You are on page 1of 11

Soils and Foundations 2015;55(3):626636

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

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

Seismic performance of reinforced soil wall with untreated and


cement-treated soils as backll using a 1-g shaking table
Motoyuki Suzukia,n, Naoki Shimurab, Takuto Fukumurac, Osamu Yonedad, Yukio Tasakad
a
Yamaguchi University, Japan
b
Hirose Co., Ltd., Japan
c
Fukken Co., Ltd., Japan
d
Ube Industries, Ltd., Japan
Received 15 June 2014; received in revised form 27 December 2014; accepted 6 February 2015
Available online 21 May 2015

Abstract

Reinforced soil walls were only slightly damaged in the recent earthquake disasters in Japan. However, because sandy soils are generally used as
backll in steel-strip reinforced soil walls, these walls may deform during or immediately after an earthquake. Therefore, it is important to clarify the
resistance of this type of wall to seismic shocks and to improve the wall's seismic performance for stabilisation through such methods as cement
treatment. For this purpose, we developed a reinforced soil wall model to simulate the stress and deformation around an embedded reinforcement.
The reinforcement used here was a steel strip with ribs. First, a series of conventional pullout tests was performed to examine the pullout resistance
of the strip with its ribs embedded in the soil layer. Next, shaking table tests were performed using a reinforced soil wall model with different initial
pullout loads applied to the strip. The conditions of the horizontal shaking table tests briey consisted of a maximum acceleration of 1000 Gal and a
frequency of 3 Hz. The initial pullout load before the shaking tests was set at 75% of the maximum load obtained from the pullout tests. Sand, clay,
and cement-treated sand and clay were used for the soil samples. The clay had a ne fraction content of over 25%. This paper discusses the seismic
behaviour of a reinforced soil wall composed of sand, clay, and cement-treated clay based on the results of pullout and shaking table tests. The
primary conclusions can be summarised as follows: 1) When no overburden pressure was applied, shaking caused the strip in the untreated sand
layer to slip within a couple of seconds. Conversely, when the overburden pressure was above 50 kPa, the strip did not slip at all during 3 min of
shaking. When the initial pullout load was decreased by 5%, the facing wall moved forward by only 0.2% of its height. Therefore, no large
deformations affecting the instability of the wall were recognised during the shaking tests. 2) Furthermore, in the case of untreated clay with or
without overburden pressure, the resistance between the strip and the soil was negligible, and the strip was simultaneously made to slip from the
shaking table tests to the same extent as the untreated sand without overburden pressure. 3) After the cement treatment was applied to the clay, the
strip did not slip during 3 min of shaking, the initial pullout load did not decrease, and the facing wall did not move during the shaking period. 4)
When the strip was connected to the facing wall, the strip slipped more slowly than when it was not connected. 5) The stability and seismic
performance of the reinforced soil wall were not affected by the ratio of the initial pullout load to the maximum pullout load within the tested range.
6) Conversely, when the acceleration was set to 1500 Gal, the initial pullout load decreased by only a maximum of 5%. Therefore, the use of
cement-treated soil can crucially improve the stability of a reinforced soil wall, even if a large acceleration occurs during an earthquake.
& 2015 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Reinforced soil wall; Steel strip; Retaining wall; Seismic performance; Shaking table; Stabilisation; Pullout resistance

n
Corresponding author. Tel.: +81 836 85 9303; fax: +81 836 85 9301.
E-mail addresses: msuzuki@yamaguchi-u.ac.jp (Motoyuki Suzuki), n-shimura@hirose-net.co.jp (Naoki Shimura).
Peer review under responsibility of The Japanese Geotechnical Society.

http://dx.doi.org/10.1016/j.sandf.2015.04.013
0038-0806/& 2015 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.
M. Suzuki et al. / Soils and Foundations 55 (2015) 626636 627

1. Introduction describes the results of pullout and shaking table tests. The
pullout tests were performed under conditions of monotonic
It is well known that reinforced soil walls generally showed high loading and stage loading to determine the maximum pullout
levels of anti-earthquake capability in the aftermath of both the load of the strip. The shaking tests were performed using a
South Hyogo Prefecture Earthquake of 1995 and the Noto Hanto reinforced soil wall model that was designed based on an
Earthquake of 2007 (Koseki et al., 2006, 2007). After the 2011 earlier pullout test apparatus (Suzuki et al., 2007) and was set
Great East Japan Earthquake (off the Pacic Coast of Tohoku), it up on a shaking table. While carrying out the shaking table
was reported that whole foundations below reinforced soil walls tests, we assumed that the reinforced soil wall generally
slipped due to the strong motion and that crests settled and retaining exerted its retaining effect by retaining the pretension with
walls tilted. Inherent factors, such as poor drainage or frost heaving, respect to the strip. Thus, the shaking table tests were
may have contributed to these failures (Sahara, 2012). Such damage performed using pretension based on the results of the pullout
can lead to the instability of reinforced soil walls (Sato et al., 2006). tests. The effects of the overburden pressure, the cement
However, the established theory suggests that reinforced soil walls treatment, the pretension load, and the acceleration on the
still exhibit strong earthquake resistance. The seismic performance shaking behaviour of the reinforced soil wall were examined
of reinforced soil walls using geosynthetics, such as geogrids, has separately through a series of tests. Based on the test results,
been examined through experiments and numerical analyses of we mainly discuss the characteristics of the shaking pullout
construction materials conducted under various conditions behaviour of the strip and the simultaneous displacement of the
(Richardson and Lee, 1975; Richardson et al., 1977; Matsuo retaining wall.
et al., 1998; Ling et al., 1997; Watanabe et al., 2003; Koseki
et al., 2006, 2007). Provided that a reinforced soil wall is designed 2. Reinforced soil wall model on shaking table
and constructed under appropriate standards, it will exhibit excellent
seismic resistance compared with other soil structures. However, the Fig. 1(a)(c) show our newly developed reinforced soil wall
seismic performance of a retaining wall with a steel strip reinforce- model. The sketch in Fig. 1(a) indicates how the model
ment has been examined in only a few studies (Uezawa et al., 1974;
Richardson et al., 1977; Yogendrakumar et al., 1992; Futaki et al.,
1996). Additionally, the behaviour of a reinforced soil wall over a
long duration of strong ground motion, as occurred in the recent
great earthquakes, remains unclear.
The ne fraction content and maximum grain size of backll
soil are regulated according to the Geosynthetic Reinforced
Soil Wall (GRSW), the Reinforced Railroad/Road with Rigid
Facing (RRR), Terre Arme, and Multi-Anchored Retaining
Wall (MARW) methods (Miyata et al., 2001). In a reinforced
soil wall, such as that by Terre Armee, frictional resistance is
expected to be well mobilised upon contact between the
backll soil and the steel strip. Backll soil with a ne fraction Tube for air bag Confining plate Dial gauge for
content of no more than 25% is suitable for a reinforced soil displacement of wall

wall (PWRC, 2003). A recent problem is that the supply of


suitable backll soil has been nearly exhausted, making it
difcult to procure sufcient amounts. Accordingly, there is an
increasing demand to make better use of on-site soil with high Soil tank
ne fraction content. Even construction-generated soil, such as Sh
ak
ne-particle soil, has been occasionally used in chemical ing
dir
ec
stabilisation techniques (JGS, 2006). The application of tio Bellofram cylinder for
n
pullout loading
cement or lime treatment to the backll material of a reinforced
soil wall has been conducted in a practical manner. However,
the strength characteristics of cement-treated soil, whose
Facing wall
cohesion gradually increases, have not been sufciently
considered in the design code for reinforced soil walls. The
Side wall
authors previously claried the pullout resistance character- Rubber for prevention
of sample leakage
istics of a steel strip embedded in a cement-treated soil layer
(Tasaka et al., 2010). Consequently, it remains unknown how Strip*
Teflon for reducing
much the seismic performance of a reinforced soil wall is friction
improved when using cement-treated soil as a backll material.
With this as our background and using a shaking table, the aim Fig. 1. Reinforced soil wall model on shaking table., (a) Field and laboratory
of the present study is to clarify the seismic resistance of conditions, (b) Overview of reinforced retaining wall model, and (c) Close up
retaining walls reinforced with cement-treated soil. This paper of internal structure. n: Smooth-type strip (not used in the experiments).
628 M. Suzuki et al. / Soils and Foundations 55 (2015) 626636

simulates the conditions under which tension and pullout


resistance were applied around the reinforcement. Fig. 1(b)
shows that the longer side of the model is set in the direction of
the horizontal shaking of the table. Fig. 1(c) indicates the
detailed apparatus around the facing wall. Fig. 2(a)(c) present
the cross-sectional and the top and side views of the test
apparatus. This apparatus mainly consists of a soil tank, a
facing wall, a horizontal displacement gauge, a Bellofram
cylinder, load cells, a strip, and air bags. The soil tank is
700 mm long, 300 mm high, and 200 mm wide. The horizontal
displacement of the facing wall was measured from the part of
the model connected to the displacement gauge. The strip with
ribs was 350 mm long, 60 mm wide, and 5 mm thick.
Although the strip shown in Fig. 1(c) is smooth, a ribbed strip
was used in our series of tests. The ribs were 3 mm high and
6 mm wide and were located at intervals of approximately 50
100 mm along the strip, as shown in Fig. 2(d).
The test apparatus was designed to shake when loading was
applied to the strip and to simulate the actual conditions under
which pullout stress is loaded onto the strip. The boundaries where
skin friction may be generated are at the following four locations: 1)
the interface between the airbags and the top surface of the soil
layer (a conning plate was not set between them), 2) the interface
between the side faces of the soil layer and the side wall of the soil
tank, 3) the interface between the end surface of the soil layer and
the bottom of the soil tank, and 4) the motion part between the
facing wall and the side wall of the soil tank. From our perspective,
we made the following improvements to the conditions: 1) The
airbags made contact with the top surface of the soil, but not with
the conning plate, which constrained the airbags. Hence, the
airbags did not intervene in the displacement of the soil layer. For
this reason, the shear stress acting on the top surface of the soil layer
was not signicant. 2) Silicon grease was carefully applied to the
side faces of the soil tank to remove or reduce the inuence of skin
friction. 3) The movable part of the facing wall came into contact
with other parts that were made of Teon, and the silicon grease
was able to be spread. However, it is possible that the boundary
conditions would have affected the test results regardless of whether
any mitigating measures were taken for the skin friction. In our
previous tests, the strip was not connected to the facing wall
because we had intended to compare the results with those of a test
in which the facing wall was unmovable. In fact, reinforcing strips
must be connected to facing walls in engineering practice; and thus, Fig. 2. Schematic diagrams and photograph of reinforced soil wall model, (a)
the strip and the wall were clamped together when the strip was Top view, (b) Side view, (c) Sectional image, and (d) Size and intervals of ribs
on strip.
loaded during the experiments in this study. We were able to apply
an overburden pressure through the air bags within a range of 0
hole in the wall, which goes through the strip, was protected by
100 kPa and to consolidate the soil sample embedded in the soil
rubber to avoid any leakage of water or soil.
tank under double drainage conditions.
Fig. 3 illustrates the essential features of the test system. The
overburden pressure and pullout load were controlled through the 3. Test procedure
supply of air pressure. Every measurement set in the model was
recorded through a dynamic strain amplier during static and We conducted two types of pullout tests, one in which the
dynamic tests. The pullout load mobilising on the strip, the horizontal pullout load was controlled under monotonic loading and the other
displacements of the facing wall and the strip, and the acceleration of in which it was controlled under step loading. Based on the results,
the shaking table were recorded during the shaking tests. As shown we investigated the maximum pullout load that caused the strip to
in Fig. 1(c), the outside of the facing wall was protected by Teon to slip. Next, based on the slipped load, we determined the initial
decrease the friction between the facing wall and the side wall. The pullout load for the shaking table tests. For these tests, the pullout
M. Suzuki et al. / Soils and Foundations 55 (2015) 626636 629

Data
Table 1
Recorder
Physical properties of samples used.
Data Air pressure
control panel Sample name Chiba sanda Mito clayb
Acceleration
control panel
Natural water content (%) 4.3 27.5
Density of soil particles (g/cm3) 2.719 2.751
v PC
PC
Fine fraction content (%) 14.8 68.6
Initial water content (%) 19.3 45.0
Pullout load
Soil classication [S-F] [CH]
a
Soil tank Compressor Chiba sand: Optimum water content 19.3%, Maximum dry density 1.66 g/cm3.
b
Mito clay: Liquid limit 45.0%, Plastic limit: 28.9%.
Actuator Shaking table

Fig. 3. Shaking table test and data acquisition system. 100

load was actively induced by directly pulling the strip out from the 80
soil layer. Although there was a difference between the actual and

Percentage of passing partics (%)


the laboratory test conditions, we attempted to reproduce the
conditions of the element around the strip of the reinforced soil 60

wall with our test apparatus, as shown in Fig. 1(a). In element tests,
such as triaxial compressive or hollow torsional tests, we generally 40
consider it reasonable to directly control the stress or deformation of Mito clay
the soil specimen (e.g., Hyodo et al., 1994). The tests in this study Chiba sand
were not intended to simulate all of the reinforced soil wall, but 20
only a certain part of it. Therefore, we directly controlled the
mechanical conditions of part of the reinforced soil wall. Further- 0
more, for the test apparatus, the facing wall was movable and in 0.001 0.01 0.1 1 10 100
synch with the displacement of the soil layer. Thus, we considered Grain size (mm)
that the conditions reproduced in this test were nearly the same as Fig. 4. Grading curves of soil samples.
those under the actual conditions.

facing wall was xed to the original position. An untreated or


3.1. Sample preparation cement-treated soil sample of Chiba sand was placed into the
soil tank and compacted to a wet density of 1.80 g/cm3 using a
The soil samples were Chiba sand collected from Chiba special 4.0-kg rammer. The soil layer was compacted to a
Prefecture and Mito clay collected from Yamaguchi Prefecture. compaction ratio of more than 90%. On the other hand, an
Table 1 shows the physical properties of the samples, and untreated or cement-treated soil sample of Mito clay was
Fig. 4 shows their grading curves. The ne fraction content of placed into the soil tank and densely lled with a vibrator
the sand was 14.8%, which is acceptable for backll material compacting it to a wet density of 1.70 g/cm3. After the top
standardised for a steel strip reinforced soil wall method, surface of the soil layer was levelled and smoothed to a height
whereas the clay was not acceptable unless improved through 150 mm from the bottom, a 350-mm-long strip was laid on the
stabilisation. The cementing agent was an ordinary general- surface of the soil. After the soil tank was lled to the top with
purpose cement stabiliser. The content of the cementing agent, the soil sample, a drainage layer was set on the smoothed top
Qc, was set to a constant 100 kg/m3 for the Chiba sand and 60 surface of the soil. The surface of the sample was loaded with
or 80 kg/m3 for the Mito clay. The initial water content of the the prescribed overburden pressure, v, based on the pressure
soil sample, w0, was 19.3% for the Chiba sand, which is exerted by the expanded air bags. After consolidation, the
equivalent to its optimum water content, and was 45.0% for xation of the facing wall was released. The wet density of the
the Mito clay, which is equivalent to its liquid limit. soil layer in this state remained at 1.80 g/cm3 for the Chiba
The cement stabiliser was added to the soil in powder form. sand and 1.70 g/cm3 for the Mito clay. During the tests, as the
The soil and cement mixtures were both stirred for 5 min using pullout load was increased either monotonously or in a step-
a Hobart mixer. No separation of the materials was detected by-step manner, the pullout load and horizontal displacement
when the treated samples were prepared. For the untreated of the facing wall were measured until the strip slipped.
samples, this process was ignored as a matter of course.

3.2. Pullout tests 3.3. Shaking table tests

In the pullout tests, a drainage layer of unwoven cloth and In the shaking table tests, the soil tank was lled with a soil
lter paper was set on the bottom of the soil tank, and the sample similar to the above-described pullout tests. An
630 M. Suzuki et al. / Soils and Foundations 55 (2015) 626636

1500 1.5
Chiba sand (Untreated)
1000 v = 50 kPa
1.2

Pullout load T( ( kN )
v = 0 kPa
Acceleration (gal)

500
0.9
0

0.6
-500

0.3
-1000

-1500 0.0
0 30 60 90 120 150 180 0.00 0.05 0.10 0.15
Time (sec) Displacement of facing wall d ( mm )
Fig. 5. Acceleration measured in upper frame of soil tank (Test no. S-2). Fig. 6. Relationship between pullout load and displacement of facing wall
during monotonic loading (Test nos. C-1 and C-2).
overburden pressure was then loaded onto the surface of the
soil sample. After consolidation, the initial pullout load, Ti, Chiba sand (Untreated)
obtained from the above pullout tests, was loaded onto the strip 3.0 = 100 kPa
in a step-by-step manner. The initial pullout load acted on the = 50 kPa
retaining wall immediately as a result of the active earth 2.5 = 0 kPa
pressure created by the construction of the backll. Accord-
Pullout load T (kN)

ingly, the initial pullout load corresponded to a horizontal force 2.0


through a reinforcement connected to the wall, which was
mainly due to the active earth pressure depending on the earth 1.5
pressure coefcient, the unit weight of the ll material, and the
height of the wall (PWRC, 2003; Miyata and Bathurst, 2012). 1.0
Therefore, Ti depends not only on the earth pressure of the
embankment, but also on the setting layout of the reinforce- 0.5
ment. In this study, the value for Ti was regarded merely as a
parameter. After the strip was either clamped or not clamped 0.0
with the facing wall, the horizontal shaking tests were begun 0 1 2 3 4 5 6
and continued for 3 min. The maximum acceleration was set at Displacement of facing wall d (mm)
1000 or 1500 Gal using the shaking table controller and was Fig. 7. Relationship between pullout load and displacement of facing wall
measured using an accelerometer xed on the upper frame of during stage loading (Test nos. C-3, C-4, and C-5).
the soil tank. The typical results of the measured acceleration
history are shown in Fig. 5. The shaking waveform was a
sinusoidal waveform at 3 Hz. The maximum acceleration and
frequency were determined based on observations at the speed was a constant 1.0 kN/min, and the facing wall was
ground surfaces during the recent great earthquakes. During horizontally movable. The maximum value for the pullout
testing, the pullout load and horizontal displacement of the load, Tslip, was small compared to that obtained from other
facing wall and strip were measured and recorded. When the pullout tests. We could not expect the pullout resistance to be
test procedure was nished, we observed and considered the efciently mobilised because the contact between the strip and
failure status or the conditions of the backll soil. the soil layer had been loosened by the motion of the facing
wall. A comparison of the displacement of the facing wall at
the same pullout load showed that the value of d at v 0 kPa
4. Results and discussion was larger than that at v 50 kPa. As the pullout load was
released, the displacement of the facing wall decreased
4.1. Pullout behaviour under monotonic and step loading slightly.
conditions Fig. 7 shows the relationship between the pullout load and
the displacement of the facing wall at v 0, 50 and 100 kPa.
First, we reported the results of the pullout tests under the One loading step during the experiment was used to increase
conditions of both monotonic loading and step loading. Fig. 6 the pullout load by 0.2 kN during a period of 30 s and was then
shows the relationship between the pullout load, T, and the stopped for 30 min. This step was repeated until the strip
displacement of the facing wall, d, during monotonic loading. slipped. In the case of v 0 kPa, when the pullout load
The sample used here was untreated Chiba sand. The loading reached 0.47 kN, the strip slipped suddenly. While the pullout
M. Suzuki et al. / Soils and Foundations 55 (2015) 626636 631

load was nearly constant, the displacement of the facing wall Chiba sand
5
increased rapidly. After the strip slipped, the displacement v = 0 kPa
returned to approximately 3 mm. When v 50 and 100 kPa, Treated
3
the strip slipped when T reached 2.1 and 2.7 kN, respectively. 4 Qc = 100 kg/m
In these cases, the pullout load decreased rapidly to 0 kN, but

Pullout load T (kN)


Tc = 3 days
unlike v 0 kPa, the displacement did not return. When the 3 Untreated
above results were compared with those of the test under a
continuously increasing pullout load, the pullout load at
v 0 kPa was nearly the same, whereas at v 50 kPa, it 2
was twice as large as that under a continuously increasing load.
The displacement of the facing wall under step-by-step loading 1
was considerably larger than that under continuously increas-
ing loading. The pullout resistance is essentially dependent on
0
the pullout speed (Tasaka et al., 2010). The maximum value of 0 2 4 6 8
the pullout load in the test results may be affected by the Displacement of facing wall d ( mm )
difference in the loading method used, that is, monotonic
Fig. 8. Relationship between pullout load and displacement of facing wall in
loading or step-by-step loading. Once the reinforcing strip
cases of cement-treated and untreated soils (Test nos. C-3 and C-6).
initiated slipping under continuous loading conditions, the
subsequent pullout resistance no longer increased. This nding Table 2
may be the result of an overly rapid monotonic loading speed. Test cases and results of pullout tests.
These ndings indicate that the loading speed has an inuence Test Sample Qc v Tc Tslip Pullout loading
under the conditions of an overburden pressure. no. (kg/m3) (kPa) (days) (kN) condition
Fig. 8 shows the relationship between the pullout load and the
displacement of the facing wall in cases of both untreated and C-1 Chiba 0 0 0 0.44 Monotonic loading
sand
cement-treated sand. It can be observed from this gure that the
C-2 0 50 0 1.11
maximum pullout load increases remarkably from the cement
treatment and that brittleness occurs during a pullout. The values C-3 0 0 0 0.47 Stage loading
for Tslip for all the pullout tests are summarised in Table 2. C-4 0 50 0 2.10
C-5 0 100 0 2.74
C-6 100 0 3 4.58
4.2. Shaking behaviour of reinforced soil wall model
C-7 Mito 80 50 1 8.02
clay
To reproduce the real conditions of a facing wall and a strip C-8 60 0 1 3.37
in a reinforced soil wall, a shaking table test was performed
while an initial pullout load, Ti, was applied to the strip. Qc: Content of cement.
v: Overburden pressure.
During the shaking, the pullout load, the displacement of the
Tc: Curing period of cement-treated sample.
facing wall, and the strip were all measured (Table 3). To Tslip: Maximum pullout load of strip in stage loading tests.
evaluate the seismic performance of the reinforced soil wall
model, the rate of change in the pullout load, Rt, and the rate of
change in the displacement of the facing wall, Rd, are dened Table 3
using the following equations. Here, T is the measured value of Test cases of shaking table tests.
the pullout load and H is the height of the facing wall. Test Sample Qc v Tc Ti Connection of facing
T no. (kg/m3) (kPa) (days) /Tslip (gal) wall and strip
Rt  100 % 1
Ti S-1 Chiba 0 0 0 0.75 1000 Not clamped
sand
d
Rd  100 % 2 S-2 0 50 0 0.75 1000
H S-3 0 100 0 0.75 1000
S-4 100 0 3 0.75 1000
Fig. 9(a) and (b) shows the relationships among Rt, Rd, and
the shaking time for untreated Chiba sand. The maximum S-5 0 0 0 0.75 1000 Clamped
acceleration was a constant 1000 Gal. The initial pullout load
S-6 Mito 0 0 0 0.75 1000
was set to 75% of Tslip. The strip was not clamped to the wall clay
during shaking. In the case of v 0 kPa, the strip began to slip S-7 0 50 0 0.75 1000
10 s after the initiation of shaking, and Rt was approximately S-8 60 0 1 0.75 1000
0%; that is, the pullout load decreased to nearly 0 kN. The S-9 60 0 1 0.85 1000
displacement of the wall then increased rapidly within the next S-10 60 0 1 0.50 1500
5 s. The change in displacement was too rapid and too large for Ti: Initial pullout load of strip in shaking table tests.
the displacement gauge to measure. The pullout behaviour was : Maximum acceleration of shaking table.
632 M. Suzuki et al. / Soils and Foundations 55 (2015) 626636

Chiba sand (Untreated) Mito clay (Untreated)


200
Ti/Tslip= 0.75

Rate of change of pullout load Rt (%)


200
Rate of change of pullout load Rt (%)

Ti/Tslip= 0.75
= 1000 gal = 1000 gal
v = 100 kPa = 50 kPa
v
150
150 v = 50 kPa = 0 kPa
v

v = 0 kPa
100
100

50
50

0 0
0 60 120 180 0 60 120 180

Shaking time (sec) Shaking time (sec)

1.0 Chiba sand (Untreated) Mito clay (Untreated)


Limit of Ti/Tslip= 0.75 3.0 Ti/Tslip= 0.75
Rate of change of displacement

measurement = 1000 gal


= 1000 gal Limit of measurement
v = 50 kPa

Rate of change of displacement


0.8 2.5
v = 100 kPa
of facning wall Rd (%)

= 0 kPa

of facning wall Rd (%)


v = 50 kPa
v

0.6 2.0
v = 0 kPa
1.5
0.4

1.0
0.2
0.5
0.0
0 60 120 180 0.0
Shaking time (sec) 0 60 120 180
Shaking time (sec)
Fig. 9. Seismic behaviour of facing wall with cement-treated sand under
different overburden pressures (Test nos. S-1, S-2, and S-3)., (a) Pullout load Fig. 10. Seismic behaviour of facing wall with untreated clay (Test nos. S-6
of strip), and (b) Displacement of facing wall. and S-7), (a) Pullout load, and (b Displacement of facing wall).

caused by a loss in frictional resistance between the strip and the strip did not slip considerably. The displacement of the
the backll soil, an increase in the active earth pressure acting facing wall was 0.5 mm at the front, which is 0.2% of its
on the facing wall or both. After the test, the maximum height. If the wall is 10 m in height, the displacement of the
displacement of the wall was more than 10 mm. When the wall wall will be nearly 20 mm. Thus, the displacement might not
is assumed to be 10-m high, the displacement will be be too damaging to the stability of a reinforced soil wall.
equivalent to nearly 30 cm. This means the displacement of However, if the effect of the skin friction is not perfectly
the wall may be larger in the case of an earthquake whose removed, the wall displacement may be underestimated.
horizontal acceleration is over 1000 Gal. We conducted a Therefore, an evaluation of the wall displacement should
shaking test under the condition of an overburden pressure include any inuence of the boundary conditions. The ten-
0 kPa. According to the design and construction manual, the dency at v 100 kPa was similar to that at v 50 kPa.
minimum overburden thickness was prescribed as 0.6 m, Displacement was less likely to occur in the presence of
which is equivalent to an overburden pressure of approxi- overburden pressure. During the shaking test, the wall dis-
mately 10 kPa. Although the overburden pressure acting on the placement was bound to the pullout direction when the strip
top surface was set to 0 kPa, the actual overburden pressure experienced tension. When the strip did not slip during
acting on the strip was estimated to be 3 kPa because the strip shaking, the pullout load uctuated by a few percentages of
was embedded at a 0.15-m depth from the top surface. To the pullout load during shaking. It is considered that the
clarify the inuence of the overburden pressure within the pullout load may be comparatively affected by the inertia force
estimated range on the behaviour and characteristics of the of the soil layer. The mass of the soil layer in the tank is
seismic activity, it was necessary to add test cases without an approximately 75 kg; and thus, the inertia force corresponds to
overburden pressure, i.e., v 0 kPa. On the other hand, with approximately 0.74 kN when the acceleration is 980 Gal. This
regard to the relationships among the pullout load, the wall can be equivalent to 10% of the maximum pullout resistance in
displacement, and the shaking duration at v 50 kPa and Ti/ the case of treated Chiba sand under an overburden pressure of
Tslip 0.75, the pullout load barely decreased during the 0 kPa. Although the inertia force was added or reduced to the
shaking table test. However, unlike the case of v 0 kPa, pretention pullout load, the extent of the contribution was
M. Suzuki et al. / Soils and Foundations 55 (2015) 626636 633

relatively low. We concluded that the inuence of inertia on


200
the uctuation of the pullout load was slight.

Rate of change of pullout load Rt (%)


Fig. 10 shows the test results for Mito clay in the presence
or absence of overburden pressure. The maximum acceleration 150
was 1000 Gal. The strip was clamped rigidly to the wall before
shaking commenced. In addition, Ti was set to 0.027 kN and
0.17 kN for v 0 kPa and 50 kPa, respectively. The pullout 100
resistance of the clay was very small and was determined
based on the results of another pullout test. At 15 s, after the
start of the shaking, the strip slipped and Ti decreased to 0%. 50
Simultaneously, the wall displacement was approximately
2.3%, which is equal to the measurement limit at the front.
Thus, when an improper backll material was used, the 0
0 60 120 180
resistance between the soil and the strip was negligible and Shaking time ( sec )
the wall had a poorer anti-earthquake performance. In the case
of v 50 kPa, the strip slipped at 150 s of shaking and the
pullout load decreased to 0%. In addition, the displacement of
3.0
the wall was increased 1.8%. When clay is used as a backll

Rate of change of displacement


material, the pullout load is inuenced by the overburden 2.5
pressure because the resistance is negligible. It is clear that the

of facing wall Rd (%)


strip was entirely pulled out from the clay layer even at a 2.0
higher overburden pressure.
In a real reinforced soil wall, two strips are rigidly connected 1.5

to an individual facing. We examined the interaction between


1.0
the facing wall and the strip observed during the shaking
period. Fig. 11 compares the behaviour of a wall to which a 0.5
strip was clamped or not clamped. In the test apparatus, the
displacement or the deformation of the soil layer could be 0.0
observed by the displacement of the facing wall or strip. To 0 60 120 180
evaluate the displacement of the facing wall from the Shaking time (sec)
deformation of the soil layer, a series of tests were performed Fig. 11. Seismic behaviour of facing wall clamped with strip (Test nos. S-1
under the condition in which a strip was not clamped to the and S-5), (a Pullout load of strip), and (b) Displacement of facing wall.
facing wall. The test results showed the behaviour of a wall not
connected with a strip, and that damage can be caused by
deterioration. The sample used here was untreated Chiba sand, Facing wall

and the applied overburden pressure was zero during the Frame of soil tank

shaking test. The initial pullout load was set to 0.34 kN. The
strip clamped to the wall was pulled out from the soil layer
Surface of soil
within 60 s. The changes in Rt and Rd under the clamped
condition were slower than those under the non-clamped
condition. Here, Rt was higher than under the non-clamped
condition. The actual seismic resistance was improved through Strip
an integration of a wall and a strip. For only the soil layer in
the test apparatus, we did not directly observe or measure the
deformation or displacement. The soil layer was bound to Shaking direction

displace in the shaking direction due to restraining from the


lateral displacement. Accordingly, the soil layer and the strip
Fig. 12. Deformation of soil after shaking (Test no. S-1).
were thought to be displaced in the body until the strip started
to slip because the surrounding soil layer was restrained by the
strip. Therefore, the displacement of the soil layer was tensioned strip. Fig. 12 shows that no vacancy between the
observed as the displacement of the facing wall connected facing wall and the soil layer could be conrmed after the
with the strip. On the other hand, when the strip and the facing shaking test (S-1). Therefore, the behaviour of the soil layer
wall were not clamped during the shaking test, the displace- was approximated as the displacement of the facing wall even
ment of the facing wall gradually increased, whereas the under this condition.
pullout load of the strip restraining the soil layer decreased. Fig. 13(a) and (b) shows the seismic behaviour of the
According to the observation, the displacement of the facing reinforced soil wall model using untreated and cement-treated
wall was induced by the soil deformation around the pre- Mito clay at v 0 kPa. Here, the Ti values applied to the
634 M. Suzuki et al. / Soils and Foundations 55 (2015) 626636

200 200

Rate of change of pullout load Rt (%)


Rate of change of pullout load Rt (%)

150 150

100 100

50 50

0 0
0 60 120 180 0 60 120 180
Shaking time (sec) Shaking time ( sec )

3
Rate of change of displacement

2
of retaining wall Rd (%)

0.10

0.05

0.00
0 60 120 180
Shaking time (sec)

Fig. 13. Seismic behaviours of cement-treated and untreated clay (Test nos. S- Fig. 14. Seismic behaviours of cement-treated and untreated sand (Test nos. S-1
6 and S-8), (a) Pullout load, and (b) Displacement of facing wall. and S-4) (a) Pullout load, and (b) Displacement of facing wall.

untreated and cement-treated samples were set to 0.027 kN and soil wall may not be signicantly inuenced by the initial
2.47 kN, respectively. In contrast with the untreated clay, the pullout load acting on the strip, which is smaller than Tslip.
pullout load ratio using the cement-treated clay as a backll Fig. 16 compares the results between 1000 and 1500 Gal of
material remained approximately 100% during shaking. The acceleration. Both samples used in the shaking test were
facing wall was not displaced during shaking. Based on these cement-treated clay. It should be noted that their values for
results, we conclude that the cementation of the backll Ti/Tslip differed. In the case of the 1500 Gal acceleration, the
material remarkably improves the brittle behaviour of a value for Rt decreased approximately 5% at maximum during
reinforced soil wall when using untreated clay. Fig. 14(a) the shaking period, whereas the value for Rd increased only
and (b) shows the test results for untreated and cement-treated 0.07% at maximum. These results suggest that when cement-
Chiba sand. The overburden pressure was zero, and the value treated soil is used as a backll material, the acceleration
of Ti/Tslip was set to 0.75 for both samples. The pullout load of seldom changes the pullout load. Moreover, even in the event
the cement-treated sand did not change, and the wall was of a large earthquake, whose input value of acceleration is
barely displaced as the shaking proceeded. These behaviours more than 1500 Gal, a reinforced soil wall will not be damaged
were consistent with those of the cement-treated clay. as the pullout load decreases. If the wall is 10 m in height,
Figs. 15 and 16 show the relationships among Rt, Rd, and the 0.07% of the wall's displacement ratio will be approximately
shaking time under different values of Ti and acceleration , 7 mm. The internal stability of a reinforced soil wall can be
respectively. The sample used here was treated Mito clay, and impaired only if displacement continues to occur.
the applied overburden pressure was zero during shaking. Fig. 17 shows the slipping surface of a soil tank after the shaking
When Ti was set to 85% of Tslip (Ti/Tslip=0.85), the rate of table test. A slipping surface observed during a previously
change in the pullout load increased 3% at maximum during conducted pullout test is overlaid in the image. It can be observed
the shaking period, but did not decrease below 100%. The wall from this gure that a tensile failure of the soil occurred locally
displacement was nearly the same as when Ti was set to 75% behind the facing; and thus, the collapsed soil fell. In addition, the
of Tslip (Ti/Tslip=0.75). The strip pulls out more easily as the slipping line was formed obliquely toward the pulling direction
value of Ti/Tslip increases. Nevertheless, the initial pullout load through the strip. Accordingly, the active and passive regions were
did not disappear, and the wall did not move in the direction of conrmed to be clearly divided in the soil layer. Although the two
the pullout load. Thus, the seismic performance of a reinforced slipping surfaces differed from each other around the zone in which
M. Suzuki et al. / Soils and Foundations 55 (2015) 626636 635

120 120 Mito clay (Treated)


Mito clay (Treated)
v = 0 kPa
Rate of change of pullout load Rt (%)

Rate of change of pullout load Rt (%)


= 1000 gal
3 3
Qc= 60 kg/m , Tc= 1 day Qc = 60 kg/m , Tc = 1 day
110 Ti/Tslip= 0.75 110 = 1500 gal, T /T = 0.50
i slip

Ti/Tslip= 0.85 = 1000 gal, Ti/Tslip= 0.75

100 100

90 90

80 80
0 60 120 180
0 60 120 180
Shaking time (sec)
Shaking time (sec)

0.20 Mito clay (Treated)


= 1000 gal
Rate of change of displacement

3 0.20 Mito clay (Treated)


Qc= 60kg/m , Tc= 1 day
v = 0 kPa

Rate of change of displacement


of facing wall Rd (%)

0.15 Ti/Tslip= 0.75 3


Qc = 60 kg/m , Tc = 1 day

of facing wall Rd (%)


Ti/Tslip= 0.85
0.15 = 1500 gal, Ti/Tslip= 0.50
0.10 = 1000 gal, Ti/Tslip= 0.75

0.10

0.05

0.05

0.00
0 60 120 180
Shaking time (sec) 0.00
0 60 120 180
Fig. 15. Inuence of initial pulling load on seismic behaviour of facing wall Shaking time (sec)
with cement-treated clay (Test nos. S-8 and S-9) (a) (a) Pullout load, and
Displacement of facing wall. Fig. 16. Inuence of acceleration on seismic behaviour of facing wall with
cement-treated clay (Test nos. S-8 and S-10) (a) (a) Pullout load of strip, and
(b) Displacement of facing wall.
the strip was embedded, they showed nearly the same lines at the
top and bottom of the crack.
pressure. However, deformation of the soil is likely to be caused by
shaking under low overburden pressure. It can be seen from Fig. 14
4.3. Effect of stabilisation of ll material that for untreated Chiba sand, the pre-tensioned pullout load
decreased and the wall displacement increased under low over-
The advantages of stabilization obtained from this study are burden pressure within approximately 20 s. On the other hand, both
summarized in comparison with the case of untreated soil. It can be the pullout load and the displacement for treated sand were
seen from Fig. 8 that the peak value of the pullout resistance for maintained and were not affected by the shaking. Accordingly, it
cement-treated soil obtained from the static pullout test was nearly was conrmed that the pullout resistance of a strip is signicantly
nine times that for untreated soil. In the case of untreated Mito clay, increased by the application of stabilisation to a low frictional
a small pullout resistance was exhibited under no overburden material. An effective suppression of the reduction in the pullout
pressure. By stabilizing the Mito clay, the peak value of the pullout load and the increase in the wall displacement was recognised
resistance became more than that for Chiba sand under the same during the shaking test.
overburden pressure. These tendencies are very similar to those
from a previous study (e.g., Tasaka et al., 2010). This may be
attributed to the adhesion generated in between the particles or 5. Conclusion
between the particles and the strip. Fig. 13 shows the shaking test
results for both untreated and treated Mito clay. It can also be seen The conclusions of this study can be summarised as follows:
from Fig. 13 that for untreated clay, the pullout load disappeared
and the displacement of the facing wall rapidly increased immedi- 1) As the pullout load was monotonously increased, the
ately after shaking. Conversely, both the pullout load and the wall displacement of the facing wall during the test at an
displacement for treated clay were maintained and were unaffected overburden pressure of 50 kPa became smaller than that
by the shaking. In the case of sandy soil, the pullout resistance is at 0 kPa. As the pullout load was decreased, the displace-
sufcient during an earthquake under the application of overburden ment of the facing wall was not fully reversible.
636 M. Suzuki et al. / Soils and Foundations 55 (2015) 626636

References
260 mm
m
Slipping surface Futaki, M., Ogawa, N., Sato, M., Kumada, T., and Natsume, S., 1996.
of pullout test Experiments about seismic performance of reinforced earth retaining wall.
In: Proceedings of the 11th World Conference on Earthquake Engineering,
200 mm Paper no. 1083.
Hyodo, M., Yamamoto, Y., Sugiyama, M., 1994. Undrained cyclic shear
Strip behaviour of normally consolidated clay subjected to initial static shear
stress. Soils Found. 34 (4), 111.
Slipping surface Koseki, J., Bathrust, R.J., Gler, E., Kuwano, J., and Maugeri, M., 2006.
of shaking test Seismic stability of reinforced soil walls, Keynote Lecture. In: Proceedings
of the 8th International Conference on Geosynthetics. Yokohama, vol. 1,
pp. 5177.
Koseki, J., Tateyama, M., Watanabe, K., and Nakajima S., 2007. Stability of
earth structures against high seismic loads. In: Proceedings of the 13th
Fig. 17. Comparison of failure lines after pullout and shaking table tests. Asian Regional Conference on Soil Mechanics and Geotechnical Engineer-
ing, Post-Conference Volume. pp. 222241.
Ling, H.I., Leshchinsky, D., Perry, E.B., 1997. Seismic design and perfor-
2) During the shaking test on untreated Chiba sand at an over- mance of geosynthetic-reinforced soil structures. Gotechnique 47 (5),
burden pressure of 0 kPa, when the initial pullout load was set to 933952.
Matsuo, O., Tsutsumi, T., Yokoyama, K., Saito, Y., 1998. Shaking table tests
75% of the maximum level and an acceleration of 1000 Gal was
and analyses of geosynthetic-reinforced soil retaining walls. Geosynth. Int.
applied, the strip slipped within 10 s and the displacement of the 5 (12), 97126.
facing wall increased rapidly to more than 10 mm. Miyata, T., Fukuda, N., Kojima, K., Konami, T. and Otani, Y., 2001. Design
3) In the shaking table test on untreated Chiba sand at an of reinforced soil wall: overview of design manuals in Japan. Landmarks in
overburden pressure of more than 50 kPa, the displacement Earth Reinforcement. In: Proceedings of the International Symposium on
Earth Reinforcement. vol. 2, pp. 11071114.
of the wall was 0.5 mm in front, which was 0.2% of the
Miyata, Y., Bathurst, R.J., 2012. Measured and predicted loads in steel strip
height of the facing wall, and the rate of change in the reinforced c - soil walls in Japan. Soils Found. 52 (1), 117.
pullout load ratio remained at approximately 100%. In this Public Works Research Center (PWRC), 2003. Design Method, Construction
test case, the strip did not slip. Manual and Specications for Steel Strip Reinforced Retaining Walls, third
4) In contrast to when sand was used as a backll material, the ed. Public Works Research Center, Tsukuba, Ibaraki, Japan (in Japanese).
Richardson, G.N., Lee, K.L., 1975. Seismic design of reinforced earth walls. J.
rate of change in the pullout load decreased to 0% when
Geotech. Eng. Div. ASCE 101 (2), 167188.
clay was used, even when an overburden pressure was Richardson, G.N., Feger, D., Fong, A., Lee, K.L., 1977. Seismic testing of
applied. However, the cementation of the backll material reinforced earth walls. J. Geotech. Eng. Div. ASCE 103 (1), 117.
improved the pullout behaviour of the soil that had been Sahara, K., 2012. Investigation of damage to reinforced earth walls in the great
deemed unsuitable by the regulations. east Japan earthquake and related maintenance. In: Proceedings of the
International Joint Symposium on Urban Geotechnics for Sustainable
5) Comparing the seismic behaviours when the initial pullout
Development. JS-Seoul, pp. 7477.
load was set at 75% and 85% of the maximum pullout load, Sato, M., Onodera, S., and Naemura, S., 2006. A proposal of damage grade
we found that the behaviours were nearly the same. estimation for reinforced retaining wall. In: Proceedings of the 41st Japan
Therefore, we can conclude that the value of the initial National Conference on Geotechnical Engineering. pp. 18471848 (in
pullout load does not signicantly affect the seismic Japanese).
Suzuki, M., Tasaka, Y., Yoneda, O., Kubota, A., and Yamamoto, T., 2007.
performance within the tested range.
Pullout resistance of strip embedded in cement-treated soil layer for
6) When the reinforced soil wall model, mounted on the reinforced soil walls. In: Proceedings of the 5th International Symposium
shaking table, was vibrating at an acceleration of 1500 Gal, on Earth Reinforcement. IS-KYUSHU07. pp. 693699.
the rate of change in the pullout load decreased by no more Tasaka, Y., Suzuki, M., Yoneda, O., Shimura, N., Sugiyama, Y., 2010. Pullout
than 5% of the initial value. The resistance between the resistance characteristics of strips from cement-treated soil and design
method of strip in reinforced earth wall. J. JSCE Geotech. Eng.(C) 66 (3),
strip and the backll material can be inuenced by the
516529 (in Japanese).
acceleration. However, it is possible that the above results The Japanese Geotechnical Society, 2006. New application of earth
could include the inuence, if any, of those boundary reinforcement-combined geotechnical technology (in Japanese).
conditions peculiar to laboratory testing. Uezawa, H., Yasuda, Y., Menjo, S., 1974. Experiment and analysis of the
failure of model reinforced earth (Terre Arme) Railway Technical
Research Report. 889. The Railway Technical Research Institute, Japanese
National Railways. 116 (in Japanese).
Watanabe, K., Munaf, Y., Koseki, J., Tateyama, M., Kojima, K., 2003.
Acknowledgements Behaviors of several types of model retaining walls subjected to irregular
excitation. Soils Found. 43 (5), 1327.
The authors are grateful to the late Professor Tetsuro Yogendrakumar, M., Bathurst, R.J., Finn, W.D.L., 1992. Dynamic response
analysis of reinforced-soil retaining wall. J. Geotech. Eng. ASCE 118,
Yamamoto for providing helpful advice and to Mr. Masanori
11581167.
Fujii and Mr. Noboru Sato for assisting in the setting up of the
experimental facilities.

You might also like