You are on page 1of 18

METHOD TO ESTIMATE THE SYSTEM PROBABILITY OF FAILURE OF

MECHANICALLY STABILIZED EARTH (MSE) WALLS
Tanit Chalermyanont
1
and Craig H. Benson
2
Members ASCE
Abstract: This paper describes a technique to calculate the system probability of failure (or
system reliability) of MSE walls. The system probability of failure is defined as the
complement of the sum of the probabilities of failure corresponding to internal failure and
external failure (sliding, overturning, and bearing capacity failure). The reliability is the
complement of the system probability of failure. Probability of internal failure is obtained
from analysis of MSE walls using design charts. Probabilities for each external failure
mode are calculated using a mean factor of safety computed with conventional design
equations and mean values for the soil properties as input, and a standard deviation of the
factor of safety computed using an empirical equation. The empirical equation relates the
standard deviation of the factor of safety for each external failure mode to uncertainty in the
design parameters. Analyses of three existing walls suggest that conventionally designed
MSE walls have a system probability of failure 0.0003, which is one to two orders of
magnitude lower than the failure rate for similar geotechnical structures.
INTRODUCTION
The impact of uncertainties in soil properties, reinforcement properties, and loads is
generally accounted for in geotechnical design using a factor of safety approach. An
alternative approach is to use a probabilistic analysis, where uncertainties in the design
parameters are considered in a mathematical framework. The advantage of the probabilistic
approach is a direct linkage between uncertainty in the design parameters and the risk
(probability) of failure (Whitman 2000, Phoon et al. 2003, Christian 2004).
Chalermyanont and Benson (2004, 2005) developed a probabilistic method for design
of MSE walls that consists of selecting the tensile strength and length of the reinforcement
to achieve the target probability of failure for internal (failure within the reinforced zone)
and external (sliding, overturning, and bearing capacity) failure modes. Tensile strength of
the reinforcement is selected based on the internal analysis, whereas length of the
reinforcement is selected as the maximum of lengths obtained from internal and external
analyses. This paper presents a method to determine the system probability of failure of an
1
Lecturer, Dept. of Civil Engineering, Faculty of Engineering, Prince of Songkla University, Songkla,
90112,Thailand, chtanit@ratree.psu.ac.th
2
Professor, Dept. of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison, WI
53706, USA, benson@engr.wisc.edu
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
MSE wall (or the complement, the system reliability). The method is based on walls having
a length-to-height ratio (L/H) of 0.7, which is commonly used in design (Ho and Rowe
1997).
BACKGROUND
A variety of methods can be used to determine the probability of failure (P
f
) for
geotechnical problems. Approximate analytical techniques, such as the first order second
moment or point estimate methods, can be used when relatively simple analytical
expressions describe the factor of safety (FS) and the distribution of the FS is
approximately normal. For more complicated problems, the Monte Carlo method is often
used to determine the distribution of the FS and P
f
(Harr 1987). In all cases, P
f
is related to
the geometry of the problem as well as statistical parameters describing variability in the
loads and properties of the soil and reinforcement.
Chalermyanont and Benson (2004, 2005) used Monte Carlo simulation to develop a
probabilistic design method for internal and external stability of MSE walls backfilled with
and founded on cohesionless soil. A schematic of the wall considered in their analyses is
shown in Fig. 1. Walls having different height (H), length of reinforcement (L), vertical
spacing of reinforcement (V
S
), backfill friction angle (|), backfill unit weight (¸), friction
angle of foundation soil (|
f
), unit weight of foundation soil (¸
f
), surcharge (q), and tensile
strength of reinforcement (T) were considered. Probability distributions were used to
describe spatial variability in properties of the soil and reinforcement. The Monte Carlo
simulation process consisted of a series of realizations where the soil and reinforcement
properties were randomly sampled from their probability distributions and the FS was
calculated for each realization using conventional methods [e.g., Bishop’s simplified
method for internal stability or Elias-Christopher method for external stability (Siegel 1975,
Elias and Christopher 1997)]. P
f
was calculated from the distribution of FS obtained from
the realizations; i.e., P
f
= P(FS<1). The methods used by Chalermyanont and Benson
(2004, 2005) were used in this study as well and are summarized in the following sections.
Internal Stability Analysis
A schematic showing how MSE walls were analyzed for internal stability is shown
schematically in Fig. 2. The reinforcement length to wall height ratio (L/H) was set at 0.7,
which is typical for MSE walls (AASHTO 1992, Ho and Rowe 1997). The backfill and
retained soil were assumed to be the same, to be cohesionless, and to have a mean friction
angle (µ
|
) ranging from 25° to 40° and a mean unit weight (µ
¸
) ranging from 16 to 21
kN/m
3
. The foundation was assumed to be a strong and homogeneous cohesionless soil
having a friction angle of 40° and a unit weight of 19.5 kN/m
3
so that failure would not
occur through the foundation.
Friction angle and unit weight of the backfill were assumed to be normally distributed
(parameterized with a mean, µ, and coefficient of variation, COV) as has been shown by
others (Lumb 1966, Hoeg and Murarka 1974, Lacasse and Nadim 1996, Basheer and Najjar
1996, Low and Tang 1997, Phoon and Kulhawy 1999). Friction angle and unit weight of
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Fig. 1. Schematic of MSE wall c onsidered in study.
Note: W
i
= weight of each slice, F
si
= interface friction along slip surface
Fig. 2. Forces and geometry for internal stability analysis.
H
10 m
Foundation soil ( µ µµ µ
| || |f
, COV
| || |f
, µ µµ µ
¸ ¸¸ ¸f
, COV
¸ ¸¸ ¸f
)
L
V
S
Reinforcement (µ µµ µ
T
, COV
T
)
Backfill (µ µµ µ
| || |
, COV
| || |
, µ µµ µ
¸ ¸¸ ¸
, COV
¸ ¸¸ ¸
) )) )
Retained soil
Wall facing
Ground surface
Surcharge
H
10 m
Foundation soil ( µ µµ µ
| || |f
, COV
| || |f
, µ µµ µ
¸ ¸¸ ¸f
, COV
¸ ¸¸ ¸f
)
L
V
S
Reinforcement (µ µµ µ
T
, COV
T
)
Backfill (µ µµ µ
| || |
, COV
| || |
, µ µµ µ
¸ ¸¸ ¸
, COV
¸ ¸¸ ¸
) )) )
Retained soil
Wall facing
Ground surface
Surcharge
H
Surcharge
L
e
Circular Slip Surface
W
i
F
si
F
i
i
m
T
H
Surcharge
L
e
Circular Slip Surface
W
i
F
si
F
i
i
m
T
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
the backfill were assumed to vary randomly from lift-to-lift, and all lifts were assumed to be
0.3 m thick (based on Elias and Christopher 1997). Tensile strength of the reinforcement
and the surcharge load were assumed to be normally distributed based on distributions
characteristic of structural systems (e.g., Alpsten 1972, Pier and Cornell 1973).
The slope stability approach was used to calculate the FS for internal stability using
Bishop’s simplified method (Fig. 2). The mobilized resisting force due to the reinforcement
(T
m
) was set as the smaller of the force developed by the soil-reinforcement interface
friction (F) and the tensile strength of the reinforcement (T). Only the effective
reinforcement length behind the slip surface was used to calculate the force developed by
the soil-reinforcement friction (Broms 1978, Bonaparte et al. 1987, Leschinsky and Perry
1987, Schmertmann et al. 1987). Resisting forces generated by the reinforcement were
assumed to act horizontally. Forces generated at the soil-reinforcement interface were
calculated as:
F = 2o
v
L
e
tano (1)
where o
v
is the vertical stress, L
e
is the effective length behind the slip surface, and o is the
soil-reinforcement interface friction angle.
External Stability Analyses
The MSE wall model used for external stability analyses is shown in Fig. 3. Spatial
variability of soil properties within the wall backfill and retained soil was simulated by
dividing the soil into a series of cells. Soil properties were assumed to be uniform within a
cell and independent of those in surrounding cells. Sizes of the cells in the backfill and the
foundation soil were varied based on their correlation lengths. The size of the cells in the
vertical direction (AH) of the backfill was set at 0.3 m based on the typical thickness of a
lift (Elias and Christopher 1997). The size of cells in horizontal direction (AL) of the
backfill was set at 1 m, based on correlation lengths typical for compacted soil (Benson
1991). Sizes of the cells of the foundation soil in vertical direction (AD
f
) and horizontal
direction (AL
f
) were set at 1 m and 20 m, respectively, based on correlation lengths of
natural soils reported by Cherubini (2000).
Factors of safety corresponding to sliding, overturning, and bearing capacity modes
were computed using classical analyses for gravity-type wall systems (Chalermyanont
2002). The MSE wall is treated as a rigid mass subjected to a body force and an active
thrust (Fig. 4). Factors of safety for sliding (FS
S
), overturning (FS
O
), and bearing capacity
(FS
BC
) are computed as:
a
t
S
P
S
FS = (2)
H P 2
L W 3
FS
a
t
O
= (3)
( ) e 2 L / W
q
FS
t
ult
BC
÷
= (4)
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Fig. 3. MSE wall model showing cells in backfill, retained soil, and foundation.
Fig. 4. Forces and stresses on an MSE wall for overturning, bearing capacity and sliding.
H
L
L
f
D
f
W
t
P
a
S
t
q
ult
45°-|/2 45°-|/2
i
ii
iii
H
L
L
f
D
f
W
t
P
a
S
t
q
ult
45°-|/2 45°-|/2
i
ii
iii
1
2
m-1
m
1 2 n-1 n
H
L
AH
AL
i
j
L
f
D
f
AL
f
AD
f
10 m
n+1
Backfill Retained soil
Foundation Soil
1
2
m-1
m
1 2 n-1 n
H
L
AH
AL
i
j
L
f
D
f
AL
f
AD
f
10 m
n+1
Backfill Retained soil
Foundation Soil
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
where W
t
= total weight of the backfill, P
a
= total active thrust, S
t
= total shear resistance,
q
ult
= ultimate bearing capacity of the foundation soil, and e = eccentricity due to active
thrust = P
a
H/3W
t
(Fig. 4). The weight, W
t
, is computed as:
( )
¿¿
= =
A A ¸ =
n
1 i
m
1 j
j , i t
L H W (5)
where i and j are counters, n and m are the number cells in the vertical and horizontal
directions, and ¸
i,j
is the unit weight of cell (i,j). The shear force along the base, S
t
, is
computed as:
( )
¿¿
= =
| A A ¸ =
n
1 i
m
1 j
m , i j , i t
tan L H S (6)
where |
i,m
is the friction angle of the backfill at the backfill-foundation soil interface. The
active thrust, P
a
, is computed as the sum of the thrust applied by each of j elements along
the back of backfill soil (P
j
):
( ) ( )
¿ ¿
=
÷ ÷
=
(
¸
(

¸

A ÷ + A = =
m
1 j
1 j j 1 j
m
1 j
j a
H p p
2
1
H p P P (7)
where
j
a j , 1 n 1 j j
HK p p A ¸ + =
+ ÷
. The coefficient of active earth pressure, K
aj
, is computed
using Rankine theory with friction angles corresponding to the cells next to the backfill soil
mass (i.e., cells in column n+1).
Meyerhof’s bearing capacity equation (Meyerhof 1963) was used to calculate the
ultimate bearing capacity of the foundation using the average friction angle (|
fave
) and unit
weight (¸
fave
) of the foundation soil over an “effective area.” Ultimate bearing capacity of
the foundation was calculated as:
¸
÷ ¸ = N ) e 2 L ( 5 . 0 q
fave ult
(8)
where ( ) ( )
fave q
4 . 1 tan 1 N N | ÷ =
¸
and ( ) |
.
|

\
| |
+ | t =
2
45 tan tan exp N
fave 2
fave q
. The effective
area was defined by a depth (D
f
) and breadth (L + L
f
) (Fig. 4) where:
( )
f f
tan 90
o f
e r D
| |
µ µ ÷ °
= (9a)
and
(
(
¸
(

¸

|
|
.
|

\
| µ
÷ ° =
|
µ
|
|
.
|

\
| µ
÷ °
|
|
2
45 cos e r 2 L
f
tan
2
135
o f
f
f
(9b)
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
In Eq. 9b, r
o
= L / (2cosµ
|f
).
SYSTEM RELIABILITY ANALYSIS
The system probability of failure (P
fSYS
) has four potential modes of failure: internal
failure, sliding failure, overturning failure, and bearing capacity failure. These modes of
failure are assumed to be independent, with the MSE wall failing if any of the modes of
failure occurs. Thus, P
fSYS
is calculated as:
fBC fO fS fINT fSYS
P P P P P + + + = (10)
where the subscripts INT, S, O, and BC refer to internal stability, sliding, overturning, and
bearing capacity, respectively. By definition, R = 1- P
fSYS
. Accordingly, the P
f
for each
failure mode must be determined to compute P
fSYS
or R.
The P
f
for each of the failure modes can be determined by interpolating between curves
on the design charts in Chalermyanont and Benson (2004, 2005). An alternative semi-
empirical approach can also be used to determine these probabilities of failure, at least for
the external failure modes, because the distribution of FS for the external failure modes is
normal (sliding and overturning) or log-normal (bearing capacity) and the parameters
(mean and standard deviation) for these FS distributions can be derived directly from
statistics of the input parameters. The semi-empirical method is useful because it is simpler
and more direct relative to interpolation.
The mean FS for each external failure mode can be computed analytically using the
conventional stability equations (Eqs. 2-4) with the mean soil properties used as input.
Mean factors of safety computed using this approach are compared in Fig. 5 with those
obtained directly using the Monte Carlo simulation method described in Chalermyanont
and Benson (2005). For bearing capacity, the comparison shows the mean of the natural
logarithm of FS, because bearing capacity is log-normally distributed. For all three external
failure modes, excellent agreement exists between the mean FS computed using means as
input to Eqs. 2-4 and the means obtained directly from Monte Carlo simulation.
Analytical methods were also evaluated for computing the standard deviation of FS for
each the external failure modes. However, the analytical methods were not successful.
Thus, an empirical approach was followed. Multivariate regression was used to relate the
standard deviation of FS for the external failure modes (o
FSS
for sliding, o
FSO
for
overturning, and o
FSBC
for bearing capacity) to the parameters identified by Chalermyanont
(2002) to have a significant effect on P
f
for each external failure mode. The most precise
and accurate relationships were defined using an equation having the following form
(Chalermyanont 2002):
g fCOV e dCOV cCOV b aH log
f f FS 10
+ + µ + + + µ + = o
| | ¸ | |
(11)
where a, b, c, d, e, f, and g are empirical coefficients obtained from regression analyses. A
unique set of coefficients was obtained for each failure mode. These empirical coefficients
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Fig. 5. Comparison between µ
FS
or lnµ
FS
obtained from Monte Carlo simulation and FS or lnFS obtained from Eqs. 2-9 using
mean parameters as input: (a) sliding, (b) overturning, and (c) bearing capacity.
3 4 5 6 7
FS
O
Obtained from Deterministic Analysis
Using Mean Parameters
3
4
5
6
7
µ
F
S
O

O
b
t
a
i
n
e
d

f
r
o
m

M
o
n
t
e

C
a
r
l
o

S
i
m
u
l
a
t
i
o
n
s
(b) Overturning
1 2 3 4 5 6
FS
S
Obtained from Deterministic Analysis
Using Mean Parameters
1
2
3
4
5
6
µ
F
S
S

O
b
t
a
i
n
e
d

f
r
o
m

M
o
n
t
e

C
a
r
l
o

S
i
m
u
l
a
t
i
o
n
s
(a) Sliding
0 1 2 3 4 5
lnFS
BC
Obtained from Deterministic Analysis
Using Mean Parameters
0
1
2
3
4
5
µ
l
n
F
S
B
C

O
b
t
a
i
n
e
d

f
r
o
m

M
o
n
t
e

C
a
r
l
o

S
i
m
u
l
a
t
i
o
n
s
(c) Bearing capacity
G
S
P

1
4
0

S
l
o
p
e
s

a
n
d

R
e
t
a
i
n
i
n
g

S
t
r
u
c
t
u
r
e
s

u
n
d
e
r

S
e
i
s
m
i
c

a
n
d

S
t
a
t
i
c

C
o
n
d
i
t
i
o
n
s
are tabulated in Table 1, along with their coefficients of determination (R
2
) and standard
error (o
c
). Fits of Eq. 11 to the output from the Monte Carlo simulation are shown in Fig.
6. For all external modes of failure, o
FS
computed using Eq. 11 compares well with the o
FS
obtained from Monte Carlo simulation
Table 1. Empirical coefficients, coefficients of determination, and
standard error obtained from multivariate regression analyses.
The method used to compute P
fSYS
and R can be summarized in the following steps and
applies to walls having L/H = 0.7.
1. Determine the geometry of the wall and the spacing of the reinforcement.
2. Determine µ
|
, COV
|
, µ
¸
, COV
¸
, µ
|f
, COV
|f
, µ
T
, and COV
T
(T subscript corresponds to
tensile strength of the reinforcement). Guidance on selecting these parameters can be
found in Chalermyanont and Benson (2004).
3. Determine P
fINT
by interpolation between the design curves on the charts in
Chalermyanont and Benson (2004).
4. Compute µ
FS
for each of the external failure modes using the conventional stability
equations (Eqs. 2-4) with means input for each of the independent variables.
5. Compute o
FS
for each of the external failure modes using Eq. 11 and the coefficients in
Table 1.
6. Compute P
fS
, P
fO
, and P
fBC
as P(FS<1) using µ
FS
and o
FS
from steps 4 and 5 as input to
the cumulative distribution functions (CDFs) for the normal (sliding and overturning)
and log-normal (bearing capacity) distributions. Most modern spreadsheet programs
include functions for these CDFs. Methods to make these computations can also be
found in statistical textbooks and Chalermyanont and Benson (2004).
7. Compute P
fSYS
=P
fINT
+ P
fS
+ P
fO
+ P
fBC
and R = 1- P
fSYS
.
Mode
R
2 o
c
a b c d e f g
Sliding -0.038 0.044 0.038 0 0 0 -2.299 0.947 0.055
Overturning -0.042 0.028 0.021 0.016 0 0 -1.678 0.947 0.055
Bearing -0.077 0.009 0 0 0.083 0.043 -2.886 0.987 0.078
Capacity
Empirical Coefficient
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Fig. 6. Comparison of o
FS
from simulation and o
FS
from Eq. 11: (a) sliding, (b) overturning, and (c) bearing capacity.
25 30 35 40 45
Mean Foundation Soil Friction Angle, µ
|f
(Degrees)
0
5
10
15
20
25
30
35
S
t
a
n
d
a
r
d

D
e
v
i
a
t
i
o
n

o
f

B
e
a
r
i
n
g

C
a
p
a
c
i
t
y

F
a
c
t
o
r

o
f

S
a
f
e
t
y
,

o
F
S
B
C

Simulation Results
COV
|f
=5%
COV
|f
=10%
COV
|f
=15%
COV
|f
=20%
Calculated using Eq. 11
(c) Bearing Capacity with COV
|
= 10%, µ
|
= 30°
25 30 35 40
Mean Backfill Friction Angle, µ
|
(Degrees)
0
0.2
0.4
0.6
S
t
a
n
d
a
r
d

D
e
v
i
a
t
i
o
n

o
f

O
v
e
r
t
u
r
n
i
n
g
F
a
c
t
o
r

o
f

S
a
f
e
t
y
,

o
F
S
O
Simulation Results
COV
|
=5%
COV
|
=10%
COV
|
=15%
COV
|
=20%
Calculated using Eq. 11
(b) Overturning mode with COV¸ = 10%
25 30 35 40
Mean Backfill Friction Angle, µ
|
(Degrees)
0
0.2
0.4
0.6
0.8
1
S
t
a
n
d
a
r
d

D
e
v
i
a
t
i
o
n

o
f

S
l
i
d
i
n
g

F
a
c
t
o
r

o
f

S
a
f
e
t
y
,

o
F
S
S

Simulation Results
COV
|
=5%
COV
|
=10%
COV
|
=15%
COV
|
=20%
Calculated using Eq. 11
(a) Sliding mode
G
S
P

1
4
0

S
l
o
p
e
s

a
n
d

R
e
t
a
i
n
i
n
g

S
t
r
u
c
t
u
r
e
s

u
n
d
e
r

S
e
i
s
m
i
c

a
n
d

S
t
a
t
i
c

C
o
n
d
i
t
i
o
n
s
ANALYSIS OF SYSTEM PROBABILITY OF FAILURE FOR EXISTING WALLS
MSE walls designed based on deterministic limit equilibrium principles are ubiquitous
and most are performing successfully. To assess the P
fSYS
inherent in existing walls, three
well-documented MSE walls were analyzed: a model wall referred to as the “RMC wall”
(Bathurst et al. 1988), a full-scale test wall referred to as the “Simac wall” (Simac et al.
1990), and a real wall referred to as the “Wilson wall” (Wilson 1993). The Simac and
Wilson walls exhibited little deformation; the RMC wall was loaded until failure occurred.
The Simac wall was designed with L/H = 0.7. Thus, for the Simac wall, P
f
for each of
external failure modes was computed using the semi-empirical method described in this
paper. For internal stability, P
fINT
was interpolated from the design charts in
Chalermyanont and Benson (2004). The RMC and Wilson walls do not have L/H = 0.7 and
thus the design charts in Chalermyanont and Benson (2004) and the semi-empirical method
presented in this paper could not be used directly for analysis. Thus, for these walls, P
f
, µ
FS
,
and o
FS
for each failure mode were calculated directly using Monte Carlo simulation
following the methods in Chalermyanont and Benson (2004, 2005). Thus, the analyses of
the RMC and Wilson walls provide an assessment of the system reliability method, but not
the efficacy of Eq. 11. In all cases, P
fSYS
was computed using Eq. 10.
Wall Characteristics
The RMC wall is a 3-m-high large-scale geogrid-reinforced model wall constructed at
the Royal Military College of Canada (RMC) and is referred to as Test 3 in Bathurst et al.
(1988). The RMC wall was backfilled with granular material having a friction angle of 53°
and unit weight of 19.6 kN/m
3
. The wall was reinforced with four layers of geogrid having
an ultimate tensile strength of 14 kN/m and length of 3 m. The wall was uniformly
surcharged with 50 kPa until failure. Reinforcement creep was identified as the failure
mode and the maximum mobilized tensile stress of the reinforcement at failure was
reported by to be 7.3 kN/m (Bathurst et al. 1988).
The Simac wall is a 6.1-m-high full-scale geogrid-reinforced test wall constructed at the
FHWA research study site at Algonquin, Illinois. The wall was constructed with dry-
stacked soil-filled incremental facing units. The Simac wall was reinforced with eight
geogrid layers having an ultimate tensile strength of 39 kN/m, a long-term allowable tensile
strength of 15.5 kN/m, and length of 4.3 m. A homogeneous well-graded sand and gravel
(internal friction angle of 39°, unit weight of 20.1 kN/m
3
) was used as the backfill and as
the foundation soil. The Simac wall was tested with and without surcharge loading. Failure
did not occur in either case.
The Wilson wall is a 4.9-m-high geogrid-reinforced wall with precast concrete face
panels that is located at Highway 97 in Westbank, British Columbia, Canada. The backfill
was sandy gravel with cobbles having an internal friction angle of 35° and unit weight of 20
kN/m
3
. The wall was reinforced with eight geogrid layers having an allowable tensile
strength of 22 kN/m and length of 2.9 m.
Characteristics of these walls reported in the literature were assumed to be mean values.
No foundation soil properties or statistics representing uncertainty in the parameters were
reported. Thus, the mean foundation soil properties were assumed to be the same as the
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
mean backfill properties and typical values were assumed for the COVs (COV
|
= COV
|f
=
10%, COV
¸
= COV
¸f
= 5%, and COV
T
= 5%).
Assessment of the System Probability of Failure
The P
f
, µ
FS
, o
FS
, and P
fSYS
of the walls are shown in Table 2. Results of two analyses are
shown for the RMC and Simac walls. RMC-1 corresponds to an analysis using the
mobilized tensile strength reported by Bathurst et al. (1988), whereas RMC-2 corresponds
to an analysis with an adjusted mean tensile strength of the reinforcement. Simac-1
corresponds to an analysis of the Simac wall without surcharge loading, whereas Simac- 2
corresponds to an analysis with surcharge loading. For all analyses, P
fSYS
was found to
depend almost exclusively on P
fINT
rather than P
fS
, P
fO
, or P
fBC
. In general, P
fINT
was at least
100 times larger than any of the P
f
for the external failure modes.
Table 2. Probabilities of failure of existing walls.
The Simac-1

and Wilson walls have P
fSYS
s 0.0003 (Table 2). Both of these walls were
designed with factors of safety exceeding unity (the internal factor of safety was 1.1 for
Simac-1; the Wilson had FS = 1.5 for internal stability, 2.0 for sliding, 1.5 for overturning,
and 2.0 for bearing capacity). Thus, these case histories suggest that the inherent P
fSYS
for
conventionally designed MSE walls is approximately one to two orders of magnitude lower
than the failure rate for similar geotechnical structures (0.01 to 0.001, Baecher 1987).
Higher P
fSYS
was obtained for the Simac-2 wall, which is consistent with the absence of
failure but also reflects the effect that heavy surcharging had on the likelihood of failure.
P
fSYS
of the RMC wall was less than 0.0001 when the geogrid tensile strength (µ
T
) was
set at 7.3 kN/m, as reported by Bathurst et al. (1988) (RMC-1). This P
fSYS
is much lower
than expected for a wall that failed (P
f
closer to unity was expected). Another analysis
(RMC-2), performed using µ
T
=6.7 kN/m, yielded P
fSYS
= 0.017. Thus, P
fSYS
for the RMC
wall is very sensitive to the input parameters that are assumed, which is reasonable given
that the RMC wall was designed to fail. Another reason for P
fSYS
being lower than
expected is that the RMC wall failed due to excessive deformation attributed to creep. This
effect can be dealt with using the system reliability approach by carefully selecting µ
T
so
that it accurately reflects the operative tensile strength in the field (see Chalermyanont and
Benson 2004 for a discussion of this issue).
Wall P
fSYS
P
fINT
P
FSINT
P
FSINT
P
fS
µ
FSS
o
FSS
P
fO µ
FSO
o
FSO
P
fBC
µ
FSBC
o
FSBC
RMC-1 <0.0001 1.12 0.042 <0.0001 16.32 2.519 <0.0001 13.69 1.353 <0.0001 506.75 184.144 <0.0001
Simac-1 0.0003 N/A N/A <0.0001 5.02 0.367 <0.0001 6.55 0.280 <0.0001 19.57 4.600 0.0003
Simac-2 > 0.01 N/A N/A <0.0001 4.92 0.367 <0.0001 6.21 0.280 <0.0001 18.27 4.600 > 0.01
Wilson <0.0001 1.41 0.033 <0.0001 3.10 0.276 <0.0001 3.96 0.180 <0.0001 6.65 2.359 <0.0001
Internal Stability External Stability
Sliding Bearing Capacity Overturning Slope Stability
RMC-2 0.0171 1.08 0.041 <0.0001 16.32 2.519 <0.0001 13.69 1.353 <0.0001 506.75 184.144 0.0171
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
SUMMARY
A method for calculating the system probability of failure and reliability of MSE walls
having a length-to-height ratio of 0.7 has been described. The system probability of failure
(P
fSYS
) is the sum of the probabilities of failure corresponding to the four common modes of
failure for MSE walls: slope stability, sliding, overturning, and bearing capacity. System
reliability is the complement of the system probability of failure. Probabilities of internal
failure (slope stability) are obtained from analysis of MSE walls using design charts
presented in Chalermyanont and Benson (2004). Probabilities of failure due to sliding,
overturning, and bearing capacity are calculated using the mean factor of safety computed
using traditional deterministic methods and standard deviations computed using an
empirical equation.
P
fSYS
of existing walls were computed to determine the probability of failure inherent in
typical MSE wall designs (i.e., FS
INT
= 1.5 for internal stability, FS
S
= 1.5 for sliding, FS
O
=
2.0 for overturning, and FS
BC
= 2.0 for bearing capacity). This analysis showed that the
probability of failure typically is < 0.0003, which is approximately one order of magnitude
lower than the failure rate (0.01 to 0.001) for similar geotechnical structures
ACKNOWLEDGEMENT
The Royal Thai Government provided financial support for this study. This support is
gratefully acknowledged.
REFERENCES
AASHTO (1992), Standard Specifications for Highway Bridges (15
th
Edition), American
Association of State Highway and Transportation Officials, Washington, DC.
Alpsten, G. (1972), “Variations in Mechanical and Cross-Sectional Properties of Steel,”
Proc. International Conf. on Planning and Design of Tall Buildings, Lehigh University,
Bethlehem, PA, 775-805.
Baecher, G. (1987), “Geotechnical Risk Analysis User’s Guide,” FHWA/RD-87-011,
Federal Highway Administration, McLean, Virginia, 55 p.
Basheer, I. and Najjar, Y. (1996), “Reliability-Based Design of Reinforced Earth Retaining
Walls,” Transportation Research Record 1526, National Academy Press, Washington
DC, 64-78.
Bathurst, R., Benjamin, D., and Jarrett, P. (1988), ”Laboratory Study of Geogrid Reinforced
Soil Walls,” Proceeding of Symposium on Geosynthetics for Soil Improvement,
Geotechnical Special Publication No. 18, ASCE, 178-192.
Benson, C. (1991), “Predicting Excursions Beyond Regulatory Thresholds of Hydraulic
Conductivity from Quality Control Measurements,” Proceedings of the First Canadian
Conference on Environmental Geotechnics, CGS, Montreal, Quebec, Canada, 447-454.
Bishop, A. (1955), “The Use of the Slip Circle in the Stability Analysis of Slopes,”
Geotechnique, 5(11), 7-17.
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Bonaparte, R., Holtz, R., and Giroud, J. (1987), “Soil Reinforcement Design Using
Geotextiles and Geogrids,” Geotextile Testing and the Design Engineer, STP 952, J.
Fluet, Jr., Ed., American Society for Testing Materials, Philadelphia, 69-116.
Broms, B.. (1978), “Design of Fabric Reinforced Retaining Structures,” Proceedings of the
Symposium on Earth Reinforcement, ASCE, 282-304.
Chalermyanont, T. (2002), “Reliability Analysis of Mechanically Stabilized Earth (MSE)
Walls,” Ph.D. Dissertation, University of Wisconsin-Madison.
Chalermyanont, T. and Benson, C. (2004), “Reliability-Based Design for Internal Stability
of Mechanically Stabilized Earth (MSE) Walls,” Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 130(2),163-173.
Chalermyanont, T. and Benson, C. (2005), “Reliability-Based Design for External Stability
of Mechanically Stabilized Earth (MSE) Walls,” International Journal of Geomechanics,
ASCE, in press.
Cherubini, C. (2000), “Reliability Evaluation of Shallow Foundation Bearing Capacity on
c’, |’ Soil,” Canadian Geotechnical Journal, 37, 264-269.
Christian, J. (2004), “Geotechnical Engineering Reliability-How Well Do We Know What
We are Doing?” Journal of Geotechnical and Geoenvironmental Engineering, ASCE,
130(10), in press.
Claybourn, A. and Wu, J. (1991), “Case History Comparison of Geosynthetic-Reinforced
Soil Walls,” Geosynthetics‘91, Industrial Fabrics Association International, St. Paul, MN,
2, 549-559.
Duncan, J. (2000), “Factors of Safety and Reliability in Geotechnical Engineering,” Journal
of Geotechnical and Geoenvironmental Engineering, ASCE, 126(4), 7-316.
Elias, V. and Christopher, B. (1997), Mechanically Stabilized Earth Walls and Reinforced
Soil Slopes, Design and Construction Guidelines, FHWA Demo Project 82-1,
Washington DC.
Harr, M. (1987), Reliability-Based Design in Civil Engineering, Dover Publication, New
York.
Ho, S. and Rowe, R. (1997), “Effect of Wall Geometry on the Behavior of Reinforced Soil
Walls,” Geotextiles and Geomembranes, 14, 521-541.
Hoeg, H. and Murarka, R. (1974), “Probabilistic Analysis and Design of a Retaining Wall,”
Journal of the Geotechnical Engineering Division, ASCE, 100(3), 349-366.
Koerner, R. (1994), Design with Geosynthetics, Prentice Hall, Englewood Cliffs, New
Jersey.
Lacasse, S. and Nadim, F. (1996), “Uncertainties in Characterizing Soil Properties,”
Uncertainty’ 96, Geotechnical Special Publication No. 58, ASCE, 1, 49-75.
Leshchinsky, D. and Perry, E. (1987), “A Design Procedure for Geotextile-Reinforced
Walls,” Geosynthetics ‘87, Industrial Fabrics Association International, St. Paul, MN, 95-
107.
Low, B. and Tang, W. (1997), “Reliability Analysis of Reinforced Embankments on Soft
Ground,” Canadian Geotechnical Journal, 34, 672-685.
Lumb, P. (1966), “The Variability of Natural Soils,” Canadian Geotechnical Journal, 3(2),
74 – 97.
Meyerhof, G. (1963), “Some Recent Research on the Bearing Capacity of Foundations,”
Canadian Geotechnical Journal, 1, 16-26.
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions
Phoon, K. and Kulhawy, F. (1999), “Evaluation of Geotechnical Property Variability,”
Canadian Geotechnical Journal, 36, 625-639.
Phoon, K., Kulhawy, F., and Grigoriu, M. (2003), “Development of a Reliability-Based
Design Framework for Transmission Line Structure Foundations,” Journal of
Geotechnical and Geoenvironmental Engineering, ASCE, 129(9), 798-806.
Pier, J. and Cornell, C. (1973), “Spatial and Temporal Variability of Live Loads,” Journal
of the Structural Division, ASCE, 99(ST5), 903-922.
Tobias, P. and Trindade, D. (1995), Applied Reliability, Van Nostrand Reinhold, NY.
Schmertmann, G., Chouery-Curtis, V., and Johnson, R. (1987), “Design Charts for
Geogrid-Reinforced Soil Slopes,” Geosynthetics ‘87, Industrial Fabrics Association
International, St. Paul, MN, 108-120.
Siegel, R. (1975), STABL User Manual, Purdue University, West Lafayette, Indiana.
Simac, M., Christopher, B., and Bonczkiewicz, C. (1990), “Instrumented Field Performance
of a 6 m Geogrid Soil Wall,” Proceedings of 4
th
Conference on Geotextiles,
Geomembranes, and Related Products, Balkema, Rotterdam, 1, 53-59.
Whitman, R. (2000), “Organizing and Evaluating Uncertainty in Geotechnical
Engineering,” Journal of Geotechnical and Geoenvironmental Engineering, ASCE,
126(7), 583-593.
Wilson, H. (1993), “Geogrid Reinforced Soil Against Precast Concrete Building Wall
British Columbia, Canada,” Geosynthetics Case Histories, International Society of Soil
Mechanics and Foundation Engineering, 256-257.
GSP 140 Slopes and Retaining Structures under Seismic and Static Conditions

G E O T E C H N I C A L S P E C I A L P U B L I C A T I O N N O . 1 4 0
SLOPES AND RETAINING
STRUCTURES UNDER SEISMIC
AND STATIC CONDITIONS
PROCEEDINGS OF SESSIONS OF THE GEO-FRONTIERS 2005 CONGRESS

January 24–26, 2005
Austin, Texas



SPONSORED BY
Geosynthetics Committee of
The Geo-Institute of the American Society of Civil Engineers



EDITED BY
Mohamed A. Gabr, Ph.D., P.E.
John J. Bowders, Ph.D., P.E.
David Elton, Ph.D., P.E.
Jorge G. Zornberg, Ph.D., P.E.



Published by the American Society of Civil Engineers
Notices
ISBN 0-7844-0769-X

Any statements expressed in these materials are those of the individual authors and do
not necessarily represent the views of ASCE, which takes no responsibility for any
statement made herein. No reference made in this publication to any specific method,
product, process or service constitutes or implies an endorsement, recommendation,
or warranty thereof by ASCE. The materials are for general information only and do
not represent a standard of ASCE, nor are they intended as a reference in purchase
specifications, contracts, regulations, statutes, or any other legal document.

ASCE makes no representation or warranty of any kind, whether express or implied,
concerning the accuracy, completeness, suitability, or utility of any information,
apparatus, product, or process discussed in this publication, and assumes no liability
therefore. This information should not be used without first securing competent
advice with respect to its suitability for any general or specific application. Anyone
utilizing this information assumes all liability arising from such use, including but not
limited to infringement of any patent or patents.

Copyright © 2005 by the American Society of Civil Engineers.
All Rights Reserved.
ISBN 0-7844-0769-X
Manufactured in the United States of America.

American Society of Civil Engineers
ASCE International Headquarters
1801 Alexander Bell Drive
Reston, VA 20191-4400 USA

Call Toll-Free in the U.S.: 1-800-548-2723 (ASCE)
Call from anywhere in the world: 1-703-295-6300
Internet: http://www.pubs.asce.org



Preface
Design of slopes and retaining structures continue to pose wide challenges to
geotechnical engineers. These include long term performance under static and seismic
conditions and in harsh environments as well as the issues related to use of non-
traditional backfills and the reliability of the constructed systems throughout the
design life. Several of the challenges related to design and construction of slopes and
retaining structures are addressed with the increased utilization of geosynthetics over
the past two decades. The level of activities associated with geosynthetics
reinforcement range from paved and unpaved roads to the construction of segmental
retaining walls and slopes. In these proceedings, field and laboratory performance of
retaining walls and slopes under seismic and static conditions are premiered. National
and international experts present their perspectives regarding performance of
reinforced and unreinforced pavement systems, slopes and retaining structures in
harsh environments and under severe conditions.

This Geotechnical Specialty Publication (GSP) includes papers presented in the Slope
and Retaining Structures (SRS) track at the ASCE Geo-Institute specialty conference,
Geo-Frontiers 2005 in Austin, Texas. Several of the sessions were sponsored by the
G-I Geosynthetics committee. All papers in these proceedings have been peer-
reviewed by two anonymous reviewers in accordance with ASCE and the Geo-
Institute. Required revisions were made by the authors prior to final acceptance and
publication. All papers are eligible for discussion in the ASCE Journal of
Geotechnical and Geoenvironmental Engineering and for ASCE awards. The editors
wish to express sincere thanks to the session chairs for their diligence, and to all of
the reviewers for their willingness to help and prompt responses.

Mohamed Gabr
North Carolina State University-Raleigh

John J. Bowders
University of Missouri-Columbia

David Elton
Auburn University

Jorge G. Zornberg
University of Texas at Austin