Soils and Foundations 2014;54(6):1175–1187

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

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

Failure behavior and mechanism of slopes reinforced using soil nail wall
under various loading conditions
Ga Zhangn,1, Jie Cao, Liping Wang
State Key Laboratory of Hydroscience and Engineering, Tsinghua University, Beijing, PR China
Received 26 September 2013; received in revised form 28 July 2014; accepted 7 September 2014
Available online 11 December 2014

Abstract

Soil–nailing technology is widely applied in practice for reinforcing slopes. A series of centrifuge model tests was conducted on slopes
reinforced with a soil nail wall under three types of loading conditions. The behavior and mechanism of failure process of the reinforced slopes
were studied using image-based observation and displacement measurements for the slope, nails, and cement layer. The nailing significantly
increased the stability level and restricted the tension cracks of the slopes. Increasing the nail length improved the stability of the reinforced slopes
with deeper slip surfaces. The reinforced slope exhibited a significant failure process, in which slope slippage failure and cement layer fracture
occurred in conjunction with a coupling effect. The deformation localization was induced by the loading within the slope and ultimately
developed into a slip surface. The nailing reinforced the slope by significantly delaying the occurrence of the deformation localization within the
slope. The failure of nails was recognized as a combination of pull-out failure and bend deformation. The loading conditions were shown to have
a significant effect on slope deformation and nail deflection, and they consequently influenced the failure behavior and its formation sequence.
& 2014 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.

Keywords: Slope; Reinforced soils; Soil nailing; Failure; Centrifuge model test

1. Introduction A diverse range of methods have been proposed for analyzing

the stability of nail-reinforced slopes, including the limit equili-
As soil–nailing technology is increasingly applied to reinforce brium method, the finite-element method, and the kinematics
slopes, the number of studies on the design of nail-reinforced method. Based on the different hypotheses for slope failure
slopes has also increased. The effective designing method surface and/or the soil–nail interaction model, soil nail behavior
depends on the robust evaluation of the stability level of the was analyzed and the design was optimized with respect to
reinforced slopes, which should be based on a sound under- various parameters, including length, spacing, inclination, dia-
standing of failure behavior and reinforcement mechanisms. meter and geometric arrangements (e.g., Shen et al., 1981; Juran
et al., 1990; Kim et al., 1997; Yang and Drumm, 2000; Guler and
n
Correspondence to: Institute of Geotechnical Engineering, Department of Bozkurt, 2004; Cheuk et al., 2005; Patra and Basudhar, 2005;
Hydraulic Engineering, Tsinghua University, Beijing 100084, PR China. Gui and Ng, 2006). A non-linear finite element program,
Tel./fax: þ86 10 62795679. PLAXIS, used to analyze the stability level of nail-reinforced
E-mail addresses: zhangga@tsinghua.edu.cn (G. Zhang), slopes has revealed that the optimal nail orientation decreases as
caojie.06@gmail.com (J. Cao), wlp04@mails.tsinghua.edu.cn (L. Wang).
1
Institute of Geotechnical Engineering, Department of Hydraulic Engineering,
the gradient of the slope increases (Fan and Luo, 2008).
Tsinghua University, Beijing 100084, PR China. The reinforcement effect and mechanism for soil nailing in
Peer review under responsibility of The Japanese Geotechnical Society. slopes was investigated via observations of the strain-stress

http://dx.doi.org/10.1016/j.sandf.2014.11.011
0038-0806/& 2014 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. All rights reserved.
1176 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

performance of nails, the safety level of soil nail wall, and the Table 1
soil–nail interaction (e.g., Wong et al., 1997; Junaideen et al., List of centrifuge model tests.
2004; Zhou and Yin, 2008; Zhou et al., 2009). The strength Loading condition Gradient (V:H) Nail length (cm) No.
reduction method and the limit equilibrium method were
utilized to analyze the soil–nail interaction in reinforced slopes Self-weight loading 5:1 8 G5-R8-S
and showed that failure modes influence the line of maximum 4 G5-R4-S
– G5-U-S
tension (Wei and Cheng, 2010).
3:1 8 G3-R8-S
Field observation has played an important role in under- 6 G3-R6-S
standing the behavior of slopes reinforced with soil nailing 4 G3-R4-S
(e.g., Andrzej et al., 1988; Nowatzki and Samtani, 2004). Excavation 3:1 8 G3-R8-E
Another study analyzed the response of a full-scale nail- 2:1 8 G2-R8-E
6 G2-R6-E
reinforced slope using observed and measured data (Turner 4 G2-R4-E
and Jensen, 2005). Centrifuge model tests have been effec- 1.2:1 8 G1.2-R4-E
tively used to explore the deformation and failure behavior of Vertical loading 3:1 8 G3-R8-L
nail-reinforced slopes under various loading conditions 6 G3-R6-L
through producing an equivalent gravity-induced stress field 2:1 6 G2-R6-L
between the model and prototype (e.g., Zornberg et al., 1997;
Zhang et al., 2001; Wang et al., 2010). The aseismic nail-
reinforcement mechanism was demonstrated by comparing the surface loading
18
deformation between reinforced and unreinforced slopes via 5 loading plate
centrifuge model tests under earthquake conditions (Wang
et al., 2010) and 1g shaking table tests (Tatsuoka et al., 2012).
Zhang et al., (2013) employed centrifuge model tests to
analyze the deformation behavior and failure process of a u
cement layer
nail-reinforced slope under the surface loading condition, and
they illustrated the influence rules of different nail layouts.
33

excavation face
The objective of this paper is to investigate the failure y
v
mechanism of slopes reinforced with soil nail wall using nail

h
serialized centrifuge model tests. The steps were as follows: (1) x
the shotcrete wall and different loading conditions were o

5
simulated to systematically understand the actual response of R
the reinforced slopes; (2) the failure modes were clarified 50
according to the test observations; (3) the slope failure process,
together with the fracture and deformation of the reinforcement Fig. 1. Schematic views of test model under different loading conditions
applied to the slope in various separate tests (unit: cm).
structures, was captured according to the test observations and
measurement results: and (4) the failure and reinforcement
mechanisms were elucidated through integrated analysis of the for comparison with the reinforced slopes, to analyze the
deformation and failure processes, which was accomplished by reinforced mechanism.
accurately measuring the displacement of entire slopes.

2. Tests 2.2. Model preparation

2.1. Schemes The model slope was placed in an aluminum alloy model
container that was 50 cm in length, 20 cm in width, and 35 cm
All centrifuge model tests were conducted using a 50g-ton in height. All the slope models were 28 cm in height with a
geotechnical centrifuge at Tsinghua University, which has an 5-cm-deep horizontal ground layer. The container sides in
effective radius of 2 m and a maximum centrifugal acceleration contact with the slope were coated with silicone. This measure,
of 250g. The deformation and failure processes of nail- together with the ground layer, was used to diminish the
reinforced cohesive soil slopes were observed under different influence of the model container on the slope response.
loading conditions, including self-weight loading, vertical A cohesive soil with a specific gravity of 2.7 was retrieved
loading on the top of the slope, and excavation at the toe of from the soil base of a high-rise building in Beijing for use in
the slope (Table 1, Fig. 1). the slope model in the tests. The liquid limit and plastic limit of
The slope gradient and nail length were varied to investigate the soil were 33.5% and 15.5%, respectively; indicating that
their influence on the response of the reinforced slopes. The the soil is a type of clay. The average particle size (d50) of the
gradients of slope for the tests ranged from 1.2:1 to 5:1 clay was 0.01 mm, and particle sizes d10 and d60 of the clay
(Horizontal:Vertical). Three nail lengths from 4 cm to 8 cm were 0.001 mm and 0.02 mm, respectively. The maximum
were used in the tests. Unreinforced slopes were also simulated density of the clay was about 1.8 g/cm3. The soil was
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1177

compacted to the design dry density of 1.51 g/cm3, with a 2.3. Test process
water content of 17% (i.e., saturation degree: 75%), by
horizontal layer with thickness of 5–6 cm in the model For the self-weight loading tests, the centrifugal acceleration
container. Excess soil was then removed to form the slope was applied on the model by a step-wise increase of 5g. The
model. The shear strength of the soil were determined using centrifugal acceleration was maintained for a few minutes at
the consolidated-drained triaxial tests and can be described each stage to assure the deformation of the slope resulting from
using the Mohr–Coulomb criterion, with 241 as the internal the self-weight loading becoming stable. The centrifugal
frictional angle and 26 kPa as the cohesion as the effective acceleration increment decreased to 1–2g when a slope failure
stress parameters. was predicted to occur. The loading was ceased upon observed
The nail was simulated using a thin steel needle with a slope failure.
diameter of 1 mm in the tests. The tensile strength and the For the vertical loading and excavation tests, the centrifugal
elastic modulus of the steel were 200 MPa and 210 GPa, acceleration was gradually increased to 50g. The loading plate
respectively. For simplicity in model preparation, the nails and excavation blade did not contact the slope while increasing
were inserted orthogonally with the slope surface into the slope the centrifuge acceleration. Vertical loading or excavation was
with a uniform distribution spacing of 2 cm. It can be conducted to the slope after the slope settlement became stable
concluded that the inclination of nails would have a small at 50g. Vertical loading was applied to the top of the slope at a
effect on the fundamental rules obtained in this paper. Note rate of 1 mm/s via a loading plate with a width of 5 cm using a
that the nails were arranged only in the upper part of the slope vertical loading device (Zhang et al., 2013) (Fig. 1). The
in the excavation tests because the excavation was conducted loading device consists of an electric motor and corresponding
on the lower part of the slope (Fig. 1). reducer, which applies a vertical load with capacity of 10 kN
A new technique was developed to simulate shotcrete and on the loading plate through the shaft. Excavation was
the rock bolt support of the soil nail wall using a cement layer conducted near the toe of the slope using an excavation
during centrifuge model tests (Fig. 2). After preparing the simulator (Li et al., 2011) (Fig. 1). The simulator realizes the
slope model, a type of special-made Portland cement was excavation by using the steel blade, which is pushed vertically
smeared on the slope surface with thickness of 2 mm at 1g- down into the slope and then cut soil away from the slope.
level (earth gravity). At this moment, the cement layer was
very soft; thus the nails were then easily inserted into the slope 2.4. Measurement
and remained an outcrop on the slope surface with several
centimeters (Fig. 2). The nails were kept vertical with the slope The vertical load was measured using a load transducer
surface with aid of a drawing square. The cement reaches a installed on the loading plate during the vertical loading tests,
compression strength of 15 MPa from the fully flexible state and the loading pressure was obtained as the load divided by
within two hours, which satisfies the simulation time require- the area of the loading plate. Settlement of the loading plate
ment for soil nailing during centrifuge model tests. In other was measured using a laser displacement transducer.
words, the cement layer reached the required strength within The images of lateral side of the slope were recorded using
two hours while maintaining a centrifugal acceleration of 50g. an image-recording and displacement measurement system
The friction behavior of the nail–soil interface has been during the tests (Zhang et al., 2009). The displacement history
investigated using a series of nail pull-out tests under diffe- of an arbitrary point on the lateral side of the slope, where a
rent constant overburden pressures (Zhang et al., 2013). The number of white granite particles were embedded to obtain a
Mohr–Coulomb criterion was found to describe the relation- gray-variance region required by the measurement, was
ship between the friction stress and overburden pressure, with determined based on an image-correlation algorithm using
16 kPa in the cohesion and 231 in the frictional angle as the the captured image series (Zhang et al., 2009). In this paper,
effective stress parameters. the measurement accuracy of displacement reached 0.03 mm at

Fig. 2. Precasting of soil nail wall before tests. (a) Cement layer on the slope; (b) nails.
1178 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

the model dimension. The slip surface of the slope, which

appeared near the failure of the slope were directly by hand
determined according to the captured images because it could
be easily distinguished on the image. The Cartesian coordinate
system was established with the center of the slope toe as the
origin. Positive was defined as to the right in the horizontal
direction (x-axis) and upward in the vertical direction (y-axis)
(Fig. 1).
All measured results were presented at the model dimension
in this paper. The prototype dimension of the length and
measured displacements can be obtained by multiplying the
model dimension of those with the centrifugal acceleration.
The prototype dimension of the loading pressure was equal to
the model dimensions.

3. Failure mode and limit bearing capacity

Fig. 3a shows an image of a slope at failure state under the

self-weight loading condition. It can be seen that an evident
slip surface appeared in the slope and that significant fracture
occurred in the lower part of cement layer. The locations of the
slip surface and the cement layer’s fracture were easily
distinguished on the image (Fig. 3a). Thus, they were
determined directly by hand according to the image and drawn
in Fig. 3b by outlining the nail-reinforced zone. The slip
surface passed by the nails in the upper part of the slope but
passed through the nails in the lower part of the slope. Similar
phenomena were confirmed by observations from other tests.
Fig. 4 compares the slip surfaces of the slopes with different
nail lengths and different gradients under the self-weight
loading condition. The slip surface was shown to depend
significantly on the nail length (Fig. 4a). For example, the slip
surface moved deeply into the slope if the nail length was
increased from 4 cm to 8 cm. The slip surface of the reinforced
slope with a nail length of 4 cm was fairly close to that of the
unreinforced slope in the middle and lower parts of the slope.
This result demonstrated that the nails did not significantly
affect the failure features of the slope when the nail length was
relatively small. The nail length should be much longer than
the maximum depth of the critical slip surface of the
unreinforced slope for a better reinforcement effect. In other
Fig. 3. Failure features of the 5:1 slope reinforced with 8-cm-long nails under
words, the slip surface of an unreinforced slope can be used to self-weight loading conditions (G5-R8-S). (a) Image; (b) schematic view.
direct the nail design. The slip surfaces of reinforced slopes Numbers in square brackets show centrifugal acceleration corresponding to
became deeper as the gradient of the slope decreased, e.g., fraction height of cement layer.
from 5:1 to 3:1 (Fig. 4b).
Fig. 5 shows the slip surfaces of the 3:1 slope reinforced
with 6-cm-long nails under the vertical loading condition. Close comparison showed that the slip surface depended
Vertical loading caused two slip surfaces that differed sig- significantly on the loading conditions (Figs. 4–6). For example,
nificantly from the slip surface induced by the self-weight self-weight loading, which applied the volumetric force on the
loading outlined in Fig. 4. The slip surfaces nearly passed slope, induced the deepest slip surface compared with the other
through the inner edge of the loading plate and the toe of the two loading conditions that apply loads on the slope surface.
slope. This result demonstrated that the slip surface was To ensure a unified description of different loading types,
controlled by the feature of the vertical load, as was confirmed the limit bearing capacity of a slope was defined as the load
in previous centrifuge model tests (Zhang et al., 2013). at which the slope failed, meaning that the slip surface of
Fig. 6 shows the slip surfaces of the 3:1 slope reinforced with the slope fully appeared as determined based on the
8-cm-long nails under the excavation condition. It can be seen captured images. For the self-weight loading condition, the
that the slip surface also passed through the nail-reinforced zone. limit bearing capacity was defined based on the ultimate
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1179

Fig. 5. Slip surface of a 3:1 slope under vertical loading conditions (G3-R6-L).

30
y (cm)

25

boundary of nail-
20
reinforced zone

15

10
slip surface
h=7.5cm
5

x (cm)
0
-25 -20 -15 -10 -5 0
Fig. 4. Slip surfaces of slope in different self-weight loading tests. (a) varying
nail lengths for the 5:1 slope; (b) varying gradients for the slope reinforced Fig. 6. Slip surface of a 3:1 slope under excavation conditions (G3-R8-E).
with 4-cm-long nails. h, excavation height.

centrifugal acceleration at which the slope failed; the ultimate indicating that this is the benchmark depth that can be used to
centrifugal accelerations in different tests are summarized in distinguish the reinforcement effect. Thus, it can be concluded
Fig. 7. It can be inferred that as the nail length increased, small that the nail reinforcement length should largely exceed the
increases were observed in the limit bearing capacity for nails maximum depth of the slip surface of the unreinforced slope.
shorter than 4 cm in length, while more significant increases The bearing capacity of the slope failure (Pu) under the
were observed when the nail length was greater than 4 cm. vertical loading condition was described using the ultimate
This was consistent with the influence rules of the nail length loading pressure at slope failure, as shown in Fig. 8. The
on the slip surface of the slope. The maximum depth of the slip ultimate loading pressure was determined to be approximately
surface of the unreinforced slope was approximately 3 cm, equal to the stable value, or peak value, of the loading pressure
1180 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

75 125
ultimate centrifugal acceleration (g)

70
gradient of slope 100
65
3:1
60
5:1 75

Pu (kPa)
55

50 50
45 gradient of slope
25 2:1
40
3:1
35
0 2 4 6 8 10 0
nail length (cm) 4 6 8 10
nail length (cm)
Fig. 7. Ultimate centrifugal accelerations for the slopes under self-weight
loading conditions. Fig. 8. Ultimate loading pressures of the slopes under vertical loading
conditions. Pu, ultimate loading pressure.

in the load-settlement curves of the loading plate. It can be

seen that the limit bearing capacity of the slope decreased as 15
the nail length decreased and the slope gradient increased gradient of slope
under the vertical loading condition. Under the excavation 2:1
condition, the limit bearing capacity of the slope was described 3:1
using the excavation height (hu) at slope failure (Fig. 9). The 10

limit bearing capacity of the slope was shown to increase with

hu (cm)

increasing nail length and with decreasing slope gradient,

which was similar to other loading conditions.
5

4. Failure process

Close examination of captured images of reinforced slopes 0

4 6 8 10
revealed that the slope exhibited significant failure process nail length(cm)
during loading. The full failure process of slopes reinforced
Fig. 9. Ultimate excavation heights for slopes under excavation conditions.
using soil nailing was examined through observations in the
hu, ultimate excavation height.
centrifuge model tests, which included fracture of the cement
layer and formation of the slip surface.
A typical fracture process of the cement layer of a reinforced acceleration. For every point couple, the tangential relative
slope was analyzed based on the images captured throughout displacement was larger than the normal relative displacement
the tests (Fig. 10). First, a few tiny cracks appeared in the at a given centrifugal acceleration, demonstrating that the
cement layer near the toe of the slope at the 35g-level slippage failure occurred along the slip surface. In all cases, the
(Fig. 10a). The cracks developed with increasing centrifugal tangential relative displacement increased slowly with increas-
acceleration and proliferated horizontally at the 45g-level ing centrifugal acceleration, followed by a remarkable inflec-
(Fig. 10b). The local failure expanded upwards in the cement tion in each of the history curves, after which the tangential
layer as the centrifugal acceleration increased and a small area relative displacement increased significantly as the centrifugal
of the slope was crushed at the 47.6g-level (Fig. 10c). Finally, acceleration increased. Such an inflection can be regarded as
the local failure of the cement layer occurred within an area of an indicator of the appearance of slippage failure in the slope.
the lower part of the slope (Fig. 10d). Thus, the images The inflection can be accurately determined as the intersection
revealed that local failure of the cement layer appeared and of tangents to the history curve near the inflection, as indicated
progressively developed with the application of loading. by the dashed line in Fig. 12. This phenomenon allowed clear
The appearance and development of slip surface of the slope determination of the centrifugal accelerations at which the
was analyzed on the basis of the relative displacement of point local slippage appeared along the slip surface.
couples on opposite sides of the slip surface, selected with a Fig. 11 identifies the centrifugal accelerations at which local
spacing of 10 mm. Fig. 11 shows the point couples for analysis failures occurred to describe the development process of the
of a typical reinforced slope. Their relative displacements slip surface under the self-weight loading condition. The local
under self-weight loading conditions, tangential and normal to failure first appeared near the toe of the slope and it gradually
the slip surface, were obtained using image-based measure- developed upwards until finally reaching the top of the slope.
ments (Fig. 12). The relative displacements of all the point This result demonstrated that the nail-reinforced slope under-
couples were found to increase with increasing centrifugal went a significant failure process. The area of the cement layer
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1181

Fig. 10. Fracture process of the cement layer of 5:1 slope with 8-cm-long nails under self-weight loading conditions (G5-R8-S). (a) 35g; (b) 45g; (c) 47.6g;
(d) 48.3g.

tangential relative displacement at the point couples. However,

local slippage failure occurred at this location when reinforce-
ment nails of adequate length were used (e.g., 8 cm). This result
demonstrated that use of the nailing changed the failure
properties of the slope, which restricted the tension crack in
the slope.
Fig. 14 shows the failure sequence of a reinforced slope
under vertical loading conditions that was investigated via
point couple analysis. The failure sequence was shown to be a
relatively more complex process compared with the self-
weight loading conditions. The vertical load first induced local
slippage near the toe of the slope and at the inner edge of the
loading plate. These local slippages extended to the interior of
the slope and eventually developed into a full slip surface.
Similar failure sequences under vertical loading conditions
have been demonstrated in centrifuge model tests on nail-
reinforced slopes without cement layers (Zhang et al., 2013).
Fig. 15 shows the failure processes of different reinforced
slopes under excavation conditions. The slip surface first
Fig. 11. Failure process of 5:1 slope with 8-cm-long nails under self-weight
loading conditions (G5-R8-S). Arrow, direction of the failure sequence;
occurred near the toe of the slope and then developed upwards
numbers in round bracket and square bracket, centrifugal accelerations that to the top of the slope. Whereas the nail length did not affect
indicate the appearance of local failure and fraction zone height of the cement this pattern of slip surface development, it was shown to
layer, respectively. influence the local failure properties of the slope, similar to the
self-weight loading condition results (Fig. 13). For example,
local tension failure appeared near the top of slope with a nail
fracture is also outlined at different centrifugal accelerations in length of 4 cm, but this changed to slippage failure at nail
Fig. 11. The cement layer was observed to exhibit a fracture lengths of 6 cm and 8 cm.
prior to the occurrence of the slip surface, demonstrating that In summary, the reinforced slopes underwent significant
the cement layer fracture, which reduced the restriction of the failure processes that were influenced by both the loading
cement layer on the slope, was an important prerequisite for conditions and reinforcement arrangements. Stability analysis
slippage failure of the slope. The slippage failure of the slope methods should capture the features of such a failure process.
caused a significant movement of the slippage body, in turn
aggravating the cement layer fracture. Therefore, the slippage 5. Failure process mechanism
failure of the slope and fracture of the cement layer developed
with a coupling effect until the final landslide occurred. 5.1. Deformation-based analysis
Comparison of the local failure sequences of slopes indicated
that the slip surface developed gradually from the toe of the The failure process mechanism of the reinforced slope was
slope upwards to the top of the slope for all the tests under self- analyzed on the basis of the deformation behaviors observed via
weight loading conditions (Fig. 13). This result demonstrated image-based measurements. Fig. 16 shows a series of horizontal
that the failure sequence was not influenced by the length of the displacement contours of a reinforced slope at different loads
nail or the gradient of the slope, but instead depended primarily under self-weight loading conditions. Significant horizontal dis-
on the loading condition. For the unreinforced and 4-cm-long placement was observed at the 35g-level (Fig. 16a). The contour
nail-reinforced slopes with a gradient of 5:1, local tension failure lines of horizontal displacement exhibited a tendency of concen-
occurred near the top of the slope, which was determined tration, demonstrating significant deformation localization near
because the normal relative displacement was larger than the the toe of the slope as the centrifugal acceleration increased to the
1182 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

2.5 45g-level (Fig. 16c). This localized deformation caused the

relative displacement (mm)

2 point couple a nearby cement layer to crack because the cement was unable to
bear such a significant non-uniform deformation (Fig. 10b). The
1.5
extent of deformation localization near the toe of the slope
1
0.5
0
-0.5
2
relative displacement (mm)

1.5 point couple b

0.5

-0.5
2.5
relative displacement (mm)

2 point couple c
1.5
1
0.5
0
-0.5
3
relative displacement (mm)

2.5 point couple d

2 Fig. 13. Failure processes of slopes in different self-weight loading tests.
1.5 arrow, direction of the failure sequence.
1
0.5
0
-0.5
2
relative displacement (mm)

1.5 point couple e

0.5

-0.5
1
relative displacement (mm)

point couple f
0.5

0
ds
dn
-0.5
45 46 47 48 49 50
centrifugal acceleration (g)

Fig. 12. Histories of relative displacement of point couples of a 5:1 slope with Fig. 14. Failure process of 3:1 slope with 6-cm-long nails under vertical
8-cm-long nails under self-weight loading conditions at model dimension (G5- loading conditions (G3-R6-L). Arrow, direction of failure sequence; number in
R8-S). ds and dn, relative displacements tangential and perpendicular to the bracket, settlement (unit: mm) of loading plate corresponding to the appearance
slip surface, respectively. of slippage.
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1183

increased following the rupture of the cement layer, and the zone Fig. 17a shows the horizontal distribution of the horizontal
with the localized deformation developed upwards to the top of displacement of a reinforced slope at several elevations at
the slope with increasing centrifugal acceleration (Fig. 16d–g). different g-levels. The horizontal displacement increased mono-
The development of the deformation localization zone was shown tonically from the interior to the surface of the slope. Before the
to be similar to the failure sequence of the reinforced slope. This centrifugal acceleration approached the range at which the
implied that the deformation localization within the slope was the failure of the slope occurred, a segment can be observed with
primary driver of the slope failure process and can therefore be significant gradient along the horizontal distribution curve of the
used to analyze the failure mechanism of the slope. horizontal displacement, indicating significant deformation
localization there. The gradient of the horizontal displacement,
defined as the ratio of the increment of displacement to the
increment of x-coordinate, was used to quantify the extent of
the deformation localization (Fig. 17b). Fig. 17b shows that the
gradient of horizontal displacement was fairly small across the
entire slope at the 40g-level and increased significantly with an
evident peak value as centrifugal acceleration increased to the
45g-level. At that point, the gradient of horizontal displacement
around the location of the peak was significantly larger than in
other locations, which demonstrates that a deformation localiza-
tion zone was induced in the slope by the self-weight loading.
After the deformation localization zone appeared, the gradient of
horizontal displacement in the deformation localization zone
increased at a significantly more rapid rate than in other areas
during the loading process. The slip surface that ultimately
appeared in the zone is outlined with dashed lines in Fig. 17b, it
was denoted g-level based on the analytical results of the failure
process shown in Fig. 11. The slip surface was found to appear
after, and is therefore inferred to result from, the occurrence of a
localized deformation zone. Close examination showed that the
Fig. 15. Failure sequence of 2:1 slopes in excavation tests. h, excavation deformation localization zone moved from the interior to the
height; arrow, direction of the failure sequence. surface of the slope with decreasing elevation from the top of

Fig. 16. Horizontal displacement contours of the 5:1 slope with 8-cm-long nails in self-weight loading test at model dimension (G5-R8-S) (unit: mm). (a) 35g;
(b) 40g; (c) 45g; (d) 45.7g; (e) 46.3g; (f) 47g; (g) 47.7g.
1184 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

Fig. 17. Horizontal distributions of horizontal displacement of 5:1 slope with 8-cm-long nails in self-weight loading test at model dimension (G5-R8-S).
(a) Horizontal displacement; (b) gradient of horizontal displacement. u, horizontal displacement; Δu/Δx, gradient of horizontal displacement; dashed line, location
of slip surface; number in round bracket, centrifugal accelerations that indicate the appearance of local failure.

the slope, which was consistent with the shape of the final slip of horizontal displacements of the unreinforced slope showed
surface. Therefore, it can be concluded that the deformation similar tendency as the reinforced slopes, although the
localization was induced by the loading and ultimately resulted deformation localization zone appeared earlier (Fig. 18). This
in the development of the slip surface. The slip surface can be indicated that both the reinforced and unreinforced slopes are
estimated using the deformation localization zone prior to slope driven by the same failure mechanism. The use of a soil nailing
failure. significantly decreased the slope deformation and subsequently
delayed the occurrence of the deformation localization within
5.2. Reinforcement mechanism the slope, therefore arresting the failure of the slope. This
conclusion was confirmed by the observations in other groups
The nail reinforcement mechanism can be explained via the of centrifuge model tests with varied slope gradients and
deformation-based analysis. The distributions of the gradient loading conditions.
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1185

reinforced with soil nailing should consider not only nail pull-
out but also nail bending deformation as well as their
combination. In other words, both types of nail failure
mechanisms, including pull-out failure and bending deforma-
tion, and their combination need to be analyzed to evaluate
their effect on stability of a reinforced slope.
The nail deflections, which described the bending deforma-
tion and reflect the complicated nail–soil interaction, can be
obtained directly from the corresponding displacement of soil
that was measured using the image-based analysis. The
validity of this method has been preliminarily verified in
previous studies (Zhang et al., 2013).
The distributions of nail deflections at different locations were
determined and described along the nail using a local w–X
coordinate system under different loading conditions (Fig. 19).
In the local w–X coordinate system, the X-axis and w-axis were
defined as parallel to the nail and the slope surface, respectively;
this is depicted in Fig. 19a along with the positive direction of
the coordinate axes and the location of measurement nails. Note
that the deflection of nails was only presented in upper part of
slope for the excavation tests because the nails were limited to
those locations (Fig. 19b). The deflections of nails increased
during loading, reflecting an increase in nail–soil interaction due
to application of loading. The distribution of nail deflections
exhibited different characteristics in different locations
(Fig. 19b–d). For example, under self-weight loading condi-
tions, the deflection of the nail in upper part of the slope
increased from the interior of the slope, whereas it was nearly a
parabola with a peak value near the slope surface in the lower
part of the slope (Fig. 19c). On the other hand, the loading
conditions had a significant effect on the distribution of nail
deflection. The vertical loading induced a nearly contrary rule in
the distribution of deflection compared with the self-weight
loading condition. A parabolic distribution curve was observed
in the upper part of the slope, with a monotonic increase in the
lower part of the slope (Fig. 19d). The distribution of nail
deflection was similar for both the vertical loading and excava-
tion conditions (Fig. 19b and d).

6. Conclusions

A series of centrifuge model tests was conducted on the

Fig. 18. Horizontal distributions of gradient of horizontal displacement of 5:1 slopes reinforced with soil nail wall under three types of
unreinforced slope in self-weight loading test at model dimension (G5-U-S). loading conditions: a self-weight loading applied with a
Δu/Δx, gradient of horizontal displacement; dashed line, location of slip volumetric force, and a vertical loading and excavation applied
surface; number in round bracket, centrifugal accelerations that indicate the to different types of surface loads on the slope. Image-based
appearance of local failure.
observation and measurements of the slope, nails, and cement
layer were used to study the behavior and mechanism of failure
5.3. Nail responses process of reinforced slopes. The main conclusions drawn are
as follows:
The nails were examined after the tests. The results showed
that the nails did not break due to tension and exhibited evident (1) The soil nailing significantly increased the stability level
flexural deformation near the slip surface. Pull-out deformation and restricted the tension cracks of the slopes. Increasing
of nails was inferred but could not be measured accurately. the nail length improved the stability of reinforced slopes
Note that, despite experiencing significant bending, nails with deeper slip surfaces.
still passed through the slip surface over a large area in the (2) The reinforced slope exhibited a significant failure process
slope. This result indicated that the failure analysis of slopes during loading. The slippage failure of the slope and
1186 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187

Fig. 19. Distributions of nail deflections of reinforced slopes at model dimension in different tests. (a) Locations of measurement nails; (b) 2:1 slope with 6-cm-long
nails under excavation conditions (G2-R6-E); (c) 5:1 slope with 8-cm-long nails under self-weight loading conditions (G5-R8-S); (d) 3:1 slope with 6-cm-long nails
under vertical loading conditions (G3-R6-L). w, nail deflection; X, distance from slope surface.

fracture of the cement layer developed in turn with a the deformation localization zone: the soil nailing significantly
coupling effect until final landslide occurred. reduced the deformation of the slope, and accordingly delayed
(3) The deformation localization was induced by loading within the the occurrence of the deformation localization.
slope and ultimately developed into a slip surface. Slope failure (4) The nails exhibited increasing deflections with different
features, such as the slip surface, can therefore be estimated distributions in different locations of slope during loading.
according to the deformation localization zone. Similarly, the Nail failure was recognized as a combination of pull-out
reinforcement mechanism of the nailing can be described using failure and bend deformation.
 G. Zhang et al. / Soils and Foundations 54 (2014) 1175–1187 1187

(5) Slope deformation and nail deflection were significantly Junaideen, S.M., Tham, L.G., Law, K.T., Lee, C.F., Yue, Z.Q., 2004.
dependent on the loading conditions. Consequently, the Laboratory study of soil–nail interaction in loose, completely decomposed
granite. Can. Geotech. J. 41, 274–286.
loading condition had a large effect on the slip surface with
Juran, I., Baudrand, G., Farrag, K., Elias, V., 1990. Kinematical limit analysis
the formation sequence and on the features of soil–nail of soil–nailed structures. J. Geotech. Div. ASCE 116 (1), 54–73.
interaction. Self-weight loading induced a failure process Kim, J.S., Kim, J.Y., Lee, S.R., 1997. Analysis of soil nailed earth slope by
from the toe to the top of the slope; while vertical loading discrete element method. Comput. Geotech. 20 (1), 1–14.
and excavation caused failure sequences that are opposite Li, M., Zhang, G., Zhang, J.M., Lee, C.F., 2011. Centrifuge model tests on a
to one another and have a relatively shallow slip surface. cohesive soil slope under excavation conditions. Soils Founda. 51 (5),
801–812.
Nowatzki, E., Samtani, N., 2004. Design, construction, and performance of an
18-meter soil nail wall in Tucson, AZ. Geotech. Spec. Publ. ASCE 124,
It should be noted that the centrifuge model tests used only a 741–752.
type of cohesive soil and inadequately considered different Patra, C.R., Basudhar, P.K., 2005. Optimum design of nailed soil slopes.
Geotech. Geol. Eng. 23, 273–296.
physical state of the soil. For example, the saturated slopes Shen, C.K., Bang, S., Hermann, L.R., 1981. Ground movement and analysis of
may exhibit more complicated response near failure due to earth support system. J. Geotech. Div. ASCE 107 (12), 1609–1624.
effect of pore pressure. Therefore, the conclusions should be Tatsuoka, F., Muñoz, H., Kuroda, T., Nishikiori, H., Soma, R., Kiyota, T.,
carefully applied in the real soil nail wall design. Systematic Tateyama, M., Watanabe, K., 2012. Stability of existing bridges improved
investigations which consider more factors are expected to be by structural integration and nailing. Soils Found. 52 (3), 430–448.
Turner, J.P., Jensen, W.G., 2005. Landslide stabilization using soil nail and
conducted in a future study based on the findings of this study. mechanically stabilized earth walls: case study. J. Geotech. Geoenviron.
Eng. 131 (2), 141–150.
Acknowledgements Wang, L.P., Zhang, G., Zhang, J.M., 2010. Nail reinforcement mechanism of
cohesive soil slopes under earthquake conditions. Soils Found. 50 (4),
459–469.
The study is supported by the State Key Laboratory of Wei, W.B., Cheng, Y.M., 2010. Soil nailed slope by strength reduction and
Hydroscience and Engineering (no. 2014-KY-1), National limit equilibrium methods. Comput. Geotech. 37, 602–618.
Program for Support of Top-notch Young Professionals, and Wong, I.H., Low, B.K., Pand, P.Y., 1997. Field performance of soil wall in
National Natural Science Foundation of China (no. 51479096). residual soil. J. Perform. Constr. Facil. 11 (3), 105–112.
Yang, M.Z., Drumm, E.C., 2000. Numerical analysis of the load transfer and
deformation in a soil nailed slope. Geotech. Spec. Publ. ASCE 96,
References 102–116.
Zhang, G., Cao, J., Wang, L.P., 2013. Centrifuge model tests of deformation
Andrzej, S., Danuta, L., Marek, K., 1988. Measured and predicted stresses and and failure of nailing-reinforced slopes under vertical surface loading
bearing capacity of a full scale slope reinforced with nails. Soils Found. 28 condition. Soils Found. 53 (1), 117–129.
(4), 47–56. Zhang, G., Hu, Y., Zhang, J.M., 2009. New image analysis-based displace-
Cheuk, C.Y., Ng, C.W.W., Sun, H.W., 2005. Numerical experiments of soil ment-measurement system for geotechnical centrifuge modeling tests.
nails in loose fill slopes subjected to rainfall infiltration effects. Comput. Measurement 42 (1), 87–96.
Geotech. 32, 290–303. Zhang, J., Pu, J., Zhang, M., Qiu, T., 2001. Model tests by centrifuge of soil
Fan, C.C., Luo, J.H., 2008. Numerical study on the optimum layout of soil– nail reinforcements. J. Test. Eval. 29 (4), 315–328.
nailed slopes. Comput. Geotech. 35, 585–599. Zhou, W.H., Yin, J.H., 2008. A simple mathematical model for soil nail and
Gui, M.W., Ng, C.W.W., 2006. Numerical study of a nailed slope excavation. soil interaction analysis. Comput. Geotech. 35, 479–488.
Geotech. Eng. 37 (1), 1–12. Zhou, Y.D., Cheuk, C.Y., Tham, L.G., 2009. Numerical modeling of soil nails
Guler, E., Bozkurt, C.F., 2004. The effect of upward nail inclination to the in loose fill slope under surcharge loading. Comput. Geotech. 36, 837–850.
stability of soil nailed structures. Geotech. Eng. Transp. Projects, Zornberg, J.G., Mitchell, J.K., Sitar, N., 1997. Testing of reinforced slopes in a
2213–2220. geotechnical centrifuge. Geotech. Test. J. 20, 470–480.