was destroyed, the applied force reached the peak value, and
decreased afterwards (Fig. 7). However, the vertical and horizontal
displacements still increased with the loading time (Figs. 5 and 6).
According to the test theory proposed in this study (Section 2), the
Fig. 3. Result of grain-size analysis.
variation processes of the elastic modulus of the five samples were
determined, as shown in Figs. 8 and 9.
3.2. Sample preparation
As shown in Fig. 8, it can be seen that during the loading procedure,
the tensile elastic modulus Es is not a constant, but a variable, in the
The testing material was fine-grained silty clay, collected from the
range of 1.5–15 MPa. As presented in Fig. 8, two steps are involved in
alluvial plain of the Yellow River (Zhengzhou, China). The main
the loading process. The obvious trend is that the tensile elastic
physical parameters are as follows: liquid limit is 27.4%, plastic limit
modulus increases as the time and loading force increases, and rises
is 18.5% and plasticity index is 8.9. Fig. 3 presents the result of grain-
up to a peak value before the sample breaking. When the sample was
size analysis.
destroyed, the tensile elastic modulus decreased as the loading force
According to the previous studies (Shen et al., 1994b), the most
decreased. At the beginning of the Brazilian test, there is a compaction
appropriate height/diameter ratio of the sample ranged from 0.4 to 2.1.
process, resulting in the micro fractures in the vertical direction are
In this paper, the sample was designed as 61.8 mm in height and
closed and the micro fractures in the horizontal direction are opened.
61.8 mm in diameter (height/diameter = 1). The soil was mixed with
As a result, the horizontal displacement at point C or D increased, and
distill water to obtain a water content of 20%, with an approximate dry
the tensile elastic modulus generally increased. After the applied force
density about 1.78 g/cm3. The soil was placed in a sealed container for
reached the peak value, the tensile elastic modulus began to decreases,
24 h to allow for water equilibration. Then, the soil sample was made
while the horizontal displacement at point C or D still increased.
into columns, and placed into the copper jacket. Finally, put the copper
Similarly, the compressive elastic modulus Ec is not a constant, which
jacket into a freezing cabinet with a temperature of −20 °C. The copper
changes with the loading time (Fig. 9).
jacket was removed after freezing for 24 h. Ten samples were prepared,
of which five samples were used in the Brazilian test, and the others
5. Discussions
were used in the conventional test method to determine the Poisson's
ratio.
As we know, some materials, such as, soil and rock contain many
microcracks, show non-homogeneity to some extent. Accordingly, there
3.3. Experimental process are two kinds of elastic modulus: the compressive elastic modulus and
the tensile elastic modulus. In actual project, the damage or instability
In the artificial ground freezing method, the temperature of the of numerous geotechnical engineering structures is often caused by
freezing wall is of great importance. Generally, the mean temperature partial or whole structure failure under tensile stress. Furthermore, the
of the freezing wall is always around − 10 °C. Therefore, the test ratio of rock or soil's compressive and tensile strength is larger than one.
temperature in this paper was chosen as −10 °C. Before each test, the Therefore, we should treat the tensile elastic modulus differently from
soil sample was cooled to a uniform initial temperature of −10 °C. For the compressive elastic modulus. In this paper, according to the
displacement-controlled test, the strain rate of 6.18 mm/min was used theoretical elastic solution of Brazilian disc, an analytical algorithm,
in the Brazilian tests. In order to examine the reliability of the test which considered the difference between the compressive elastic
system, five repeatability tests were carried out in the laboratory. modulus and tensile elastic modulus, was presented to calculate the
Poisson's ratio was determined as 0.36 through the conventional test elastic modulus of material. This is different from the previous studies,
method. In the conventional test, the test temperature and the loading most of which only focus on the compressive elastic modulus (Wang
rate were same as the Brazilian test. et al., 2004; Yu and Wang, 2004) or the tensile elastic modulus (Liu,
In this paper, a displacement sensor was placed at the top of disc to 1996; Ye et al., 2009; Gong et al., 2010).
record the vertical displacement of points A, two displacement sensors In early time, people improved the Brazilian test to obtain more
were placed at the center on the both side faces of disc along the accurate useful information, such as compressive elastic modulus
direction to record the horizontal displacement of point C and point D, (Wang et al., 2004; Yu and Wang, 2004), tensile elastic modulus (Liu,
and a force sensor was used to record the applied force. The force and 1996; Gong et al., 2010), both the compressive elastic modulus and
displacements data were picked automatically by the acquisition system tensile elastic modulus (Ye et al., 2012). In order to obtain the required
(DT80). The measuring system was shown in Fig. 4. parameters (strain or displacement) used in these test theories men-
tioned above, the strain gauge was pasted at the center of the Brazilian
4. Experimental results disc (Liu, 1996). As an improvement, two strain gauges were pasted on
the compressive zone and tensile zone, to record the compressive and
In this section, the test results of five samples (T1–T5) are used to tensile strain, respectively (Li and Yin, 1998). However, the accuracy of
check out the feasible of the proposed test theory. Based on the above this method is difficult to be ensured (Yu and Wang, 2004; Gong et al.,
proposed analytical expressions, the elastic modulus are determined 2010). There are two reasons. The first one is that the finite length
from the experimental data. Figs. 5–7 show the recorded results of the strain is smaller than the length of the pieces resistance, which cannot
five samples. According to the experimental scheme, Brazilian splitting reflect the true strain of the disc center (Gong et al., 2010). The second

4
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

Fig. 4. The method of measuring elastic modulus with Brazilian test.

Fig. 5. Test curves of vertical deformation versus loading time. Fig. 7. Test curves of applied force versus loading time.

Fig. 8. Variation processes of the tensile elastic modulus Es.

Fig. 6. Test curves of horizontal deformation versus loading time.

has a large lateral deformation, was used in this paper, and the vertical
one is, it is difficult to for us to paste the strain gauge at the center of the
and horizontal displacements were used to calculate the elastic
tensile zone, and completely perpendicular to the tensile stress direc-
modulus. To a certain degree, this method can reduce the dating error.
tion (Ye et al., 2009). Due to these problems existing in the method of
In a conclusion, the measurement of displacement is much more
strain gauge, people are more like to use the displacement to calculate
convenient than the stain measurement by using the strain gauge.
the elastic modulus (Wang et al., 2004; Gong et al., 2010). However, as
In the previous studies, due to the vertical displacement in the
the horizontal displacement is very small, only 0.02–0.06 mm, in the
Brazilian disc under a concentrated force loading is singular at the
rock material with the Brazilian test, it is hard to measure the
contact point. For simplification, the elastic modulus was explicitly
horizontal displacement. As an improvement, the frozen soil, which

5
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

Fig. 9. Variation processes of the compressive elastic modulus Ec.

expressed only by the horizontal displacement (Ye et al., 2012). In this

paper, to solve the problem that is the displacement is singular at the
contact point, we introduced the loading flat, and the concentrated
force was spread over a certain horizontal finite width, and the total
vertical displacement can be described by Eq. (21). In the Brazilian test,
the vertical and horizontal displacements can be easily measured by the
displacement sensors. Therefore, the compressive elastic modulus and
tensile elastic modulus can be calculated by the vertical and horizontal
displacements. Compared with other methods (Gong et al., 2010; Ye
et al., 2012), this method considered the difference between the
compressive and tensile modulus.
Based on the theoretical analysis and experimental results, the
variation regularities of the compressive elastic modulus and tensile
elastic modulus were shown in Fig. 8 and Fig. 9, respectively. However,
due to the result was obtained on the assumption of linear elastic Fig. 10. Test curves of tensile stress-horizontal strain and tensile stress-vertical strain of
deformation. Therefore, the elastic modulus is available when the T3 specimen (horizontal strain = horizontal displacement/R; vertical strain = vertical
sample is in the elastic state. In the previous studies (Liu, 1996), the displacement/2R; R is the radius of the specimen).
tensile elastic modulus was calculated by the failure loads and the
corresponding strain at the center of the disc. This is incorrect. This can modulus changes with the loading time when the sample is still in the
be explained that before the applied load reaches the failure load, some elastic deformation stage. This means that there are serials of elastic
areas in the sample has been in the plastic status, and the stresses field moduli. However, this result is different from the previous knowledge
based on elastic theory cannot describe the real stresses in these areas that is the elastic modulus is treated as a constant during the elastic
(Ye et al., 2009). In order to determine whether the sample was in the deformation stage. Actually, this difference is caused by the different
elastic behavior or not, there is a simple method that is when the simplified methods to calculate the elastic modulus. In the traditional
sample is loaded by the half of failure load, and the elastic modulus is method, the elastic modulus was determined by the stress-strain datum
determined by the following formula, at a special point (Eq. (24)) or the straight slope of the stress-strain
curve (Eq. (25)). Different from the traditional methods, this paper used
σ 2
E= two equations (Eqs. (20) and (23)) to determine the elastic modulus Ex
ε (24)
and Ey. During the test process, the curves of tensile stress-horizontal
where σ is the maximum stress; σ/2 is the stress in stress-strain curve strain curve and tensile stress-vertical strain were recorded simulta-
which is half of the maximum stress; ε is the strain related to σ/2 in the neously. Extracted one datum from the two stress-strain curves,
stress-strain curve. combining with Eqs. (20) and (23), then a Es and a Ec can be obtained.
Generally, the straight slope of the stress-strain curve is defined as Due to the stress-strain curve have lots of data points, as a result there is
the elastic modulus, which can be expressed as, a serials of Es and Ec, as shown in Figs. 8 and 9. Now, facing such many
elastic moduli (Es and Ec), how can be determine a reasonable elastic
σb − σa
E= modulus for engineering design?” As a simple processing method, we
εb − εa (25)
can choose the elastic modulus when the sample is loaded by the half of
where σa and σb are the stress at the start and end point of the line failure load (Eq. (24)). Consequently, this result is same from the
section of stress-strain; εa and εb are the strain related to σa and σb in the previous knowledge that is the elastic parameters are treated as a
stress-strain curve. constant during the elastic deformation process.
In order to further illustrate the linear elastic deformation in the
test, Fig. 10 presents the test curves of tensile strength-horizontal 6. Conclusions
deformation and tensile strength-vertical deformation of T3 specimen.
It is found that there exists obvious linear relationship between stress (1) According to the theoretical elastic solution of Brazilian disc, an
and displacement, especially when the tensile strength is in the range of analytical algorithm to estimate the elastic modulus through the
100 kPa–400 kPa. Brazilian tests was presented, and the quantitative relationship
(a) Test curves of tensile stress-horizontal strain (b) Test curves of between the displacement and elastic modulus was derived. In the
tensile stress-vertical strain. analytical algorithm, the difference between compressive elastic
As shown in Figs. 8 and 9, it is demonstrated that the elastic modulus and tensile elastic modulus was considered. Different from

6
F. Ming et al. Cold Regions Science and Technology 141 (2017) 1–7

the previous test methods, the vertical displacement was recorded excavate a tunnel in sandy soil: measurements and back analysis. Tunn. Undergr.
Space Technol. 50, 226–238.
and then used to calculate the compressive elastic modulus and Gong, F.Q., Li, X.B., Zhao, J., 2010. Analytical algorithm to estimate tensile modulus in
tensile elastic modulus. Brazilian disk splitting tests. Chin. J. Rock Mech. Eng. 29, 881–891.
(2) The data of displacements and applied forces, measured from five Hondros, G., 1959. The evaluation of Poisson's ratio and the modulus of materials of a low
tensile resistance by the Brazilian (indirect tensile) test with particular reference to
Brazilian test, was used to validate the rationality of the analytical concrete. Aust. J. Appl. Sci. 10, 243–268.
algorithm. As the parameters (applied force, horizontal and vertical ISRM, 1978. Suggested methods for determining tensile strength of rock materials. Int. J.
displacements) were easy to measure, it was convenient to calculate Rock Mech. Min. Sci. 15, 99–103.
Li, W., Yin, J.G., 1998. A method for measuring both tensile and compressive elastic
the tensile and compressive elastic modulus simultaneously, with modulus of rock-like material. Rock Soil Mech. 19, 93–96.
the proposed theory. Liu, F.Z., 1996. Study of mechanical properties of rock in tension and tension-shear states.
(3) The calculated results showed that the compressive elastic modulus J. Yangtze River Sci. Res. Inst. 13, 35–39.
Liu, C., 2010. Elastic constants determination and deformation observation using
was all large than the tensile elastic modulus. It is obvious that the
Brazilian disk geometry. Exp. Mech. 50, 1025–1039.
elastic modulus changes with the elapsed time, when the sample Ma, W., Wang, D.Y., 2015. Frozen Soil Mechanics. Science Press, Beijing (366 p).
was still in the elastic deformation phase. The phenomenon is Ming, F., Zhang, Y., Li, D.Q., 2016. Experimental and theoretical investigations into the
different from the previous knowledge that is the elastic parameters formation of ice lenses in deformable porous media [J]. Geosci. J. 20 (5), 667–679.
Muskhelishvili, H.N., 1958. Some Basic Problems in Mathematic Elastic Mechanics
were treated as a constant during the elastic deformation stage. (Translated by Zhao Hui-yuan). Science Press, Beijing (478 p.).
Shen, Z.Y., Liu, Y.Z., Peng, W.W., Chang, X.X., 1994a. Application of the radial-splitting
Acknowledgment method to determining tensile strength of frozen soils. J. Glaciol. Geocryol. 16,
224–231.
Shen, Z.Y., Peng, W.W., Liu, Y.Z., Chang, X.X., 1994b. The effect of length of specimen on
This research was supported by the National Natural Science the in radial splitting test. J. Glaciol. Geocryol. 16, 327–332.
Foundation of China (41271080 and 41471063), the 100-Talent Shloido, G.A., 1968. Determining the tensile strength of frozen ground. Hydrotech.
Constr. 2, 238–240.
Program of the Chinese Academy of Sciences (Granted to Dr. Mingyi ASTM Standard, D3967-2008, 2008. Standard Test Method for Splitting Tensile Strength
Zhang) and the Funding of the State Key Laboratory Frozen Soil of Intact Rock Core Specimens.
Engineering (SKLFSE-ZT-17). Wang, Q.Z., Jia, X.M., Kou, S.Q., Zhang, Z.X., Lindqvist, P.A., 2004. The flattened
Brazilian disc specimen used for testing elastic modulus, tensile strength and fracture
toughness of brittle rocks: analytical and numerical results. Int. J. Rock Mech. Min.
References Sci. 41, 245–253.
Ye, J.H., Wu, F.G., Sun, J.Z., 2009. Estimation of the tensile elastic modulus using
Brazilian disc by applying diametrically opposed concentrated loads. Int. J. Rock
Akagawa, S., Nishisato, K., 2009. Tensile strength of frozen soil in the temperature range
Mech. Min. Sci. 46, 568–576.
of the frozen fringe. Cold Reg. Sci. Technol. 57, 13–22.
Ye, J.H., Wu, F.Q., Zhang, Y., Ji, H.G., 2012. Estimation of the bi-modulus of materials
Azmatch, T.F., Sego, D.C., Arenson, L.U., Biggar, K.W., 2011. Tensile strength and stress-
through deformation measurement in a Brazilian disk test. Int. J. Rock Mech. Min.
strain behavior of Devon silt under frozen fringe condition. Cold Reg. Sci. Technol.
Sci. 52, 122–131.
68, 85–90.
Yu, Y., Wang, T.X., 2004. Study on relationship between splitting behavior and elastic
Cauwellaert, F.V., Eckmann, B., 1994. Indirect tensile test applied to anisotropic
modulus of three Gorges granite. Chin. J. Rock Mech. Eng. 23, 3258–3261.
materials. Mater. Struct. 27, 54–60.
Yu, Y., Zhang, J.X., Zhang, J.C., 2009. A modified Brazilian disk tension test. Int. J. Rock
Christ, M., Park, J.B., 2010. Laboratory determination of strength properties of frozen
Mech. Min. Sci. 46 (2), 421–425.
rubber-sand mixtures. Cold Reg. Sci. Technol. 60, 169–175.
Zhou, G.Q., Hu, K., Zhao, X.D., Wang, J.Z., Liang, H.H., Lu, G.L, 2015. Laboratory
Elghazel, A., Taktak, R., Bouaziz, J., 2015. Determination of elastic modulus, tensile
investigation on tensile strength characteristics of warm frozen soils. Cold Reg. Sci.
strength and fracture toughness of bioceramics using the flattened Brazilian disc
Technol. 113, 81–90.
specimen: analytical and numerical results. Ceram. Int. 41, 12340–12348.
Gianpiero, R., Andrea, C., Filippo, C., Silvia, A., 2015. Artificial ground freezing to