You are on page 1of 21

Stability analysis of slopes reinforced with piles

E. Ausilio*, E. Conte, G. Dente
Dipartimento di Difesa del Suolo, Universita` della Calabria, 87036 Rende, Cosenza, Italy
Received 18 August 2000; received in revised form 22 March 2001; accepted 29 March 2001
Abstract
In this paper, the kinematic approach of limit analysis is used to analyse the stability of
earth slopes reinforced with piles. First, the case of slope without piles is considered and a
procedure is developed to calculate the safety factor for the slope. Results are compared with
those obtained using both the limit equilibrium method and more complex upper and lower
bound limit analysis solutions. Then, the stability of slopes reinforced with piles is analysed.
Expressions are derived allowing the force needed to increase the safety factor to a desired
value and the most suitable location of piles within the slope to be evaluated. A study is car-
ried out to illustrate the effect of piles on slope stability. # 2001 Elsevier Science Ltd. All
rights reserved.
1. Introduction
Slope stability can be increased in different ways such as: flattening of slopes by
modifying the ground surface geometry, carrying out surface and subsurface drai-
nage, using soil improvement techniques, installing continuous or discrete retaining
structures such as walls or piles. The first remedy leads to a reduction of the driving
forces for failure; the other measures in general produce an increase of the resisting
forces.
Piles have been used successfully in many situations in order to stabilise slopes or
to improve slope stability [1–9], and numerous methods have been developed for the
analysis of piled slopes.
The finite element method is certainly the most comprehensive approach to study
pile-slope stability, as this method simultaneously solves pile response and slope
Computers and Geotechnics 28 (2001) 591–611
www.elsevier.com/locate/compgeo
0266-352X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved.
PI I : S0266- 352X( 01) 00013- 1
* Corresponding author.
E-mail addresses:ausilio@dds.unical.it (E. Ausilio)
stability. However, its use generally requires high numerical costs and accurate
measurements of material properties. This makes the use of the method rather
unattractive for current applications. The finite element method has been recently
used by Cai and Ugai [10] to analyse the effect of piles on slope stability.
In practical applications, the study of a slope reinforced with piles is usually car-
ried out by extending the methods commonly used for the stability analysis of slopes
to incorporate the reaction force exerted on the unstable soil mass by the piles. To
date, the limit equilibrium method is the most widely used approach to analyse slope
stability due to its simplicity of use. Moreover, this method allows for the effect of
seepage, loading and general soil conditions without requiring additional computa-
tional efforts. Major criticisms of the limit equilibrium method are that it is generally
based on simplified assumptions, and the results obtained from this method are, in
the light of limit analysis, neither upper bounds nor lower bounds on the true solu-
tion [11].
The limit equilibrium method was used by Ito et al. [12] to deal with the problem
of the stability of slopes containing piles. In this approach the safety factor of the
piled slope was defined as the ratio of the resisting moment to the overturning moment
acting on the potentially unstable soil mass. The resisting moment consists of two
components: the moment due to soil shearing resistance along the sliding surface and
the moment provided by the reaction force fromthe piles. The driving moment and the
resisting moment due to soil shearing resistance were obtained applying the ordinary
slice method. To calculate the resisting moment due to the piles, Ito et al. [12] pro-
posed the use of the theoretical equation, derived previously by Ito and Matsui [3],
to evaluate the lateral force acting on a row of piles due to soil movement.
A similar approach was developed by Lee et al. [13] in which Bishop’s simplified
method [14] was employed to find the critical sliding surface for the slope as well as
the driving moment and resisting moment due to soil shearing resistance. The resist-
ing moment generated by the piles was obtained from the shear force and bending
moment developed in the pile at the depth of the sliding surface by the lateral soil
movement. These forces were calculated using a procedure based on the boundary
element method which was earlier proposed by Poulos [15] and later developed by
Lee et al. [16].
Recently, Hassiotis et al. [17] have extended the friction circle method to incor-
porate the pile reaction in slope stability analysis. The Ito and Matsui equation [3]
has been used to evaluate the lateral force that the failing soil mass exerts on a row
of piles.
The limit equilibrium method was also used by Chugh [18] and Poulos [8] to
analyse the stability of piled slopes. In both these approaches, it is assumed that the
piles provide an additional shear resistance along the critical sliding surface which
should increase the safety factor of the slope to a selected value.
In this paper, the stability of slopes reinforced with piles is analysed using the kine-
matic approach of limit analysis. The case of a slope without piles is first considered,
and a solution is proposed to determine the slope safety factor, which is here defined
as a reduction coefficient for the strength parameters of the soil. Then, the stability
of a slope containing piles is analysed. To account for the presence of the piles, it is
592 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
assumed that a lateral force and a moment are applied at the depth of the potential
sliding surface. Theoretical solutions are derived that allow the values of these forces
to be calculated. Moreover, conclusions are drawn regarding the most suitable
locations of the piles within the slope.
2. Method of analysis
Limit analysis takes advantages of the static and kinematic theorems of plasticity
theory to find the range in which the true solution of a stability problem falls. This
range can be narrowed finding the highest possible lower-bound solution and the
lowest possible upper-bound solution. The unknown quantity may be the bearing
capacity of a foundation, the earth pressure on a retaining wall, the safety factor or
critical height of a slope, etc. In limit analysis, soil is assumed to deform plastically
according to the normality rule associated with the Coulomb yield condition.
The static theorem considers stress fields which are in equilibrium with surface
tractions and body forces, and do not violate the yield criterion anywhere in the soil
mass (statically admissible stress field). Application of the static theorem leads to a
set of differential equations which may be solved numerically using the finite element
method [11,19].
To solve slope stability problems, use of limit analysis has almost exclusively
concentrated on the kinematic theorem [20–25], because under certain assumptions,
this is generally simpler to use than the static approach. For instance, when the
failing soil mass is assumed to move as a rigid body, the kinematic theorem neces-
sitates the solving of a simple equation.
Application of the kinematic theorem requires to equate the rate of work done by
tractions and body forces to the internal energy dissipation rate, for any assumed
strain rate field which is governed by the normality rule and is compatible with the
velocities at the boundary of the failing soil mass (kinematically admissible failure
mechanism). This can be expressed by the following work equation:
ð
S
T
i
v
i
dS þ
ð
V
X
i
v
i
dV ¼
ð
V
o
ij
c
.
ij
dV i. j ¼ 1. 2. 3 ð1Þ
where X
i
= body forces; T
i
= traction; v
i
= kinematically admissible velocity field;
c
.
ij
= strain rate field compatible with v
i
; o
ij
= stress field relating to X
i
and T
i
.
Moreover, S and V are, respectively, the loaded boundary and the volume of the
sliding soil mass. When the unknown quantity is a force that makes the soil mass
unstable, application of the kinematic theorem leads to an upper bound for the true
solution. On the contrary, this theorem yields a lower bound solution when a stabi-
lising force has to be determined.
In this study, the kinematic approach is employed to calculate the stabilising force
which must be provided by a retaining structure to increase the safety factor for a
slope of homogeneous soils to a selected value. For simplicity, the effect of pore-
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 593
water pressure on slope stability is not considered in the present work. However,
under drained loading conditions, this effect can be accounted for by expressing the
second term on the left-hand side of Eq. (1) as the rate of work done by the sub-
merged soil weight and that due to the seepage forces. In addition, the term o
ij
on
the right-hand side of (1) is the effective stress tensor.
The kinematic approach is first applied to analyse slope stability without retaining
structures.
3. Stability analysis of slopes without piles
In limit analysis, the solution of a slope stability problem is usually expressed
either in terms of the critical slope height [21] or a limit load applied on some por-
tion of the slope boundary [23]. If there is no boundary loading, collapse may be
caused by the weight of the soil itself. Thus, the limit condition has been also
expressed in terms of the unit weight of soil [11].
Slope stability analysis is traditionally formulated in terms of the safety factor
with respect to soil shearing strength parameters [26], which is analytically defined
as
FS ¼
c
c
m
¼
tg’
tg’
m
ð2Þ
where FS indicates the safety factor; c and ’ are the cohesion and the shearing
resistance angle of the soil, respectively; c
m
is the mobilized cohesion, and ’
m
is the
mobilized angle of shearing resistance. In other words, FS is defined as the factor by
which the soil shearing strength parameters should be divided to give the condition
of incipient failure. Karal [22] and Donald and Chen [25] accepted Eq. (2) as the
definition of the safety factor to analyse slope stability using the kinematic approach
of limit analysis. As pointed out by Karal [22], a direct consequence of Eq. (2) is
that, for frictional materials, the sliding surfaces are surfaces of potential yield, and
the displacements and the failure mechanism depend on the safety factor. This defi-
nition of FS is also adopted in the present study.
The kinematically admissible mechanism considered is shown in Fig. 1, where the
sliding surface is described by the log-spiral equation
r ¼ r
0
e
±À±
0
ð Þ
tg’
FS
ð3Þ
where r
0
=radius of the log-spiral with respect to angle ±
0
. The failing soil mass
rotates as a rigid body about the centre of rotation with angular velocity o
.
. This
mechanism, which was earlier considered by Chen [21], is geometrically defined by
angles [
0
, ±
0
, ±
h
(Fig. 1) and mobilized angle of shearing resistance
tg’
FS
. The slope
geometry is specified by height H, and angles o and [ which are also indicated in
Fig. 1.
594 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
The rate of external work is due to soil weight and surcharge boundary loads.
These two components of the external work rate are indicated in this study as W
.
and
Q
.
, respectively. The rate of work due to soil weight takes the form [21]
W
.
¼ , r
3
0
o
.
f
1
À f
2
À f
3
À f
4
½ Š ð4Þ
where ,=soil unit weight; functions f
1
–f
4
depend on the angles ±
0
, ±
h
, o, [ and [
0
,
and the mobilized angle of shearing resistance. Expressions for f
1
–f
4
can be found in
Chen [21]; for the sake of completeness they are also reported in the Appendix of
this paper. In deriving Eq. (4), it is assumed that the sliding surface passes below the
toe of the slope (Fig. 1). However, for the case in which the sliding surface passes
through the toe of the slope, the same expression for W
.
can be used provided f
4
=0
and [
0
=[.
When the slope is subjected to a surcharge boundary load, as shown in Fig. 1, the
rate of work done by this load is
Q
.
¼ q L o
.
r
0
cos ±
0
þ o ð Þ À
L
2
!
þ s L o
.
r
0
sin ±
0
þ o ð Þ ð5Þ
where L=distance between the failure surface at the top of the slope and the edge of
the slope (Fig. 1); q=applied normal traction; s=applied tangential traction.
For the rigid-block mechanism considered, the only energy dissipation takes place
along the sliding surface. The rate of energy dissipation, D
.
, can be written as [21]
D
.
¼
c r
2
0
o
.
2 tg’
e
2 ±
h
˱
0
ð Þ
tg’
FS
À 1
h i
ð6Þ
Fig. 1. Slope failure mechanism.
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 595
By equating the rate of external work to the rate of energy dissipation, we have
W
.
þ Q
.
¼ D
.
ð7Þ
and substituting the expressions for W
.
, Q
.
and D
.
into Eq. (7) yields
,
H
A
f
1
À f
2
À f
3
À f
4
ð Þ þ q B cos ±
0
þ o ð Þ À
B
2
!
þ s B sin ±
0
þ o ð Þ
¼
c
2 tg’
e
2 ±
h
˱
0
ð Þ
tg’
FS
À 1
h i
ð8Þ
where [21]
A ¼
sin[
0
sin [
0
À o ð Þ
sin ±
h
þ o ð Þe
±
h
˱
0
ð Þ
tg’
FS
À sin ±
0
þ o ð Þ
n o
ð9Þ
B ¼
sin ±
h
À ±
0
ð Þ
sin ±
h
þ o ð Þ
À
sin ±
h
þ [
0
ð Þ
sin ±
h
þ o ð Þ sin [
0
À o ð Þ
sin ±
h
þ o ð Þe
±
h
˱
0
ð Þ
tg’
FS
À sin ±
0
þ o ð Þ
n o
ð10Þ
The quantities A and B can be related to H and L, respectively, by the following
expressions
H ¼ A r
0
ð11aÞ
L ¼ B r
0
ð11bÞ
where distance L is indicated in Fig. 1.
For a given FS value, an upper bound for the slope height is obtained solving Eq.
(8), i.e.
H ¼
A
,
c
2 tg’
e
2 ±
h
˱
0
ð Þ
tg’
FS
À 1Š À q B cos ±
0
þ o ð Þ À
B
2
!
À s B sin ±
0
þ o ð Þ
f
1
À f
2
À f
3
À f
4
ð Þ
2
6
6
4
ð12Þ
The least upper bound for H can be found minimising the function H ¼
f ±
0
. ±
h
. [
0
ð Þ with respect to ±
0
, ±
h
and [
0
[21]. The angles thus obtained define the
potential sliding surface. In addition, substituting these angles into Eq. (12) yields
the critical height of the slope. This is the maximum height at which it is possible for
the slope to be stable with the assumed FS value. The true value of the safety factor
could be then found by an iterative procedure in which the resistance parameters of
596 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
the soil are progressively changed according to Eq. (2), until the critical height is
equal to the actual height of the slope.
Alternatively, the safety factor can be directly found by solving the following set
of equations
o H
o ±
0
¼ 0
o H
o ±
h
¼ 0
o H
o [
0
¼ 0
H ¼ H
actual
8
>
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
>
:
ð13Þ
where H
actual
denotes the actual slope height. In Eq. (13), the unknown quantities
are ±
0
, ±
h
, [
0
and FS. Therefore, the solution of Eq. (13) gives both the values of FS
and the position of the potential sliding surface.
Comparisons of the FS values derived from Eq. (13) to those obtained by other
authors using different methods are presented below.
Table 1 shows a comparison of the safety factor calculated by Eq. (13) and that
obtained by Cao and Zaman [27] using three different methods: Bishop’s method
Table 1
Comparison of slope safety factor calculated using different methods (adapted from Ref. [27])
b
(ratio)
c
(kPa)

(degree)
FS
analytical method
FS local
FS method
FS
Bishop’s method
FS
Eq. (13)
1:1 25 20 1.81 1.87 1.74 1.73
1:1 20 20 1.60 1.68 1.50 1.51
1:1 15 20 1.39 1.46 1.29 1.28
1:1 10 20 1.17 1.00 1.05 1.04
1:1 30 15 1.81 1.85 1.75 1.76
1:1 25 15 1.60 1.65 1.53 1.55
1:1 20 15 1.40 1.45 1.32 1.34
1:1 15 15 1.19 1.24 1.11 1.12
1:1 10 15 0.98 1.00 0.89 0.89
1:1 25 10 1.40 1.42 1.35 1.38
1:1 20 10 1.20 1.23 1.15 1.17
1:1 15 10 1.00 1.00 0.97 0.96
2:1 20 20 2.01 2.05 2.09 2.07
2:1 15 20 1.76 1.85 1.82 1.81
2:1 10 20 1.51 1.60 1.54 1.53
2:1 5 20 1.24 1.23 1.21 1.21
2:1 25 15 1.98 1.87 2.05 2.05
2.1 20 15 1.74 1.72 1.78 1.79
2:1 15 15 1.49 1.54 1.53 1.54
2:1 10 15 1.25 1.29 1.29 1.27
2:1 5 15 0.99 1.00 0.99 0.98
2:1 15 10 1.23 1.19 1.27 1.27
2:1 10 10 0.99 1.00 1.03 1.02
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 597
[14], the local minimum factor-of-safety method proposed by Huang and Yamasaki
[28], and the analytical method developed recently by Cao and Zaman [27]. As
shown in Table 1, safety factors derived from Eq. (13) are very close to the values
calculated using Bishop’s method, and are in good agreement with all the results
presented by Cao and Zaman [27].
Yu et al. [11] have recently presented rigorous upper and lower bound solutions
for the stability analysis of slopes. These solutions have been achieved using two
newly developed numerical procedures that are based on finite element formulations
of the bound theorems of limit analysis and linear programming techniques. For
comparison, Yu et al. [11] also applied Bishop’s limit-equilibrium method [14] to
calculate the slope safety factor. Results have been presented by Yu et al. [11] in
graphic form in terms of the stability number N
F
¼
, H FS
c
against the dimensionless
parameter l
c’
¼
, H tg’
c
earlier introduced by Janbu [29]. Figs. 2 and 3 show a com-
parison of the values of N
F
obtained in this study with those provided by Yu et al.
[11], for two values of slope angle [. As can be seen, Eq. (13) gives results that are
substantially in good agreement with those calculated using both the finite element
method and Bishop’s method. Moreover, it should be noted that the proposed upper
bound solution, based on a simple rigid-block mechanism, provides FS values that
are smaller than those obtained by Yu et al. [11] using a more complex upper bound
solution in which a failure mechanism including both rigid body motion and con-
tinuous deformation was considered.
Finally, a slope with H=13.7 m and [=30

is analysed as another example. Soil
properties are: c=23.94 KPa, ’=10

, and ,=19.63 KN/m
3
. This case was exam-
ined by Hassiotis et al. [17]; they calculated a FS value for the slope equal to 1.08.
Fig. 2. Comparison of stability number N
F
, for a slope with [=45

(adapted from Ref. [11]).
598 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
Hassiotis et al. [17] used the friction circle method. Applying Bishop’s method, Hull
and Poulos [30] found that the safety factor for the same slope is 1.12. The FS value
obtained solving Eq. (13) is 1.11, midway among those calculated by the other
authors. The potential sliding surfaces found applying the three different methodol-
ogies are shown in Fig. 4. Their positions are consistent with the corresponding FS
values.
4. Stability analysis of slopes reinforced with piles
When the safety factor for a slope is considered to be inadequate, slope stability
may be improved installing a retaining structure such as a row of piles (Fig. 5). The
piles should be designed to provide the stabilising force needed to increase the safety
factor to a selected value.
In this section, the kinematic approach is applied to assess the additional force that the
piles must provide to increase slope stability. To account for the presence of the piles, a
Fig. 3. Comparison of stability number N
F
, for a slope with [=60

(adapted from Ref. [11]).
Fig. 4. The critical sliding surfaces found by Hassiotis et al. [17], Hull and Poulos [30], and using Eq. (13).
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 599
lateral force and a moment are assumed to be applied at the depth of the potentially
sliding surface. Under this assumption, the rate of energy dissipation becomes
D
.
¼
c r
2
0
o
.
2 tg’
e
2 ±
h
˱
0
ð Þ
tg’
FS
À 1
h i
þ F r
0
sin±
F
o
.
e
±
F
˱
0
ð Þ
tg’
FS
À M o
.
ð14Þ
in which FS is the target safety factor of the slope; angle ±
F
specifies the position of
the retaining structure along the sliding surface (Fig. 5); F is the stabilising force,
per unit width of soil, which the piles have to provide to improve slope stability;
moment M accounts for F distribution with depth in the portion of the piles above
the sliding surface, it is given by
M ¼ F m h ð15Þ
where h is the height of the portion of the piles above the sliding surface (Fig. 5), and
m is a coefficient less than unity. For instance, if F is assumed to be linearly dis-
tributed between the ground surface and the sliding surface, m is set equal to 1/3.
When m=0, the presence of the piles on slope stability is expressed by an additional
shearing resistance along the potential sliding surface, as assumed also by Poulos [8].
Height h can be calculated using one of the following expressions according to the
abscissax
F
which is measured from the slope toe (Fig. 5):
h ¼ r
F
sin±
F
À r
h
sin±
h
if À D4x
F
- 0 ð16aÞ
Fig. 5. Piled slope stability problem.
600 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
h ¼ r
F
sin±
F
À r
h
sin±
h
þ x
F
tg[ if 04x
F
4H ctg[ ð16bÞ
h ¼ r
F
sin±
F
À r
h
sin±
h
þ H þ x
F
À H ctg[ ð Þtgo if x
F
> H ctg[ ð16cÞ
where
x
F
¼ r
F
cos0
F
À r
h
cos±
h
À D ð17aÞ
D ¼
sin [ À [
0
ð Þ
sin[ sin[
0
H ð17bÞ
r
F
¼
H
A
e
±
F
˱
0
ð Þ
tg’
FS
and r
h
¼
H
A
e
±
h
˱
0
ð Þ
tg’
FS
ð17cÞ
For a selected FS value, h is a function of angles ±
0
, ±
h
, [
0
and ±
F
.
The rate of external work is given again by the sum of W
.
and Q
.
. These latter are
expressed by Eqs. (4) and (5), respectively. Therefore, equating the rate of external
work to the rate of energy dissipation leads to the following expression for F:
F ¼
, H
A
f
1
Àf
2
Àf
3
Àf
4
ð Þ þq B cos ±
0
þo ð Þ À
B
2
!
þs B sin ±
0
þo ð Þ À
c
2 tg’
e
2 ±
h
˱
0
ð Þ
tg’
FS
À1
h i
A
H
sin±
F
e
±
F
˱
0
ð Þ
tg’
FS
Àmh
A
H
!
ð18Þ
Eq. (18) gives the force, per unit width of soil, which must be provided by a
retaining structure to achieve the desired value of the safety factor of the slope. If the
retaining structure consists of a row of piles, the lateral force acting on each pile may
be obtained in an approximate manner multiplying F by the centre to centre spacing
between the piles. To evaluate more suitably the force acting on the piles, arching
between adjacent piles should be considered.
When a retaining structure is inserted in a slope, the additional resistance pro-
vided by this structure changes both the slope safety factor and potential failure
mechanism with respect to the case without piles. As a consequence, other possible
sliding surfaces could be more critical than the one found for the slope without piles.
The most critical surface is that for which the highest F value is required to increase
the safety factor to the desired value. From the computational point of view, this
surface can be found maximising function F ¼ F ±
0
. ±
h
. ±
F
. [
0
ð Þ with respect to angles
±
0
, ±
h
and [
0
under the condition that the position of piles within the slope is given.
To this end, the following set of equations has to be solved
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 601
o F
o ±
0
¼ 0
o F
o ±
h
¼ 0
o F
o [
0
¼ 0
x
F
¼
H
A
cos±
F
e
±
F
˱
o
ð Þ
tg’
FS
Àcos±
h
e
±
h
˱
o
ð Þ
tg’
FS
h i
ÀH
sin [À[
0
ð Þ
sin[ sin[
0
8
>
>
>
>
>
>
>
>
>
>
>
<
>
>
>
>
>
>
>
>
>
>
>
:
ð19Þ
where x
F
specifies the position of the piles with respect to the toe of the slope (Fig. 5).
The value of x
F
should be assumed keeping in mind the critical sliding surface found
for the slope without piles. This surface indicates, in fact, the range of positions
where piles have to be placed for increasing effectively slope stability. A retaining
structure outside the region of soil affected by this sliding surface could have no
influence on slope stability [30].
In Eq. (19), the unknown quantities are ±
0
, ±
h
, [
0
and ±
F
. Angles ±
0
, ±
h
, and [
0
specify the critical potential sliding surface, and the maximum F value is calculated
substituting these angles into Eq. (18). However, it should be noted that, if m is
assumed not to be zero, F depends on height h which can be determined from ±
0
, ±
h
,
±
F
and [
0
using Eqs. (16) and (17). This implies that Eq. (19) has to be solved con-
sidering the expression for h relevant with the assumed value of x
F
, according to
Eqs. 17.
Once force F is obtained, pile geometry, centre-to-centre distance at which the
piles have to be placed, and structural requirements for the piles can be determined
from a pile-soil interaction analysis [1,4,12,16,17,31–36]. Maximum displacement,
shear and bending moments acting on the piles should be considered to assure
that the design is adequate. This matter is however outside the scope of the present
work.
The outlined approach is illustrated considering the same slope shown in Fig. 4, as
an example. The safety factor for this slope without pile reinforcement is 1.11. The
critical sliding surface is also indicated in Fig. 4. Since a safety factor of 1.11 is
considered inadequate, the slope may be reinforced installing a row of piles to
increase the safety factor to a selected value. For this example, it is assumed that the
required safety factor is 1.50. The piles are assumed to be located at x
F
¼13.7 m.
The stabilising force, for unit width of soil, which has to be provided by the piles to
increase slope stability, is evaluated using Eqs. (18) and (19) in which m is set equal
to 1/3. In the case examined, this force is equal to 515 kN/m, and is assumed to be
linearly distributed between the ground surface and the sliding surface. Moreover,
from Eq. (16) the height of the portion of the pile above the sliding surface is
h=12.7 m. Therefore, the total length of the piles may be preliminarily assumed as
L
p
%2 h=25 m [8]. The potential sliding surface for the slope without piles and that
for the piled slope are shown in Fig. 6. As can be noted, the sliding surface for the
slope reinforced with piles is deeper and passes beneath the toe of the slope. The
writers have found that this occurs generally for low values of the soil shearing
602 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
resistance angle or when the achievement of a high FS value is required, especially
when the slope is gentle.
5. Results
As can be noted from Eq. (18), force F depends on the position of the piles within
the slope which is specified by angle ±
F
or equivalently by abscissa x
F
. The most
suitable position for the piles is that where the piles are most effective for improving
slope stability.
Many studies have been conducted in order to establish the optimal location of
the piles within a slope. However, the results obtained are rather different, and in
some cases even contrasting.
Ito et al. [12] showed that the maximum effect of piles on slope stability is when
they are placed in the upper-middle part of the slope. Hassiotis et al. [17] arrived at
similar conclusions. According to these latter authors, the piles should be located
close to the top of the slope to achieve the maximum safety factor, especially when
the slope is steep. Lee et al. [13] analysed the case of a purely cohesive soil slope.
They found that when the piles are installed into a homogeneous soil the most
effective pile positions are the toe and crest of the slope. By contrast, the piles have
little effect on stability when they are located close to the middle of the slope. For a
two-layered soil slope where the upper soft layer is underlain by a stiff layer, Lee et
al. [13] showed that the piles are more effective when installed between the middle
and the crest of the slope. However, if the soil profile is reversed, according to Lee et
al. [13] the most effective positions for the piles are again the toe and the crest of the
slope. Recently, Cai and Ugai [10], using the finite element method, have pointed
out that the piles should be located in the middle of the slope to achieve the max-
imum safety factor for the slope. The same authors have also applied a modified
version of Bishop’s method in which the reaction force from the piles is expressed by
Ito–Matsui’s equation [3]. Using this approach, Cai and Ugai [10] have found that
the piles have to be installed closer to the top of the slope to give the best result.
In order to illustrate the effect of the pile position on slope stability, the example
shown in Fig. 7 is considered. Soil strength parameters are assumed to be: c=4.7
Fig. 6. Critical sliding surface for the slope without piles and for the piled slope.
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 603
kPa and ’=25

; unit weight is 20 kN/m
3
. For this slope, the safety factor is equal to
1.19 as derived from Eq. (13). The optimal position of the piles within the slope is
determined when the stabilising force needed to increase the safety factor to the
desired value takes the minimum value. Assuming that the pile position varies
between the base and the top of the slope, the force provided by the piles is calcu-
lated using Eqs. (18) and (19), and is plotted against the dimensionless abscissa
x
F
L
x
,
where L
x
¼ H ctg[ (Fig. 7). It should be noted that when
x
F
L
x
¼ 0 or
x
F
L
x
¼ 1 pile
position is the toe or the crest of the slope, respectively. Moreover, values of
x
F
L
x
greater than unity indicate that the piles are located at the top of the slope. The
stabilising force is expressed in a dimensionless form as K ¼
F
1,2 , H
2
. Coefficient m is
Fig. 7. Illustrative example of a slope reinforced with piles.
Fig. 8. Effect of pile location on dimensionless force K, when m=0.
604 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
assumed to be 0, 1/3 or 1/2. The results are given in Figs. 8–10 for three different
values of the improvement ratio j which is defined as
j ¼
FS
FS
0
ð20Þ
where FS=safety factor of the piled slope problem, and FS
0
=safety factor of the
slope without piles.
As can be expected, K increases with increasing j. The increase in K is greater
when m=1/2, although the results appear not to be greatly affected by the value of
m. In all the cases examined, the optimal location of the piles is near the toe of the
slope, where the force provided by the piles to achieve the selected value of the
improvement ratio takes the lowest value. This is due to the shape of the sliding
surface which is a log-spiral curve having a radius that increases as the surface
develops from the top to the base of the slope. For a rotational failure mechanism as
shown in Fig. 5, the required stabilising moment due to F, with respect to the rota-
tion centre, has an arm that increases as the location of the piles approaches to the
slope toe, and consequently force F decreases. However, Figs. 8–10 show that the
piles are also very effective when they are located between the middle and the toe of
the slope, especially when m is assumed to be zero. The figures also show that the
region where the piles are more effective reduces as j increases, and is located closer
to the toe of the slope. Therefore, when the achievement of a high improvement
ratio value is required, the piles should be located with greater care within the slope.
Fig. 9. Effect of pile location on dimensionless force K, when m=1/3.
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 605
Fig. 10. Effect of pile location on dimensionless force K, when m=1/2.
Fig. 11. Safety factor for the slope without piles against l
c’
.
606 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
In the following, the results of a parametric study are also presented in order to
illustrate the effect of several factors on the stabilising force that the piles have to
provide to increase slope stability. The calculations have been carried out with three
values of the slope angle, [=30, 45 and 60

. The results are given for different values
of j which are chosen to adequately improve slope stability. In all the calculations, it
is assumed that m is equal to 1/3.
The slope safety factor without piles, FS
0
, can be determined from the results
presented in Fig. 11 which have been obtained using Eq. (13). As can be noted, the
values of FS
0
decrease when increasing the slope angle and increasing the parameter
l
c’
that has been defined in a previous section. It should be noted that the value of
l
c’
also indicates the position of the potential sliding surface within the slope. As
pointed out by Duncan and Wright [37], when l
c’
is small the sliding surface
becomes deeper and expands into the soil, especially when the slope is gentle. By
contrast, as the value of l
c’
increases, the critical sliding surface becomes increas-
ingly shallow. Duncan and Wright [37] considered sliding surfaces of circular shape.
However, this also occurs when log-spiral sliding surfaces are considered, as shown
in Fig. 12.
Fig. 12. Critical sliding surfaces of a slope with [=30

for different values of l
c’
.
Fig. 13. Force K against l
c’
for a slope with [=30

.
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 607
The piles are assumed to be located at the middle of the slope which should be a
suitable location for the piles. Following Poulos [8], a retaining structure which is
located near the toe or the crest of the slope could restrain only a small mass of the
soil, while a lot of the soil mass behind or in front of the structure could be unstable.
Moreover, this location is consistent with the assumption that the piles are laterally
loaded. At the crest of the slope, the axial response could be more important than
the lateral one, because of soil movement is here predominantly down the piles and
not laterally across them [30].
Fig. 14. Force K against l
c’
for a slope with [=45

.
Fig. 15. Force K against l
c’
for a slope with [=60

.
608 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
The results are presented in Figs. 13–15 in terms of K against Janbu’s dimension-
less parameter l
c’
. These figures show that the stabilising force provided by the piles
are affected by both j and l
c’
. As previously observed, K strongly increases as the
improvement ratio increases. On the contrary, the stabilising force decreases with
Janbu’s parameter, and the lowest values of K occur when l
c’
is high. This implies
that the use of piles is a very effective measure to increase slope stability especially
when the sliding surface for the slope without piles is shallow. In this case, in fact,
the piles have to provide a smaller force to achieve the desired value of the
improvement ratio.
6. Conclusions
A kinematic approach has been described for the stability analysis of slopes rein-
forced with piles. The first step of the approach consists in finding the critical sliding
surface and the safety factor for the slope without piles. To this purpose, a proce-
dure has been developed in which the solution is expressed in terms of the safety
factor that is defined as a reduction coefficient for the shearing resistance parameters
of the soil. The results obtained using the proposed procedure are found to be in
good agreement with those derived from both Bishop’s method and more complex
upper and lower bound solutions of limit analysis.
For slopes containing piles, analytical expressions have been derived that allow
the force needed to increase the safety factor to a desired value, and the most sui-
table location of the piles within the slope to be evaluated. These expressions may be
found useful for designing piles to reinforce slopes. The calculations carried out
using the expressions obtained show that installing a row a piles is an effective
remedy to improve slope stability especially when the sliding surface for the unrein-
forced slope is relatively shallow. The results also indicate that the optimal location
of the piles within the slope is near the toe of the slope where the stabilising force
needed to increase the safety factor to the desired value takes a minimum value. Piles
appear also to be very effective when they are installed in the region from the middle
to the toe of the slope. However, this region reduces when the achievement of high
safety factor values is required.
Appendix
f
1
¼
3tg’
Ã
cos
h
þ sin
h
ð Þ exp 3
h
À
0
ð Þ tg’
Ã
½ Š À 3 tg’
Ã
cos
0
À sin
0
È É
3 1 þ 9 tg
2

Ã
ð Þ
f
2
¼
1
6
L
r
o
2cos
0
À
L
r
o
cos

sin
0
þ ð Þ
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 609
f
3
¼
1
6
exp
h
À
0
ð Þtg’
Ã
½ Š sin
h
À
0
ð Þ À
L
r
0
sin
h
þ ð Þ
!
 cos
0
À
L
r
0
cosþ cos
h
exp
h
À
0
ð Þtg’
Ã
½ Š
& '
f
4
¼
H
2
r
2
0
sin À
0
ð Þ
2sinsin
0
cos
0
À
L
r
0
cosÀ
1
3
H
r
0
cotg
0
þ cotg ð Þ
!
In these expressions, the quantities
H
r
0
and
L
r
0
are given by Eqs. (11a) and (11b),
respectively, and tg’
Ã
¼
tg’
FS
.
References
[1] De Beer E, Wallays M. Stabilization of a slope in schist by means of bored piles reinforced with steel
beams. In: Proc. 2th International Congress on Rock Mechanics, Beograd, 1970. p. 361–9.
[2] Esu F, D’Elia B. Interazione terreno-struttura in un palo sollecitato da una frana tipo colata. Rivista
Italiana di Geotecnica 1974;8(1):27–38.
[3] Ito T, Matsui T. Methods to estimate lateral force acting on stabilizing piles. Soils and Foundations
1975;15(4):43–59.
[4] Fukuoka M. The effects of horizontal loads on piles due to landslides. In: Proc. 9th International
Conference on Soil Mechanics and Foundation Engineering, Tokyo, 1977. p. 27–42.
[5] Sommer H. Creeping slope in a stiff clay. In: Proc. 9th International Conference on Soil Mechanics
and Foundation Engineering, Tokyo, 1977. p. 113–8.
[6] Gudehus G, Schwarz W. Stabilisation of creeping slopes by dowels. In: Proc. 11th International
Conference on Soil Mechanics and Foundation Engineering, San Francisco, 1985. p. 1697–700.
[7] Carruba P, Maugeri M, Motta E. Esperienze in vera grandezza sul comportamento di pali per la
stabilizzazione di un pendio. In: XVII Convegno Nazionale di Geotecnica, Taormina, 1989. p. 81–90.
[8] Poulos HG. Design of reinforcing piles to increase slope stability. Canadian Geotechnical Journal
1995;32(5):808–18.
[9] Hong WP, Han JG. The behavior of stabilizing piles installed in slopes. In: Proc. 7th International
Symposium on Landslides, Rotterdam, 1996. p. 1709–14.
[10] Cai F, Ugai K. Numerical analysis of the stability of a slope reinforced with piles. Soils and Foun-
dations 2000;40(1):73–84.
[11] Yu HS, Salgado R, Sloan SW, Kim JM. Limit analysis versus limit equilibrium for slope stability.
Journal of Geotechnical and Geoenvironmental Engineering, ASCE 1998;124(1):1–11.
[12] Ito T, Matsui T, Hong WP. Design method for the stability analysis of the slope with landing pier.
Soils and Foundations 1979;19(4):43–57.
[13] Lee CY, Hull TS, Poulos HG. Simplified pile-slope stability analysis. Computers and Geotechnics
1995;17:1–16.
[14] Bishop AW. The use of slip circle in the stability analysis of earth slopes. Geotechnique 1955;5(1):7–17.
[15] Poulos HG. Analysis of piles in soil undergoing lateral movement. Journal of Soil Mechanics and
Foundation Division, ASCE 1973;99(SM5):391–406.
[16] Lee CY, Poulos HG, Hull TS. Effect of seafloor instability on offshore pile foundations. Canadian
Geotechnical Journal 1991;28(5):729–37.
[17] Hassiotis S, Chameau JL, Gunaratne M. Design method for stabilization of slopes with piles. Jour-
nal of Geotechnical and Geoenvironmental Engineering, ASCE 1997;123(4):314–23.
[18] Chugh AK. Procedure for design of restraining structures for slope stabilization problems. Geo-
technical Engineering 1982;13:223–34.
610 E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611
[19] Lysmer J. Limit analysis of plane problems in soil mechanics. Journal of Soil Mechanics and Foun-
dation Division, ASCE 1970;96(SM4):1131–334.
[20] Finn W. Application of limit plasticity in soil mechanics. Journal of Soil Mechanics and Foundation
Division, ASCE 1967;89(SM5):101–19.
[21] Chen WF. Limit analysis and soil plasticity. Amsterdam (The Netherlands): Elsevier Science, 1975.
[22] Karal K. Energy method for soil stability analyses. Journal of the Geotechnical Engineering Divi-
sion, ASCE 1977;103(GT5):431–45.
[23] Michalowski RL. Three dimensional analysis of locally loaded slopes. Geotechnique 1989;39(1):27–
38.
[24] Michalowski RL. Slope stability analysis: a kinematical approach. Geotechnique 1995;45(2):283–93.
[25] Donald IB, Chen Z. Slope stability analysis by the upper bound approach: fundamentals and meth-
ods. Canadian Geotechnical Journal 1997;34:853–62.
[26] Taylor DW. Fundamentals of soil mechanics. New York: John Wiley and Sons, 1948.
[27] Cao J, Zaman MM. Analytical method for analysis of slope stability. International Journal for
Numerical and Analytical Methods in Geomechanics 1999;23:439–49.
[28] Huang SL, Yamasaki K. Slope failure analysis using local minimum factor-of-safety approach.
Journal of Geotechnical Engineering, ASCE 1993;119(12):1974–87.
[29] Janbu N. Stability analysis of slopes with dimensionless parameters. Harvard Soil Mechanics Series
no. 46, 1954.
[30] Hull TS, Poulos HG. Design method for stabilization of slopes with piles (discussion). Journal of
Geotechnical and Geoenvironmental Engineering, ASCE 1999;125(10):911–3.
[31] De Beer E. Piles subjected to static lateral loads. In: Proc. 9th International Conference on Soil
Mechanics and Foundation Engineering, Tokyo, 1977. p. 1–14.
[32] Viggiani C. Ultimate lateral load on piles used to stabilise landslides. In: Proc. 10th International
Conference on Soil Mechanics and Foundation Engineering, Stockholm, 1981. p. 555–60.
[33] Ito T, Matsui T, Hong WP. Design method for stabilizing piles against landslide — one row of piles.
Soils and Foundations 1981;21(1):21–37.
[34] Ito T, Matsui T, Hong WP. Extended design method for multi-row stabilizing piles against landslide.
Soils and Foundations 1982;22(1):1–13.
[35] Maugeri M, Motta E. Stresses on piles used to stabilize landslides. In: Proc. 6th International Sym-
posium on Landslides, Christchurch, 1992. p. 785–90.
[36] Chen LT, Poulos HG. Piles subjected to lateral soil movements. Journal of Geotechnical and
Geoenvironmental Engineering, ASCE 1997;123(9):802–11.
[37] Duncan JM, Wrigth SG. The accuracy of equilibrium methods of slope stability analysis. Engineer-
ing Geology 1980;16:5–17.
E. Ausilio et al. / Computers and Geotechnics 28 (2001) 591–611 611