Computers and Geotechnics 37 (2010) 602–618

Contents lists available at ScienceDirect

Computers and Geotechnics

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

Soil nailed slope by strength reduction and limit equilibrium methods

W.B. Wei, Y.M. Cheng *
Department of Civil and Structural Engineering, Hong Kong Polytechnic University, Hong Kong

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

Article history: A detailed study of nailed slopes under different conditions is reported in this paper. No major difference
Received 30 September 2009 is found in terms of safety and slip surfaces between the strength reduction method (SRM) and the limit
Received in revised form 10 February 2010 equilibrium method (LEM) in general cases. Appreciable differences between the SRM and LEM appear,
Accepted 23 March 2010
however, if the nail load is controlled by the overburden stress. Some special slip surfaces from the
SRM obtained by using a very fine mesh are discussed. Field tests demonstrated that the nail head is
important in determining the failure mode and the factor of safety of a nailed slope, while the effect of
Keywords:
the nail elastic modulus is more noticeable only when the slope is very steep. The optimum layout of
Soil nail
Strength reduction method
the soil nail was found to be longer at bottom and shorter at the top, which is contrary to some engineers’
Slope stability guidelines for soil nail design during top-down construction. The distribution of tensional force along the
Three-dimensional soil nail is influenced by the state of the slope (service state, limit state) and the failure modes (external
Slip surface failure, internal failure). In general, the line of maximum tension may not correspond to the critical slip
surface as commonly believed, except for the case where the failure mode is an internal tensile failure.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
Soil nailing is a simple and economic slope stabilization tech-
nique and is particularly useful for the strengthening of existing
slopes. The fundamental concept of soil nailing is to reinforce the
soil with closely spaced passive inclusions to create a coherent
gravity structure and thereby increase the overall shear strength
of the in situ soil, restraining its displacements. The basic design
consists of transferring the resisting tensile forces generated in
the inclusions into the slope through the friction mobilised at the
nail/soil interfaces. There are many different design methods for
soil nailing, including the limit equilibrium method (LEM), several
working stress design methods, and also the Davis method [27],
the German method [33,34] and the French method [25,26]. Based
on the limit equilibrium method, the US Federal Highway Admin-
istration has also published a series of design guidelines (1996,
1997, 1999, 2001, 2003) currently used by many engineers for
nailed soil and reinforced earth structures.
The strength reduction method (SRM) has been used for slope
stability analysis by many researchers [6,7,11,12,15,22,23,32,38,
44]. Detailed studies of the SRM and LEM for 2D and 3D slopes
have been carried out by Cheng et al. [6] and Wei et al. [41], and
interesting and sometimes surprising results have been obtained
in these studies for un-reinforced slopes. Shen et al. [27], Juran
et al. [20,21], Plumelle et al. [24], Thompson and Miller [37], Cho-
ukeir [9], Unterreiner et al. [39], Srinivasa Murthy et al. [35],
* Corresponding author.
E-mail address: ceymchen@polyu.edu.hk (Y.M. Cheng).
Sivakumar Babu et al. [30], and Cheuk et al. [8] have applied
two-dimensional finite element or finite difference methods for
the analysis of nailed slopes, techniques that may be more suitable
for slopes reinforced with geotextile but not with soil nails.
Although two-dimensional analysis provides valuable insight
into the behaviors of nailed slopes, the effect of the soil–nail inter-
action is not adequately considered. Some researchers adopt a
three-dimensional finite element method to nailed soil slope anal-
ysis [1,31,36,40,42,43], but these researchers focused mainly on
the deformation and soil–nail interaction but not the factor of
safety. For a soil nailed slope, detailed three-dimensional (3D)
studies are still in need, as there is practically no two-dimensional
failure in practice. In this paper, the authors conduct a detailed
analysis of soil nailed slope by SRM and LEM which is seldom stud-
ied in the past. Some interesting results have been obtained which
worth further consideration and discussion in analysis and design.
To define the critical failure surface from the SRM, both the
maximum shear strain rate (the rate of variation of the shear strain
at the failure state) and the maximum shear strain increment (the
accumulated total shear strain at the failure state) definitions can
be used. A previous study [6] has demonstrated that these two def-
initions will give similar locations of the critical failure surface for
most cases.
If the failure of a soil nailed slope is internal, the failure mode
can be classified into the three different types – face failure, pull-
out failure, and nail tensile failure – shown in Fig. 1 (after FHWA
[2]). One of the important factors for a proper analysis of soil nailed
slope is the determination of the nail load. According to Hong
Kong’s practice (which is a special case of Davis’s method by Shen

0266-352X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2010.03.008
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
T0 T1
T2
T0
Failure mode- Face failure and “active”
zone sliding off front of nails
(a) Face failure
Fig. 1. Potential soil nail wall internal failure modes (after FHWA [2]).
et al. [27,28], the ultimate bond strength of a soil nail sf is ex-
pressed as sf ¼ pDc0 þ 2Dr0v tan /0 , where D is the hole diameter,
r0v is the effective vertical stress on the nail, tan /0 is the frictional
coefficient between the soil and the nail, and c0 is the cohesion of
the soil. Since the adhesion and friction between the soil and nail
will be less than c0 and tan /0 , respectively, a factor of safety of
2.0 is given to the ultimate bond strength for design purposes. An-
other soil nail design practice exists in which the bond stress is as-
sumed to be independent of the confining/overburden stress. It has
been suggested from laboratory and field tests in Hong Kong that
both design methods may be appropriate under different cases,
and the actual nail load appears to be dependent on the soil type,
time, the grouting pressure, and the topography. In the following
studies, the two methods of bond load determination (both options
have been used in Hong Kong, China, Europe and many other coun-
tries) are considered and the SRM and LEM are compared on an
equal basis.
The LEM method used in nailed slope analysis is based essen-
tially on Spencer’s method, and a previous study by Cheng et al.
[6] has demonstrated that this method is good enough for normal
problems. The effect of a soil nail is considered by applying a con-
centrated load provided by the nail on the slip surface. In the pres-
ent LEM analysis, only the tensile strength and pull-out capacity of
the nail are considered. The tensile forces mobilised in the nails are
divided into tangential and normal components along the potential
slip surface and are added to the resistant forces to determine the
safety factor of the slope. The effective nail load is taken as the
minimum of: (a) the bond strength between grout and soil; (b)
the tensile strength of the nail; (c) the bond strength between
the grout and the nail. In general, only factors (a) and (b) are
important in the analysis.
During the calculation of the bond strength, the portion behind
the failure surface is taken in the calculation for pull-out failure,
while the portion in front of the failure surface is taken in the cal-
culation for face failure (if specified), and the force, as determined,
is applied as a point load on the failure surface. The software
Slope2000 developed by Cheng is used for the present LEM analysis
of soil nailed slope. In the actual analysis, the applied nail load is
the effective nail load divided by the horizontal spacing of the
nails.
In this paper, the effects of the nail head, some special results
from SRM, the effects of failure induced by external loads and
the analysis of some model tests are discussed. These results are
important for the proper understanding and analysis of a nailed
slope.
2. Verification of SRM model and influence of nail head
Since SRM analysis – a relatively tedious analysis – is seldom
conducted for soil nailed slope, a proper soil–nail model is devel-
oped in this section for SRM analysis and the results are compared
with LEM to verify the effectiveness of this model. First, a 6 m-high
603
T1 TN
T2
TN
Failure mode- Nails pull out of Failure mode- Tensile failure
resistant zone of tendon
(b) Pullout failure (c) Nail tensile failure
soil nailed slope with a slope angle of 45° is analyzed. 8 m length
nails are installed at 1.5 m centers horizontally and vertically.
The diameter of the steel bar is 40 mm and the grout hole diameter
is 100 mm. The layout of the problem and the soil properties are
shown in Fig. 2. The minimum factor of safety (FOS) by Spencer’s
method using the simulated annealing search by Cheng [5] is
1.223 for pull-out failure and 1.146 for face failure.
Jewell and Pedley [17,18] have carried out series of studies to
investigate the bending effect of nail on slope stability, and have
concluded the bending effect of nail plays a beneficial role in stabi-
lizing the slope. The bending effect is however usually neglected in
most of the routine design as a conservative practice, and the
bending effect also relies on the quality of the cement grout which
is not completely reliable [16]. In the present study, the bending
action of nail will be assessed before the detailed study of soil
nailed slopes. For the SRM analysis, if bending effects are not
important, the nails can be modeled as cable elements, because
cable elements provide a shearing resistance (by means of the
grout properties) along their length. The shear behavior of the
cable-soil interface can be represented by the model in Fig. 3 and
can be described numerically in terms of the shear stiffness, cohe-
sive strength, friction angle, and exposed perimeter (grout hole
diameter) in the shear zone and the effective confining stress rm.
The material properties of the grouted nail are calculated consider-
ing a combination of the stiffness of the steel bar and the cement
grout. The Young’s Modulus of the grouted nail is determined as
45.44 GPa. At the same time, a thin layer of material with a thick-
ness of 4.0 mm surrounding the nail is used to model the shearing
zone between the nail and the soil that is shown in Fig. 3.
In this analysis, the elastic modulus (Es) and Poisson ratio (ms) of
the soil are 15 MPa and 0.42 respectively, so the shear modulus of
the shear zone is 0.5Es/(1 + ms) = 5.28 MPa. The shear stiffness of
the shear zone kg can be estimated as [13]
2p G
kg ¼ ¼ 43:1 MPa
10 lnð1 þ 2t=DÞ
where G is shear modulus of the shear zone and is equal to
5.28 MPa; D is the grout hole diameter and is equal to 0.1 m; t is
12,11 28,11
10.5,9.5 Soil nail
9.0,8.0
7.5,6.5
6,5
0,5 c’=9kPa
φ’=18°
Z Density=19kN/m3
0,0 28,0
X
Fig. 2. Soil nailed slope model.
604 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618

Soil

Grouted nail (cable)

Shear zone (grout)

Y
X

Fig. 4. Plot of the three-dimensional soil nailed slope model.

Fig. 3. Idealization of the soil nail system.

annulus thickness of the shear zone and is equal to 0.004 m. Since a

factor of safety of 2.0 is given to the ultimate bond strength of the
shear zone in the present analysis, the ‘‘Shear zone cohesive
strength (force) per unit length” is (9 kPapD)/2.0 = 1.4135 kN/m,
and the ‘‘Shear zone friction angle” is tan1[(tan 18°)/2.0] = 9.23°.
The parameters used for the study are shown in Table 1.
The commercial software FLAC3D is used for the SRM analysis.
Two different failure modes are simulated: face failure and pull-
out failure. For pull-out failure, a 0.15 m thick continuous elastic
surface is used to simulate the nail head, and the corresponding
numerical model for the nailed slope is shown in Fig. 4 (a contin-
uous facing). The front end of the nail is fixed onto the elastic sur-
face, and no failure is allowed between the front end of the end and
the nail head (this is achieved by assigning a very large shear zone
Fig. 5. Slip surface and the tension stress of soil nail without a nail head
cohesive strength value at the joint between the soil nail and nail
(FOS = 1.20).
head). For face failure simulation, the elastic surface, as shown in
Fig. 4, is removed so that there is no strong interaction between
the front end of the nail and the soil near to the slope surface.
When no nail head is simulated, the FOS by the SRM is 1.2 (asso- LEM with no soil nail

ciated flow rule) which is slightly smaller than the FOS for pull-out
failure by the LEM analysis but slightly larger than that by the face
failure for LEM analysis. Cheng et al. [6] and Wei et al. [41] (also the
results in the following section) have established that the FOS from
the SRM is usually larger than that from the LEM. The slip surface
and the tensile stress of the nail are shown in Fig. 5 (the slip surface
is determined by the shear strain rate distribution, since at the fail-
ure state the shear strain rate around the slip surface is much lar-
ger). The critical slip surfaces corresponding to face failure and
pull-out failure from the LEM are shown in Fig. 6. The slip surface
for face failure by the LEM agrees well with the slip surface from
the SRM. For pull-out failure, the slip surface from the LEM is how-
ever greatly different from that of the SRM in Fig. 5. Another inter-
esting phenomenon is that the slip surface from the SRM for nailed
slope without nail heads subjected to pull-out failure is virtually
the same as the one when the slope is un-reinforced (FOS = 1.11).
The authors have tried several other examples with similar obser-
vations, but this phenomenon is not found in the corresponding
LEM analyses and apparently unnoticed in the past. Although this
Table 1
Parameters of the grout–soil–nail system.
Young’s modulus of the grouted nail, E (GPa)
Shear zone cohesive strength (force) per unit length, cg (kN/m)
Shear zone friction angle, /g (°)
Shear zone stiffness per unit length, kg (MPa)
Shear zone exposed perimeter, pg (m)
Cross-sectional area of the grouted nail, A (m2)
Compressive yield strength of the grouted nail (force), Fc (MN)
Tensile yield strength of the grouted nail (force), Ft (MN)
LEM with soil nail (face failure)
LEM with soil nail (pull out failure)
Fig. 6. Slip surface for soil nailed slope without nail heads by the SRM and
compared with the LEM results.
phenomenon is not universal, it does appear to be commonly
found for SRM analysis. More detailed study of this phenomenon
will be considered later in this section. If a continuous elastic sur-
face is simulated as a nail head with 0.15 m thickness, the FOS for
this model is 1.28, the slip surface is shown in Fig. 7, and this fail-
ure surface is comparable to the pull-out failure slip surface by
LEM as shown in Fig. 8.
The results in Figs. 5–8 demonstrate the effect of a nail head on
the failure mode. This factor has seldom been considered in the
past, and there is not any previous SRM study on this issue. For
45.44 the model with no nail head, the failure mode is actually a face fail-
1.4135
9.23
ure, similar to that shown in Fig. 1a. The soil nail with no nail head
43.1 is restrained by the soil behind the failure surface, while the soil
0.339 failure mass is separated from the nail, so only the bond stress
0.00785 within the failure mass is effective in the stabilization. For the
0.238
model with a nail head, the failure mode is a pull-out failure, which
0.238
is similar to that shown in Fig. 1b. The failure along the nail is ini-
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 605

To figure out the influence of the nail head with different shapes
and parameters, two different types of nail head are considered. In
the first model, the nail head is 0.5 m wide and 0.5 m high, and its
elastic modulus varies from 15 MPa to 30 GPa. In the second mod-
el, the nail head is simulated as a continuous elastic surface, and its
elastic modulus also varies from 15 MPa to 30 GPa. In the above
analysis, the nail length is 8 m, so the difference of FOS between
face failure and pull-out failure is not large. In order to investigate
clearly the influence of the nail head, a slope with a 12 m nail
Fig. 7. Slip surface and the tension stress of soil nail for model with a nail head
length is considered, and the results are shown in Table 2. These
(FOS = 1.28). results shows that there is only a very small difference between
the two nail head configurations. Namely, there are noticeable dif-
ferences for a nailed slope with and without a nail head, but if the
LEM with no soil nail size of the nail head is large enough (e.g. 0.5 m wide and 0.5 m high
for this example), the size and material parameters of the nail head
LEM with soil nail
will not be important (provided that the material parameters are
normal and reasonable). If the nail head is small, the influence of
its size will be obvious. For example, the authors have tested a nail
head with a 0.14 m width and 0.15 m height (elastic modulus is
15 GPa) and a factor of safety of 1.29, as determined from the
SRM. The importance of the size of the nail head is not considered
in many design guidelines [10] in France, various FHWA publica-
Fig. 8. Slip surface for soil nailed slope by the SRM for the model with a nail head.
tions in US), but the various slope failures in nailed loose slopes
in Hong Kong [16] have suggested that inadequate nail head size
may be a possible reason for the various nailed slope failures in
tiated along the portion behind the failure mass, so the bond load Hong Kong. Based on the present analysis, the authors suggested
from the ‘‘effective length” behind the failure surface is successful that a suitable nail head size is required for the stabilization of
in the slope stabilization. The two failure modes, as found from the slopes, or else the face failure mechanism has to be considered in
study of nailed slope with and without nail heads, can be compared the analysis of a soil nailed slope (this failure mode is seldom con-
to the design of geotextiles for embankment stabilization, where sidered at present).
the bond loads for the two sides of geotextiles have to be checked As presented above, for the soil nailed slope shown in Fig. 2, if
(there is usually no head/anchorage to the geotextile). the nail head is not modeled, face failure will occur, and the slip
surface found with the SRM happens to be virtually the same as
when the slope is not reinforced (though the FOS are not the same).
Since such phenomena have been encountered in many cases, the
Table 2
FOS of nail head with different elastic moduli for 12 m nail length. validity of this observation under different scenarios deserves
more attention. In this section, three different models are consid-
Elastic modulus 30 GPa 15 GPa 15 MPa
ered. In the first model, the horizontal and vertical interval of the
FOS (0.5 m width and 0.5 m height nail head) 1.40 1.40 1.40
FOS (continuous nail head) 1.43 1.43 1.42 nails is 1.5 m (three nails are included in the model), which is
the same as in the model shown in Fig. 2. In the second model,

Maximum nail force=13.3kN Maximum nail force=16.2kN

(a) nail interval 1.5 m,FOS=1.20 (b) nail interval 1.0 m,FOS=1.35

Maximum nail force=24.5kN

(c) nail interval 0.75 m, FOS=1.65

Fig. 9. Slip surface of soil 1 with face failure for different nail intervals.
606 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
Maximum nail force=36.6kN
(a) nail interval 1.5m, FOS=1.32
Maximum nail force=51.2kN
(c) nail interval 0.75m, FOS=1.75
Fig. 10. Slip surface of soil 2 with face failure for different nail intervals.
the nail interval is 1.0 m (five nails in the model) while the nail
interval is 0.75 m in the third model (eight nails in the model).
At the same time, two different soil properties are considered: a
soil with c0 = 9 kPa and /0 = 18° (FOS = 1.11 with no nail), and the
second case where c0 = 20 kPa and /0 = 5° (FOS = 1.2 with no nail).
The results are shown in Figs. 9 and 10, and the slip surface with
no nail is shown with a dashed line. When the nail interval is large,
the slip surface is nearly the same as the one with no nail. When
the nail interval gets smaller, the difference becomes obvious.
For the first case, where /0 is greater, the slip surface gets shallower
with the decrease of the nail interval, while the slip surface gets
deeper with the decrease of the nail interval in the second case.
For usually adopted nail spacing, however, the location of the slip
surface for ‘‘face failure” is relatively insensitive to the presence of
soil nails.
Although 3D analysis is conducted for better consideration of
the soil–nail interaction for the present soil nailed slope, the slip
surface is still basically two-dimensional (the slip surface at differ-
ent cross-sections is nearly the same as that in Fig. 5). This phe-
nomenon is due to the relatively small nail spacing used in the
analysis. If the spacing of the nails is large enough, a clear three-
dimensional failure surface may occur. For the example shown in
Fig. 2, if the horizontal and vertical intervals of the nail are 10 m
and 1 m, respectively (five nails in the model) with a nail length
of 14 m, when the soil–nail interface strength is set to be the same
as the soil shear strength, a clear three-dimensional slip surface
will be mobilised (Fig. 11). Since the basic idea of the soil nailing
technique is to install closely spaced inclusions into the soil to in-
crease the stability, a clear three-dimensional slip surface would
not be easily mobilised for practical problems except at the edge
Fig. 11. Slip surface and nail force for slope with a 10 m horizontal nail interval
(FOS = 1.30).
Maximum nail force=45.6kN
(b) nail interval 1.0m, FOS=1.49
of the nailed slope. In the above presentation (Figs. 6–10), the slip
surface is shown in cross-section, and such representation will be
adequate. For the examples shown in the later sections, all the slip
surfaces by the SRM appear to be basically two-dimensional, as the
nail interval is not large, so the results will be compared directly
with the 2D LEM analyses.
3. Soil nailed slope stability with different soil properties
In this section, a parametric study for a homogeneous soil slope
with a slope height equal to 6 m and slope angle equal to 45° is
conducted, using the same slope geometry and soil nail distribu-
tion as the example in Fig. 2. In this parametric study, different
shear strength properties are used, and both the LEM and SRM
analyses are carried out. The cohesion of the soil varies among 2,
5, 10, and 20 kPa, while the friction angle varies among 5, 15, 25,
35, and 45°, respectively. The density, elastic modulus and Poisson
ratio of the soil are kept constant at 20 kN/m3, 15 MPa, and 0.35,
respectively, in all the analyses, and the results are shown in Ta-
ble 3. For the LEM, Spencer’s method is adopted, and the tolerance
in the location of the critical failure surface is 0.0001, which is good
enough for the present study.
From Table 3, the factors of safety found by the SRM and LEM
are very similar under different combinations of soil parameters,
and all the FOS obtained from the SRM (associated flow rule) are
slightly larger than those obtained by the LEM, with a maximum
difference of 12.4%. This phenomenon is similar to that obtained
by Cheng et al. [6] and Wei et al. [41] for un-reinforced slopes.
The slip surfaces for this study are shown in Figs. 12–14. When
the friction angle of the soil is small, the slip surfaces calculated
by the SRM and LEM are in good agreement. When the friction an-
gle of the soil becomes very large (for example, / = 45°), the slip
surfaces by the SRM are deeper than those by the LEM.
The authors have observed several special cases where a com-
bined failure surface may occur. For example, when c0 = 5 kPa and
/ = 35° (Fig. 15), a combined slip surface given by Fig. 15b will be
detected with the use of a very fine mesh, and such a special slip
surface will not occur when the mesh size is slightly increased
(Fig. 15a). Even though there are major differences in the two slip
surfaces, the differences in the FOS are very small, and such phe-
nomenon does not appear for the LEM. The determination of a
combined failure surface appears to be controlled by the size of
the mesh in the computation. The authors, however, believe that
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 607

Table 3
Factors of safety by LEM and SRM.

Case c0 (kPa) /0 (°) Factor of safety (LEM) Factor of safety (SRM) FOS difference between LEM and SRM (%)
1 2 5 0.27 0.28 3.70
2 2 15 0.59 0.63 6.78
3 2 25 0.99 1.09 10.10
4 2 35 1.55 1.72 10.97
5 2 45 2.34 2.63 12.39
6 5 5 0.44 0.46 4.55
7 5 15 0.80 0.86 7.50
8 5 25 1.23 1.33 8.13
9 5 35 1.80 1.96 8.89
10 5 45 2.59 2.9 11.97
11 10 5 0.70 0.74 5.71
12 10 15 1.14 1.19 4.39
13 10 25 1.63 1.71 4.91
14 10 35 2.17 2.37 9.22
15 10 45 2.97 3.3 11.11
16 20 5 1.15 1.26 9.57
17 20 15 1.67 1.79 7.19
18 20 25 2.20 2.34 6.36
19 20 35 2.82 3.04 7.80
20 20 45 3.67 3.95 7.63

Fig. 12. Slip surface comparison with increasing friction angle (c’ = 2 kPa).

such a combined failure surface is not a true phenomenon, as such
a failure surface possesses a high factor of safety when the LEM is
used. The authors suspect that due to the use of extremely fine
mesh, stress concentration occurs around the grout/nail interfaces,
so that the combined failure surface will come out. Actually, Cheng
et al. [6] and Wei et al. [41] have demonstrated the limitations of
the nonlinear solution algorithms in evaluating the ultimate limit
states of a slope, and the authors have also sometimes experienced
strange results for un-reinforced slopes if an extremely fine mesh
is used in the analysis. The authors have carried out serious analy-
sis for projects in China using the SRM, and this kind of strange fail-
ure surface appears not to be uncommon and is a highly mesh-
dependent problem. In this respect, engineers need to be very care-
ful in assessing the results from the SRM, particularly for a three-
dimensional analysis of a nailed slope.
4. Slope with different nail inclination angle
In this section, the influence of the nail inclination is dis-
cussed. The slope geometry and the soil nail distribution used
here are the same as in the example in Fig. 2. The nail length
608 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618

Fig. 13. Slip surface comparison with increasing friction angle (c0 = 5 kPa).

Fig. 14. Slip surface comparison with increasing friction angle (c0 = 20 kPa).

is 8 m, and the nail inclination varied from zero to 60°. Both Table 4. Two different models are developed for the SRM. In
the LEM and SRM are used, and the results are compared in the first model, the soil nail is simulated by a cable structure
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 609

element that considers only the tension effect (SRM1). In the
second model, the soil nail is simulated by a pile-structure ele-
ment that can consider both the tension and bending effect
(SRM2).
With the increase of the nail inclination angle, the factor of
safety first increases and then decreases, meaning that the nail
inclination angle can be optimised in practice. The maximum FOS
occurs when the inclination angle is between 10° and 30° (10°
for LEM; 20° for SRM1; 30° for SRM2). Although the optimised in-
clined angle is different for these three methods, the variation of
the FOS is very small when the inclination angle varies within
10–30°. Two optimum nail inclinations have been proposed in
the past: one that the nail should be horizontal; the other that
the nail should be perpendicular to the slip surface. For this exam-
ple, the factor of safety is insensitive to the inclination angle (from
10° to 30°), which is basically equivalent to varying the inclination

Fig. 15. Slip surface obtained by different element sizes (c0 = 5 kPa, / = 35°).

Table 4
Results for different nail inclination.

Inclination angle (°) 0 10 20 30 40 60

FOS by LEM 1.22 1.26 1.24 1.24 1.22 1.13
Maximum force of top nail by LEM (kN) 3.71 6.54 11.26 14.17 15.63 19.00
Maximum force of middle nail by LEM (kN) 5.9 9.21 13.85 17.06 18.80 21.54
Maximum force of bottom nail by LEM (kN) 8.74 12.39 17.28 20.47 21.72 23.10
FOS by SRM1 1.28 1.30 1.33 1.33 1.31 1.17
Maximum force of top nail by SRM1 (kN) 13.85 13.22 14.78 15.27 15.94 3.79
Maximum force of middle nail by SRM1 (kN) 17.22 21.55 26.92 30.25 33.00 23.33
Maximum force of bottom nail by SRM1 (kN) 28.15 31.41 35.80 39.57 41.69 24.65
FOS by SRM2 1.28 1.30 1.33 1.36 1.36 1.20
Maximum force of top nail by SRM2 (kN) 18.52 19.56 21.54 23.79 26.22 4.99
Maximum force of middle nail by SRM2 (kN) 21.35 26.49 28.77 33.46 36.15 22.13
Maximum force of bottom nail by SRM2 (kN) 30.32 34.98 38.84 41.59 44.76 23.77
Maximum moment (kN m) 1.193 1.215 1.079 1.744 1.820 1.819

(a) nail inclination angle = 0 (b) nail inclination angle = 10°

(c) nail inclination angle = 30° (d) nail inclination angle = 60°
Fig. 16. Slip surface and nail axial force distribution for different nail inclination angles (nail simulated by cable element).
610 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618

(a) nail inclination angle = 0 (b) nail inclination angle = 10°

(c) nail inclination angle = 30° (d) nail inclination angle = 60°
Fig. 17. Slip surface and nail axial force distribution for different nail inclinations (bending effect considered).

(a) nail inclination angle = 0 (b) nail inclination angle = 10°

(c) nail inclination angle = 30° (d) nail inclination angle = 60°
Fig. 18. Slip surface and nail bending moment distribution for different nail inclinations (bending effect considered).

Maximum nail force=42.2kN Maximum nail force=22.1kN

(a) Short nail at top, FOS=1.31 (b) short nail at bottom, FOS=1.22

Fig. 19. Slip surface and the tension stress in different layouts with zero inclination angle.

from horizontal to nearly perpendicular to the slip surface. When The slip surface, nail axial force distribution, and bending mo-
the inclination is very high (say 60° or more), the FOS will then ment distribution are shown in Figs. 16–18 and Table 4. The max-
drop quickly. imum mobilised nail axial force increases initially and then
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 611

Maximum nail force=28.3kN

Maximum nail force=60.9kN

(a) short nail at top, FOS=1.40 (b) short nail at bottom, FOS=1.26
Fig. 20. Slip surface and the nail load for different layouts with a 20° nail inclination.

Maximum nail force=46.4kN Maximum nail force=30.2kN

(a) short nail at top, FOS=1.34 (b) short nail at bottom, FOS=1.22
Fig. 21. Results for different layouts with zero inclination angle (constant nail pull-out strength).

Maximum nail force=46.4kN Maximum nail force=30.7kN

(a) short nail at top, FOS=1.35 (b) short nail at bottom, FOS=1.26
Fig. 22. Results for different layouts with a 20° nail inclination (constant nail pull-out strength).

decreases with the nail inclination, a trend that also explains why
the FOS first increases and then ultimately decreases. Some bend-
ing effect is also observed to be mobilised in the nail (Fig. 18), espe-
cially when the nail inclination is large (30–60°), but the effect of
the bending moment in the nail is very small compared with the
axial nail force (Table 4), so that the contribution of the bending ef-
fect to the factor of safety is also very small. This result also sup-
ports the current method of design, where the bending effect of
soil nail is neglected in the analysis.
eliminate the influence of the overburden confining pressure,
two other models are developed in which the pull-out strength
(5.6 kN/m) is assumed to be independent of the confining pressure.
The results of these models are shown in Figs. 21 and 22. For these
new models, the FOS in case 1 is still larger than that for case 2.
Case 1 appears to be a more effective solution for enhancing the
stability of a slope.
Classically, case 2 is recommended for reinforced earth struc-
tures, as it better controls the displacement of the slope crest.

5. Slope with different soil nail layout

Table 5
Factor of safety with top pressure (bond load controlled by overburden stress).
In this section, the influence of the soil nail layout is discussed
Loading (kPa) 150 300
using two different cases. The slope geometry and the soil nail dis-
FOS by SRM 2.10 1.75
tribution are also the same as in example 1 in the previous section. FOS by LEM 1.55 1.17
In case 1, the soil nails are shorter at the upper part of the slope and Difference of FOS between SRM and LEM 35.5% 49.6%
longer at the bottom (the nail lengths of the top row, middle row
and bottom row are 4 m, 8 m, and 12 m, respectively). In case 2,
the design is reversed (the nail lengths of the top row, middle
row and bottom row are 12 m, 8 m and 4 m respectively). Two dif- Table 6
Factor of safety with top pressure (constant pull-out resistance).
ferent nail inclination angles are also considered: 0° and 20°
(Figs. 19 and 20). For both nail inclination angles, the FOS for case Loading (kPa) 150 300
1 is larger than that for case 2. Since the nail pull-out strength is FOS by SRM 1.83 1.40
FOS by LEM 1.67 1.24
assumed to be defined by the overburden/confining pressure for
Difference of FOS between SRM and LEM 9.6% 12.9%
the results in Figs. 19 and 20, such results are not surprising. To
612 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
For the ultimate limit state of slope stability, however, case 1 is
found to be more effective, as more nail forces can be mobilised.
A more detailed study on the shear strain distribution reveals that
failure initiates at the lower part of the slope. If longer soil nails are
installed at the bottom, they will take the loading more effectively
so that the factor of safety will be higher. On the other hand, if the
slope is formed by excavation, a longer nail at the top will be more
effective in the stability as the failure will start from top to bottom.
It can be concluded that the longer soil nails should be installed at
locations where failure starts for maximum efficiency.
6. Soil nailed slope with external pressure on the top
In this section, the influence of the external pressure on the SRM
analysis is discussed. There is a major difference between the SRM
and LEM in dealing with external loads that has not been discussed
in the past. The slope height, slope angle, and the soil nail distribu-
tion used here are also the same as in example 1, and the nail
length is 12 m. The cohesion and friction angle of the soil are
18 kPa and 36°, respectively. Two different loadings (150 kPa and
300 kPa) are applied on the ground surface of the slope, and both
the SRM and LEM results are shown in Tables 5 and 6. Two differ-
ent SRM models are developed for this study. In the first model, the
bond strength is assumed to be controlled by the confining pres-
sure. The results for the first model are shown in Table 5 and are
compared with the results by the LEM. For the ultimate pull-out
resistance of the nail, it is the lower of the ultimate tensile strength
of the nail and either the first or the second nail bond model. There
are major discrepancy between the results from the SRM and the
LEM, and the differences increase with increased applied loading.
The authors’ view is that such major differences arise from the in-
creased confining pressure and hence the bond strength in the SRM
analysis. On the other hand, the overburden pressure on the nail
from the external load is usually not considered in the bond
strength calculation (appear to be the practice for all commercial
programs), so a lower bond strength is determined from the LEM.
In this respect, there is a major difference in the soil nail design
by the SRM and LEM under the action of the external loads if the
bond load is the function of the confining stresses. In Table 6,
where the bond stress is independent of the effective overburden
stress, the differences in the factors of safety are small even a high
surcharge is applied. On the other hand, in Table 5 where the bond
load depends on the effective normal stresses on the nail (Hong
Table 7
Factor of safety with different slope angles (slope height 6 m, nail length 12 m).
Slope angle (°) 45 60 75
FOS by SRM (E = 45.44 GPa) 1.41 1.23 1.06
FOS by SRM (E = 4.544 GPa) 1.44 1.27 1.10
Table 8
Factor of safety with different slope angles (slope height 8 m, nail length 12 m).
Slope angle (°) 45 60 75
FOS by SRM (E = 45.44 GPa) 1.37 1.26 1.12
FOS by SRM (E = 4.544 GPa) 1.54 1.48 1.40
Table 9
Factor of safety with different slope angles (slope height 8 m, nail length 6 m).
Slope angle (°) 45 60 75
FOS by SRM (E = 45.44 GPa) 1.91 1.61 1.51
FOS by SRM (E = 4.544 GPa) 1.92 1.79 1.71
Distribution of tension
force along nail
Facing
Active zone
Resistant zone
Line of maximum tension
force
Fig. 23. Load transfer mechanism in soil nails (after FHWA [2]).
Kong and US practice), the differences in the results are much more
pronounced. If the applied loading is small, there will not be major
differences between the two bond models, and the use of a high
surcharge will amplify the differences for illustrating the effect of
the bond model.
In order to eliminate the influence of the confining stresses on
the bond strength, a second model is developed in which the bond
strength is assumed to be independent of the confining pressure
(bond strength = 11.2 kN/m). The results of these analyses are
shown in Table 6. Since the basic assumption for the bond stress
is the same for the SRM and LEM, the differences between the re-
sults from the LEM and SRM are smaller than the case in Table 5. It
can be concluded that the SRM and LEM may give very different re-
sults depending on the bond load determination method, and engi-
neers should be aware of this difference.
7. Influence of nail elastic modulus
In the previous sections, the elastic modulus of the soil nail used
for the analysis is 45.44 GPa, which is determined from a combina-
tion of the stiffness of the steel bar and the cement grout (shown in
Table 1). In the LEM, however, the nail stiffness is not required in
the analysis. To investigate the influence of the nail elastic modu-
lus on the stability of slope, two different kinds of nail elastic mod-
uli are considered for two different slope heights (6 m and 8 m),
and the slope angle is varied from 45° to 90°. The first nail stiffness
is 45.44 GPa, which corresponds to a grouted steel bar nail. The
second nail stiffness is 4.544 GPa, which approximately corre-
sponds to a glass fiber reinforced plastic nail, which is much more
flexible.
For the 6 m-high slope (nail interval is 1.5 m and nail length is
12 m), when the slope angle varies from 45° to 90°, the FOS for the
90 two different nail elastic moduli are nearly the same, as shown in
0.87 Table 7. For the 6 m slope height model, two other situations are
0.90 also considered: in the first case, the nail interval is changed to
1 m, and in the second case, the nail length is changed to 6 m.
The phenomena for this case are similar to the previous model,
where the elastic modulus of the nail has little influence on the
FOS (the detailed results are hence not shown here).
For the slope with 8 m height, two different models are consid-
90 ered. In the first model, the nail length is 12 m, nail interval is
0.91 1.5 m, and the cohesion and friction angle of the soil are 9 kPa
1.14
and 18°, respectively. For this model, the soil nail stiffness displays
noticeable effects, which are shown in Table 8. These results indi-
cate that a softer soil nail is better for slope stability, provided that
neither the nail pull-out strength nor the nail tensile strength are
reduced. In the second model, the nail length is 6 m, nail interval
is 1.5 m, and the cohesion and friction angle of the soil are
90 18 kPa and 25°, respectively. For this model, when the slope angle
1.25
is 45°, the FOS for different nail elastic modulus is nearly the same,
1.58
while when the slope angle is 60–90°, there are obvious differences
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 613

Local failure

(a) global failure (b) sliding failure (c) bearing capacity failure
Fig. 24. Axial tensile force distribution of soil nails under different external failure modes.

Locus of maximum tension

(a) no nail head, FOS=1.20, face failure (b) strong nail head, FOS=1.28, pullout failure

(c) strong nail head, nail tensile strength=10 kN, FOS=1.22 with nail tensile failure
Fig. 25. Slip surface and nail tension stress distribution under different internal failure modes.

Locus of maximum tension forces

Slip surface at limit state Locus of maximum tension forces

(a) service state (b) limitstate, FOS=1.41

Fig. 26. Axial tensile force distribution of soil nail under different state.

for the FOS, which shows that softer soil nail stiffness is usually
better. This phenomenon can be explained by the fact that more
soil movement and hence stress-redistribution is mobilised for
softer nails in a slope with great height, so that the mobilisation
of the bond stress will be greater. If the height of slope is large
but the nail length is short, when the slope angle is small, the
amount of soil mass affected, and hence the soil movement mobi-
lisation effect, will be much smaller. This phenomenon can also be
compared with the extensible and inextensible designs for rein-
forced earth design [3] (Table 9).
8. Distribution of the nail tension force and critical slip surface
The line of maximum tension within the nail is often considered
to be the failure surface, a convention that divides the soil mass
into two separate zones: (a) an ‘‘active zone” close to the slope sur-
face, where the shear stresses exerted by the soil on the reinforce-
ment are directed outward and tend to pull the reinforcement out
of the ground and (b) a ‘‘resistant zone” where the shear stresses
are directed inward and tend to restrain the reinforcements from
being pulled out. This concept is shown in Fig. 23 (after FHWA
[2]), and many reinforced earth structures are designed in this
way. It should be noted that the line of maximum tension does
not correspond to the conventional critical slip surface, as defined
in limiting equilibrium stability analyses, but reflects the results of
the soil structure interactions between the soil and the nail/facing
reinforcement system [2,3]. Classically, many researchers assume
that the maximum tensile force line coincides with the potential
sliding surface, and such a view is supported by some model tests
[19] and full-scale tests (Clouterre [10]). In this section, the authors
614 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618

50kPa

(a) limit state by reducing strength, FOS=1.31 (b) limit state by applying pressure near crest

120kPa

(c) limit state by applying 120 kPa pressure at slope top near the nail end
Fig. 27. Axial tensile stress distribution under different limit states.

50
(kN) 100 60
40 150
30
20 1
10 0
0

(kN) 40 150 100

30 60 Failure surface
20 2
10 0
0 Tensile force
110 100
(kN) 40 150
30
20 60
20kN
0 3
10
0 10kN
110
(kN) 50 150
40
30 100
20 60 0 4
10
0 110
(kN) 50 Slip surface
40 150
100
30
20 60 5
Line of maximum
0
10
0
tension force

(a) Gassler [14] (b) Stocker [33]

Observed crack
T(kN)
20

10
20 Tensile stresses
10 distribution
20

10
20

10
20

10

Slip surface

(c) Clouterre [10]

Fig. 28. Nail force distribution under the limit state. (See above-mentioned references for further information.)
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618 615

will discuss this issue by conducting strength reduction analyses of
several soil nailed slopes.
Firstly, the influence of the failure modes is investigated. Soil
nailed slope failure can be broadly classified into external failure
modes and internal failure modes. For external failure modes, the
slip surface does not intersect with the nails, so the maximum ten-
sile force line will not coincide with the potential sliding surface.
Three different external failure modes are shown in Fig. 24. In
Fig. 24a, if the soil nail is very short, a global failure will occur,
where the nails are totally within the failure mass. In Fig. 24b,
the soil nail is long and the soil mass/nail becomes an integral body
with a sliding failure mode. Fig. 24c shows one of Seattle’s first soil
nailed walls discussed in the previous section, for which the rein-
forced soil nailed wall is more stable than the toe of the slope, mak-
ing it a local bearing capacity failure. Internal failure modes are
usually classified into three different types – face failure, pull-out
failure, and nail tensile failure, all of which are shown in Fig. 1
(after FHWA [2]). These three different internal failure modes are
modeled for the slope, as shown in Fig. 2, and the results are shown
in Fig. 25. In Fig. 25a, which shows a face failure with no nail head,
and the soil nail is restrained by the soil mass behind the failure
surface so that only the friction within the failure mass is effective
in the stabilization. In this case, the line of the maximum tension
force is located behind the slip surface. In Fig. 25b, the line of the
maximum tension force is close to the nail head and is in front of
the slip surface. In Fig. 25c, where the tensile strength of the nail
is only 10 kN, the nail reaches its tensile strength at the ultimate
limit state and the line of the maximum tension force virtually
coincides with the slip surface.
Secondly, the influence of the state of the slope (service state
and limit state) is investigated. The slip surface should be refer-
enced to the condition of the failure mass at the limit state. The
maximum tension force line can be referenced at the service state
or the limit state, the choice of which should be stated clearly. To
consider this, a model is developed with 12 m length soil nail with
a nail head to compare these two states, and the slope geometry
and the soil nail distribution is the same as in example 1. The re-
sults of this analysis are shown in Fig. 26. In Fig. 26a (service state),
the slip surface at the limit state is shown by the dashed line for
comparison. This analysis illustrates that the line of the maximum
tensional force at the service state is behind the slip surface. The
force at the nail head connection in the service state is very small.
In the limit state (Fig. 26b), the line of the maximum tension force
is in front of the slip surface, and a relatively large force at the nail
head connection is mobilised. With the reduction of the soil shear
strength. Gradually, the soil nailed slope transforms from service
state to limit state, and during this process, the nail force is grad-
ually mobilised, while the line of maximum nail force will gradu-
ally move towards the nail head. In this example, which is an
internal pull-out failure, the slip surface appears to locate between

5
Predicted Locus of Maximum Tensile Force
4
Reinforced Earth
3 Hwall=30ft Method Davis Method
2 Hwall=24ft
Hwall=48ft
Axial force (kips)

1
2nd level anchor rod
0
4
Hwall=30ft
3
2
Hwall=24ft
1
3rd level anchor rod
0 KIPS kN
20
4 100

3 0
0
Hwall=30ft LOAD
2
1 4th level anchor rod
0
0 5 10 15 20
Distance behind wall facing (ft.)

(a) Shen [29] (b) Thompson [39]

Scale of stresses
σ = 40 kPa

Tensile stresses distribution

Nail length (m)
0 1 2 3 4 5 6 7 8
N4
Nail force (kN)

20 N1
N1 N2
40 N3
N4
60 N2
N3
Attempt for delimitation between
active and passive zone

(c) Cartier and Gigan [4] (d) Stocker [33]

Fig. 29. Nail force distribution under the service state. (See above-mentioned references for further information.)
616 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
the lines of maximum nail force at the service state and that at the
limit state.
In the above analysis, the limit state is achieved by reducing the
shear strength, a change that is easy to conduct in numerical sim-
ulations but not in model tests or full-scale tests. The limit state
can also be achieved by applying pressure on top of a slope, and
an example is shown in Fig. 27, where the results at the limit state
are achieved in different ways. In Fig. 27a, the limit state is
achieved by reducing soil shear strength, and the nail force at the
bottom row is slightly more mobilised than in the upper two rows.
In Fig. 27b, the limit state is achieved by applying a pressure near
the crest of the slope with an internal failure, and the nail force at
the top row is more mobilised. In Fig. 27c, the limit state, which is
an external failure, is achieved by applying a pressure on top of the
slope near to the ends of the soil nails, and the nail force at the bot-
tom row is very large, while the nail forces at the upper two rows
are much lower.
Some major test results in the literature are summarised in
Figs. 28 and 29. In Fig. 28a [14] and Fig. 28b ([33]), the limit state
is achieved by applying pressure on top of the slope. In Fig. 28c,
(Clouterre [10]), the limit state is achieved by gradually saturating
the soil, which can be viewed as approximately equivalent to
reducing the soil strength. At the limit state, the slip surface can
be observed, and both the slip surface and the nail tension force
distribution are given in Fig. 28. In Fig. 28a and b, both the slip sur-
faces only partially intersect the soil nails, and the limit state is
actually caused by the applied load on top of the slope. The loca-
tion of the slip surface is controlled mainly by the location of the
loading, as the slip surfaces starts from the edge of the loading
(similar to those examples in Fig. 27b and c). Since the locations
of the slip surfaces are controlled mainly by the location of the
loading in Fig. 28a and b, and the slip surface only partially inter-
sects the soil nails, the line of maximum tension force will not
coincide with the slip surface. It can therefore be seen that the
maximum tension force in the lower rows of nails is very close
to the nail heads. In Fig. 28c, where the ultimate state is controlled
by the shear strength reduction, an internal tension failure occurs,
and the line of the maximum tension force coincides well with the
slip surface. In this test, the structure had been designed with a
sufficiently low safety factor for failure by the breakage of the nails
(F = 1.1). When the soil is gradually saturated, nail breakages
caused system failure, which is similar to the SRM results in the
previous section.
There are also four different soil nailed test results in the service
state shown in Fig. 29. In the service state, the slip surface does not
exist, so that only the tension force distribution is presented in
these figures (in Fig. 29b, two predicted slip surfaces are also
shown). In Fig. 29a [29], the measured maximum tension force is
close to the nail head. In Fig. 29b–d, the maximum tension force
at the lower part of the slope is also very close to the nail head,
while the maximum tension force at the upper part is slightly far-
ther away from the nail head.
Based on the above analysis, the line of the maximum tension
does not correspond to the conventional critical slip surface in gen-
eral, but rather reflects the results of the soil structure interactions
between the soil and the nail/facing reinforcement system. First,
the tension stress distribution is influenced by the state of the
slope and is obviously different between the service state and
the limit state. When the soil nailed slope gradually transforms
from the service state to the limit state by reducing the shear
strength, if face failure is prevented by the use of nail head, the
maximum tension line will move towards the slope surface (such
as Fig. 26) as the nail force is gradually mobilised and the resistant
zone gets larger to maintain the slope in a stable condition. Second,
the tension stress distribution is influenced by the slope failure
modes at the limit state. For slope with internal failure modes, if
face failure occurs, the line of maximum tension is normally lo-
cated behind the slip surface (Fig. 25a). For a pull-out failure, the
line of the maximum tension is usually located in front of the slip
surface (Figs. 25b and 26b). For a nail tensile failure, the line of the
maximum tension coincides well with the slip surface. An agree-
ment could be demonstrated by both the numerical simulation
(Fig. 25c) and the full-scale test (Fig. 28c). For slopes with external
failure modes, the line of the maximum tension force will not coin-
cide with the slip surface. Besides, both the nail tension force dis-
tribution and the slip surface are greatly controlled by the location
of the external loading (Fig. 27).
9. Discussions and conclusions
This paper establishes that the Poisson ratio and the stiffness/
arrangement of the nail heads have little influence on the factor
of safety and the failure mechanism of a slope. The presence of a
nail head is important, however, and should be properly modeled.
There have been slope failures in Hong Kong that can be attributed
to the use of very small nail heads (and hence face failure), so that
this effect should be checked in the analysis and design. The possi-
bility of face failure, however, is absent in many commercial slope
stability programs and is also neglected by many engineers. Engi-
neers should generally provide nail heads of adequate size. As long
as the nail head is not too small, a pull-out failure will be control-
ling failure mechanism, as is investigated in the present study.
This study has found that the FOS from the SRM and the LEM are
similar under most cases, and that all of the FOS from the SRM are
slightly greater than those from the LEM. Although the slip surfaces
by the SRM and the LEM usually agree well, sometimes the slip
surface from the SRM is not regular, a fact that is one of the limi-
tations of SRM as mentioned by Cheng et al. [6] and Wei et al.
[41]. The analysis of a soil nailed wall in Seattle indicates that sev-
eral failure modes can easily be detected by the SRM that are ab-
sent in the corresponding LEM analyses. In this respect, the SRM
can be an alternative or complementary method to the LEM anal-
ysis. On the other hand, some special combined failure surfaces
similar to that in Fig. 12b are found when a very fine mesh is used
in the analysis, and the authors have encountered such problems
for some projects in China, which means that the engineers need
to be careful in the mesh design of a problem. In this respect, the
SRM possesses numerical problems not found in the LEM.
The nail elastic modulus appears to have little influence on the
SRM analyses except for very steep slopes. The FOS from the SRM
also appears to increase with a decreasing nail elastic modulus.
This observation can be attributed to the greater soil movement
and mobilisation of a nail bond load. This phenomenon can also ac-
count for the differences in the design of reinforced earth wall
using extensible and inextensible strip [2,3], discrepancies that
are based on field observations but are also obtained from numer-
ical modeling in this study.
When there is an external pressure on top of a slope and the
bond stress is assumed to be dependent on the overburden/confin-
ing stress, there are great differences between the SRM and LEM. In
all commercial slope stability programs based on the LEM, the in-
crease in the overburden stress from the external loads on the nails
are not included, as there is no reliable method in distributing the
external loads under the ultimate limit state. The simplified 1 in 2
rule of thumb or other empirical methods are all based on the ser-
vice state instead of the ultimate limit state and are not included in
normal LEM analysis. On the other hand, the SRM can consider the
external load easily and automatically, a capability that offers an
advantage not present in the LEM. Engineers should also be aware
of this difference in the bond load determination in LEM and SRM,
and such differences have not been considered up to present.
 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
In soil nailing design, the nail inclination angle and the layout
can be optimised by the SRM or LEM analyses. Usually, with the in-
crease of the nail inclination angle, initially the FOS slightly in-
creases and then decrease if the inclination is large. An optimum
soil nail design can be determined easily by using whether the
SRM or the LEM. The results of this study also show the mobilised
bending moment in soil nail to be small even when the nail incli-
nation angle is large. The use of the bond stress and neglect the
bending effect for soil nail design, a commonly adopted approach,
appears to be a reasonable design method.
An optimum design of longer nails at the top and shorter nails
at the bottom has been recommended by some researchers and
engineers for the control of movement for soil nails installed in a
top-down construction. For the stabilization of existing slopes,
from the safety factor point of view, the SRM analysis shows that
the reverse arrangement will be a more economic solution for
the ultimate limit state. The final recommendation to the optimum
soil nail layout from the present study is to place longer nails at
locations where failure will start. The choice of longer or shorter
nails at the top or bottom will depend on the initiation of the pos-
sible failure mechanism.
The tension force distribution along the soil nails is observed to
be controlled by the state of the slope (service state, limit state)
and the failure modes (external failure, internal failure). In general,
the line of maximum tension does not correspond to the conven-
tional critical slip surface, a finding that is different from what is
common believed. This result originates from soil structure inter-
actions between the ground and the nail/facing reinforcement sys-
tem. For a slope with external failure modes, the line of maximum
tension force will never coincide with the slip surface. For slopes
with internal face failure modes, the line of maximum tension is
usually found to be located behind the slip surface. For a pull-out
failure, the line of maximum tension is usually in front of the slip
surface, while for a tensile failure, the line of maximum tension
will coincides with the slip surface, as is the common belief.
The limit state can be achieved by reducing the soil shear
strength or by applying external loading. This study demonstrates
the differences in the nail force distribution and slip surface be-
tween these two modes. When the soil nailed slope gradually
transforms from the service state to the limit state by a reduction
in the shear strength, if the nail head is strong enough with no face
failure, the maximum tension force line will move towards the
slope surface.
Acknowledgement
The present project is funded from Research Grants Council
through the Project PolyU 513507E.
References
[1] Briaud JL, Lim Y. Soil-nailed wall under piled bridge abutment: simulation and
guidelines. J Geotech Geoenviron Eng 1997;123(11):1043–50.
[2] Byrne RJ, Cotton D, Porterfield J, Wolschlag C, Ueblacker G. Manual for design
and construction monitoring of soil nail walls. Federal Highway
Administration (FHWA) Report No. FHWA-SA-96-069, Washington, DC, USA;
1996.
[3] Byrne RJ, Cotton D, Porterfield J, Wolschlag C, Ueblacker G. Manual for design
and construction of soil nail walls, FHWA-SA-96-069R. Federal Highway
Administration (FHWA), Washington, DC, USA; 1999.
[4] Cartier G, Gigan JP. Experiments and observations on soil nailing structures. In:
Proceedings of the 7th European conference on soil mechanics and foundation
engineering, Helsinki, Finland; 1983.
[5] Cheng YM. Locations of critical failure surface and some further studies on
slope stability analysis. Comput Geotech 2003;30:255–67.
[6] Cheng YM, Lansivaara T, Wei WB. Two-dimensional slope stability analysis by
limit equilibrium and strength reduction methods. Comput Geotech
2007;34(3):137–50.
617
[7] Cheng YM, Lansivaara T, Wei WB. Reply to comments on ‘‘two-dimensional
slope stability analysis by limit equilibrium and strength reduction methods”.
Comput Geotech 2008;35(2):309–11.
[8] Cheuk CY, Ng CWW, Sun HW. Numerical experiments of soil nails in loose fill
slopes subjected to rainfall infiltration effects. Comput Geotech
2005;32(4):290–303.
[9] Choukeir, Mohamed, H. Seismic analysis of reinforced earth and soil-nailed
structures. Ph.D. Thesis, New York Polytechnic University; 1996.
[10] Clouterre. Recommendations Clouterre 1991. US Department of
Transportation, Federal Highway Administration (English translation, Report
on the French National Project Clouterre, Report No. FHWA-SA-93-026,
Washington (DC)); 1993.
[11] Dawson EM, Roth WH, Drescher A. Slope stability analysis by strength
reduction. Geotechnique 1999;49(6):835–40.
[12] Donald IB, Giam SK. Application of the nodal displacement method to slope
stability analysis. In: Proceedings of the 5th Australia–New Zealand
conference on geomechanics, Sydney, Australia; 1988. p. 456–60.
[13] FLAC 3D Version 3.1 User’s guide. Minneapolis (Minnesota, USA): Itasca
Consulting Group, Inc.; 2006.
[14] Gassler G. The first two field tests in the history of soil nailing on nailed walls
pushed to failure, soil reinforcement: full scale experiments of the 80’s. Presses
de l’ecole Nationale des Ponts et Chaussees, CEEC; 1993. p. 7–34.
[15] Griffiths DV, Lane PA. Slope stability analysis by finite elements. Geotechnique
1999;49(3):387–403.
[16] Hong Kong Institution of Engineers. Soil nails in loose fill slopes – a
preliminary study, Hong Kong; 2003.
[17] Jewell RA, Pedley MJ. Soil nailing design: the role of bending stiffness. Ground
Eng 1990;23(2):30–4.
[18] Jewell RA, Pedley MJ. Analysis for soil reinforcement with bending stiffness. J
Geotech Eng ASCE 1992;18(10):1505–28.
[19] Juran I, Beech J, Delaure E. Experimental study of the behavior of nailed soil
retaining structures on reduced scale models. In: Proceedings of the
international conference on in-situ soil and rock reinforcements, Paris; 1984.
[20] Juran I, Shafiee S, Schlosser F. Numerical study of nailed soil retaining
structures. In: Proceedings of the eleventh international conference on soil
mechanics and foundation engineering, San Francisco, vol. 3; 1985. p. 1713–6.
[21] Juran I, Elias V. In: Joseph P, editor. WelshSoil nailed retaining structures:
analysis of case histories, soil improvement – a ten year update: proceedings
of a symposium sponsored by the Committee on Placement and Improvement
of Soils of the Geotechnical Engineering Division of the American Society of
Civil Engineers in conjunction with the ASCE Convention in Atlantic City, New
Jersey, April 28, 1987; 1987. p. 232–44.
[22] Matsui T, San KC. Finite element slope stability analysis by shear strength
reduction technique. Soils Found 1992;32(1):59–70.
[23] Naylor DJ. Finite elements and slope stability. Numer meth in geomech, Lisbon
(Portugal): Peoc. NATO Advanced Study Institute; 1981. p. 229–44.
[24] Plumelle BC, Schlosser F, Delage P, Knochenmus G. In: Lambe Philip C, Hansen
Lawrence A, editors. French national research project on soil nailing Clouterre.
Design and performance of earth retaining structures: proceedings of a
conference/sponsored by the Geotechnical Engineering Division of the
American Society of Civil Engineers in cooperation with the Ithaca Section,
ASCE, Cornell University, Ithaca, New York, June 18–21, 1990; 1990. p. 660–75.
[25] Schlosser F. Behaviour and design of soil nailing. In: Proc symp on recent
developments in ground improvement techniques, Bangkok, Thailand; 1982.
p. 399–413.
[26] Schlosser FS. The multicriteria theory in soil nailing. Ground Eng; November,
1991.
[27] Shen CK, Herrmann LR, Romstad KM, Bang S, Kim YS, DeNatale JS. In situ earth
reinforcement lateral support system. Department of Civil Engineering,
University of California, Davis, Report No. 81-03; 1981.
[28] Shen CK, Bang S, Romstad KM, Kulchin L, DeNatale JS. Ground movement
analysis of an earth support system. ASCE J Geotech Eng Div
1981;107:1609–24.
[29] Shen CK, Bang S, Romstad KM, Kulchin L, DeNatale JS. Field measurements of
an earth support system. ASCE J Geotech Eng Div 1981;107:1625–42.
[30] Sivakumar Babu GL, Srinivasa Murthy BR, Srinivas A. Analysis of construction
factors influencing the behaviour of soil-nailed earth retaining walls. Ground
Improv 2002;6(3):137–43.
[31] Smith IM, Su N. Three-dimensional FE analysis of a nailed soil wall curved in
plan. Int J Numer Anal Method Geomech 1997;21:583–97.
[32] Song E. Finite element analysis of safety factor for soil structures. Chinese J
Geotech Eng 1997;19(2):1–7 [in Chinese].
[33] Stocker MF, Korber GW, Gassler G, Gudehus G. Soil nailing. In: International
conference on soil reinforcement, Paris; 1979. p. 469–74.
[34] Stocker MF, Riedinger G. The bearing behaviour of nailed retaining structures.
In: Proc conf on design and performance of earth retaining structures,
Geotechnical special publication no. 25, Ithaca, USA; 1990. p. 612–28.
[35] Srinivasa Murthy BR, Sivakumar Babu GL, Srinivas A. Analysis of construction
factors influencing the behaviour of soil-nailed earth retaining walls. Ground
Improv 2002;6(3):129–36.
[36] Tabrizi KS, Gucunski N, Maher MH. ‘‘3-D FEM analysis of excavation of a soil-
nail wall. In: Computing in civil engineering: proc. 2nd congress held in
conjunction with A/E/C Systems ‘95, Atlanta, Georgia; 1995. p. 812–9.
[37] Thompson SR, Miller IR. Design, construction and performance of a soil nailed
wall in Seattle, Washington. Design and performance of earth retaining
structures. ASCE Geotech Spec Publ 1990(25):629–43.
618 W.B. Wei, Y.M. Cheng / Computers and Geotechnics 37 (2010) 602–618
[38] Ugai K, Leshchinsky D. Three-dimensional limit equilibrium and finite element
analysis: a comparison of results. Soils Found 1995;35(4):1–7.
[39] Unterreiner P, Benhamida B, Schlosser F. Finite element modelling of the
construction of a full-scale experimental soil-nailed wall. French National
Research Project CLOUTERRE. J Ground Improv 1997;1(1):1–8.
[40] Yang MZ, Drumm EC. Numerical analysis of the load transfer and deformation
in a soil nailed slope. Numerical methods in geotechnical engineering – recent
developments. In: Proceedings of sessions of geo-denver 2000, Geotechnical
special publication no. 96, August 2000, Denver, Colorado; 2000. p. 102–15.
[41] Wei WB, Cheng YM, Li L. Three-dimensional slope failure analysis by the strength
reduction and limit equilibrium methods. Comput Geotech 2009;36:70–80.
[42] Zhang MJ, Song EX, Chen ZY. Ground movement analysis of soil nailing
construction by three-dimensional (3-D) finite element modeling (FEM).
Comput Geotech 1999;25(4):191–204.
[43] Zhou YD, Cheuk CY, Tham LG. An embedded bond-slip model for finite element
modeling of soil–nail interaction. Comput Geotech 2009;36:1090–7.
[44] Zienkiewicz OC, Humpheson C, Lewis RW. Associated and non-associated visco-
plasticity and plasticity in soil mechanics. Geotechnique 1975;25(4):671–89.