Você está na página 1de 11

Indian Geotech J (October–December 2013) 43(4):292–302

DOI 10.1007/s40098-013-0062-6

ORIGINAL PAPER

Reliability Based LRFD Approach for External Stability


of Reinforced Soil Walls
B. Munwar Basha • G. L. Sivakumar Babu

Received: 6 February 2013 / Accepted: 29 April 2013 / Published online: 14 May 2013
 Indian Geotechnical Society 2013

Abstract The main objective of reinforced soil wall List of Symbols


design is to ensure safety and performance requirements of c = cohesion of the foundation soil
a wall for target reliability levels. Analysis is conducted to e = eccentricity of the resultant force
evaluate the external stability of reinforced soil walls. As fX (x) = is a joint probability density function of X
this must be accomplished under conditions of uncertain- FSsli, FSb = factor of safety against sliding and
ties associated with reinforced backfill, retained backfill, bearing failure modes
foundation soil and surcharge load acting on the wall, FSe = factor of safety against eccentricity
probabilistic analyses are necessary in the development of failure mode
such reliability based design for external stability of rein- DL = distributed dead load surcharge
forced soil wall. A methodology for developing load and g = acceleration due to gravity
resistance factor design guidelines for reinforced soil walls g(.) = limit state function
is outlined in this paper. The potential failure mechanisms H = height of geosynthetic reinforced soil
considered in the analysis are sliding failure, eccentricity wall
failure of resultant force (or overturning failure) and L = length of the geosynthetic reinforcement
bearing capacity failure. A first order reliability method is LL = live load surcharge
used to determine appropriate ranges for the values of the Nc, Nq, Nc = bearing capacity factors
load and resistance factors. Load and resistance factors Pa = active earth pressure due to backfill soil
needed to maintain the stability against three modes of Pq = active earth pressure due to surcharge
failure by targeting component reliability indices of 2, 2.5, load
3, 3.5 and 4 are obtained for various values of coefficients Ka = active earth pressure coefficient
of variation of friction angle of backfill and foundation soil, q = surcharge load acting on the backfill soil
distributed dead load surcharge, live load surcharge and qu = ultimate bearing capacity of a shallow
cohesion of the foundation soil. foundation
Q = surcharge coefficient
Keywords Load factors  Load resistance factor design  ui = variables in standard normal space
Limit states  Reliability analysis  X ¼ fxi gni¼1 = vector of random variables representing
Resistance factors, reinforced soil wall uncertain quantities
U ¼ fuk gnk¼1 = vector of standard random variables
representing uncertain quantities
B. M. Basha (&)
Department of Civil Engineering, Indian Institute of Technology u* = most probable point of failure (MPP)
Delhi, Hauz Khas, New Delhi 110016, India xi = design variable corresponding to target
e-mail: basha@civil.iitd.ac.in; mbasha@gmail.com reliability index
WABFE = weight of the reinforced soil block
G. L. S. Babu
Department of Civil Engineering, Indian Institute of Science, ‘ABFE’
P
Bangalore 560 012, India Fr = sum of the horizontal resisting forces

123
Indian Geotech J (October–December 2013) 43(4):292–302 293

P
Fd = sum of the horizontal driving forces external failure modes of the wall. The factor of safety is an
U(.) = the standard normal cumulative empirical, but arbitrary, measure used to reduce the
distribution potential for adverse performance.
bt = target reliability index For the external stability to be satisfied, the wall must be
bsli, bb = reliability indices against sliding and safe against three modes of failures: overturning, sliding
bearing failure modes and bearing. For retaining walls, the loads to be considered
be = reliability indices against overturning and are: weight of the wall, dead earth load, lateral earth
eccentricity failure modes pressure and live load surcharge.
COV = coefficient of variation The design of any earth retaining wall requires infor-
c = unit weight of the reinforced backfill soil mation on geotechnical properties of soils to be retained
cb = unit weight of the foundation soil and also of the soils which serve as support for the
li = mean of random variable retaining wall. In practice, soil parameters are known to
lNi = equivalent mean of non normal random exhibit some degree of variability and uncertainty. An
variable important issue in the evaluation of stability of RSW is the
ri = standard deviation of random variable lack of exact knowledge of strength parameters of the
rNi = equivalent standard deviation of non backfill and foundation soil parameters. Since there are
normal random variable many uncertainties in backfill and foundation soil proper-
gi = load factor ties, the analysis from probabilistic point of view is needed.
Wi = resistance factor A few studies are available in the literature. Chalermyanont
/ = friction angle of the reinforced backfill and Benson [1] conducted a reliability study on the external
soil stability of mechanically stabilized earth walls in static
/b = friction angle of the soil below the base conditions. Basha [2] and Basha and Babu [3] presented a
slab of the retaining wall study on reliability based design optimization of external
d = interface wedge friction angle between seismic stability of reinforced soil structures under earth-
reinforced and retained backfill quake loading using pseudo-dynamic method. Again, using
db = interface friction angle between wall base similar pseudo dynamic approach, Basha and Babu [4]
and foundation soil presented a methodology for reliability assessment of
rv = vertical stress at the base and seismic internal stability of reinforced soil structures using
hw = angle of slope with vertical logarithmic spiral failure mechanism by taking into
account three failure modes such as tension failure, pullout
failure and total pullout failure. In addition, Basha and
Babu [5] reported seismic external stability analysis of
geosynthetic reinforced soil walls by taking into account
Introduction the effect of uncertainties using target reliability based
approach (TRA) considering three modes of failure such as
The design of reinforced soil walls (RSW) has traditionally sliding, bearing and overturning or eccentricity failure.
been performed using allowable stress design (ASD) in Design of a mechanically stabilized soil structure
which all uncertainty in loads and material resistance is requires explicit satisfaction of multiple performance cri-
combined in a factor of safety. Design practice of rein- teria such as sliding stability, eccentricity (or overturning)
forced soil retaining walls is based on the limit equilibrium stability and bearing capacity stability modes. Federal
approach currently. The walls are designed for both Highway Administration, FHWA [6] reported that the
external and internal stability criteria. There are several RSW must be designed to avoid external modes of failure,
possible failure modes considered in the design of rein- viz. sliding failure on its base, overturning failure (or in
forced soil retaining walls to satisfy both external and terms of eccentricity failure of the resultant force striking
internal stability. Both external and internal stability issues the footing base) and bearing capacity failure of the
need to be addressed in the design. In this study, external foundation soil. American Association of State Highway
stability is focused. The reinforcement length is governed and Transportation Officials, AASHTO [7] recommended
by the external stability conditions and the vertical spacing that for static loading, the minimum factors of safety in
of reinforcements is governed by the internal stability relation to sliding and overturning modes are 1.5 and 2.0
conditions. Further, design of the wall should be such that respectively, and eccentricity of the resultant force should
minimum required safety factors are fulfilled for all failure be less than one sixth of the width of wall. Further, the
modes. The major design considerations for a reinforced minimum factor of safety against bearing capacity failure
soil structure are the stability assessment of the potential mode should vary between 2.0 and 2.5. Currently used

123
294 Indian Geotech J (October–December 2013) 43(4):292–302

design manuals for the design of RSW rely primarily on the it possible to give proper weight to the accuracy with which
traditional allowable stress design (ASD) format in which the various loads and resistances can be determined.
the safety factors are prescribed deterministically. The load and resistance factors are dependent on the
AASHTO and FHWA are committed to transforming degree of uncertainty and influence of the relevant quan-
the current ASD method to load and resistance factor tities, and on the desired level of safety. The reference
design (LRFD). LRFD has the advantage over conven- manual by FHWA [6] reported LRFD approach (refer to
tional ASD, in that it accounts for the variability in both Chapter 14) for the mechanically stabilized earth walls, and
resistance and load, as well as provides the same consistent presented a first step toward developing load and resistance
probability of failure for the structure. LRFD is based factors, addressing static loading. For the design of rein-
primarily on a rational evaluation of performance reli- forced soil walls, the current FHWA [6] recommends the
ability. It represents an approach in which applicable fail- resistance factors in bearing capacity mode range from 0.35
ure and serviceability conditions can be evaluated to 0.60 depending on the design method, and for base
accounting for the uncertainties associated with loads and sliding mode it should be 1.0. However, the existing
material resistance. resistance factors are recommended based on the variabil-
In the LRFD, various types of loads are multiplied by ity associated with the properties of the backfill that were
load factors and the ultimate resistance is multiplied by a collected from sites that do not necessarily reflect the
resistance factor. The uncertainty in loads is represented variability of local backfill soils or design practice.
by load factors that generally have a value greater than Therefore, the resistance factors recommended by the
one, and the uncertainty in material resistance is repre- existing FHWA [6] code need to be verified before being
sented by a resistance factor that generally has a value applied to local backfill condition.
less than one. In geotechnical design, the resistance fac- In this direction, Allen et al. [8], Bathurst et al. [9] and
tors depend on the uncertainties associated with the var- Kulhawy and Phoon [10] described methodologies for LRFD
iability and reliability of different factors that include the for geotechnical and structural design. Basha and Babu [11]
extent of soil exploration and type of sampling and testing reported the LRFD approach for the reliability-based seismic
used to characterize a site; inherent soil variability; soil design of bridge abutments using pseudo-static limit equi-
property measurements; the procedures or models used librium method, considering overturning failure, eccentric-
for design; and the measures employed to monitor the ity failure, sliding failure along the base and bearing capacity
construction processes. Thus selecting resistance factors instability. Kim and Salgado [12] considered three target
that target an acceptable probability of survival is a dif- values of reliability indices (2.0, 2.33, and 3.0) to address the
ficult one. However, geotechnical engineers have the effect of redundancy on the internal stability checks of
opportunity to control the extent and type of sampling and mechanically stabilized earth walls.
testing used to characterize a site, and the procedures or
models used for design.
The selection of load factor combination will depend on Objectives of the Present Study
the mode of failure to be analyzed. The load factor com-
bination that results in the maximum vertical load controls The attempt made in this paper is perhaps the first study to
the bearing capacity consideration. Load factor combina- propose a LRFD approach for the probabilistic assessment of
tions that include minimum vertical loads and maximum stability of RSW. A reliability based optimization method-
horizontal loads control the sliding resistance as well as the ology (Basha and Babu [13]) is used to obtain the load and
overturning. Having the greatest net overturning moment resistance factors considering the variability associated with
produces the largest resultant eccentricity. distributed dead load (DL) surcharge acting on the wall,
The first objective of this paper is to evaluate the load traffic surcharge in the form of live load (LL) and properties
and resistance factors for the design of RSW against of reinforced backfill, retained backfill, foundation soil, and
external failure modes with the rigorous probability-based length of the reinforcement. The reliability analysis is
framework of LRFD approach. The designation LRFD addressed following FHWA [14] guidelines, considering the
reflects the concept of factoring both loads and resistance. variability associated with static loading.
This type of factoring differs from the ASD specification,
where only the resistance is divided by a factor of safety (to
obtain allowable stress). The LRFD approach was devised Reliability Based Load and Resistance Factor Design
to offer the designer greater flexibility, more rationality,
and possible overall economy. In the present investigation, The resistance of the wall should be set in such a way that
the use of a probabilistic mathematical model in the the factored load effects do not exceed the factored resis-
development of the load and resistance factors, which made tance for possible limit states. The term limit state refers to

123
Indian Geotech J (October–December 2013) 43(4):292–302 295

any set of conditions that may produce unsatisfactory xi ¼ ri ui þ li ð3Þ


performance of the reinforced soil wall. These limit states
would be associated with the various static and earthquake Where, i = 1, 2,……n and li is the mean of random
loads considered in the design. In this section, LRFD for- variable xi. If the variables are not normally distributed, the
mat using reliability based design optimization (RBDO) parameters of the equivalent normal distribution, lNi and
framework is presented (Ang and Tang [15]). The RBDO rNi need to be computed. It can be done by imposing the
process iteratively carries out two optimizations: a design conditions that the cumulative distribution function and the
optimization in an original random space (X—space) and a probability density functions of the actual variable and the
reliability analysis in an independent and standard normal equivalent normal variable are equal at the check point on
random space (U—space). The general form of RBDO the failure surface, that is at the check point xi,
 
problem is formulated as follows: The limit-state function xi  lNi
U ¼ F X ðx i Þ ð4aÞ
[g (x) = 0] involving ‘n’ random variables divides the rNi
design space into a failure domain (g (x) \ 0) and a safe
domain (g (x) [ 0) and hence serves as a safety criterion. where lNi and rNi are the mean and standard deviation of
During the RBDO process, a transformation between X ¼ the equivalent normal variable at the design point, and FX
fxk gnk¼1 and U ¼ fuk gnk¼1 —spaces at a design point must (xi) is the cumulative distribution function of the original
be carried out to estimate the probabilistic constraint. The non-normal variable. Rearranging Eq. (4), we get.
procedure for target reliability approach (TRA) demon- lNi ¼ xi  U1 ½FX ðxi ÞrNi ð4bÞ
strated by Basha and Babu [2, 3, 5, 11, 13] and for the
reliability analysis is used in this paper to estimate the load Equating probability density functions of the original
and resistance factors in sliding failure, overturning or variable and the equivalent normal variables at the check
eccentricity failure and bearing capacity failure which is point,
 
explained in the following section. 1 xi  lNi
/ ¼ f X ðx i Þ ð4cÞ
The load and resistance factors are estimated using rNi rNi
TRA. TRA is formulated as the inverse of the first order
reliability method (FORM), in which the most probable where /() and fX (xi) are the probability density functions
point of failure (MPPT) where the distances from the origin of the equivalent standard normal and the original random
in U—space are equal to the target reliability index (bt) are variables. From Eq. (6), we get.
sought. The procedure of calculating the target reliability  
/ U1 ½FX ðxi Þ
index is outlined in Fig. 1, which demonstrates the MPPT rNi ¼ ð4dÞ
search. The circles represent the bt (= 1, 2, 3, 4…etc.) fX ðxi Þ
spheres with the target curve represented by dotted circles. The algorithm to find the partial factors then proceeds in
The point on the target circle with the minimum and five iterative steps.
positive value of limit state equation [i.e. g (u) = 5, 1. Assume initial values of xi and compute the values of
10…etc.] is searched. This point is called the MPPT lNi and rNi if it is not normally distributed.
(Fig. 1). The first-order probabilistic performance measure, 2. Compute the design point (ui ) corresponding to given
is obtained from the optimization problem in U space, as bt using Eq. (2).
(
minimize gðuÞ 3. Compute the values of xi using xi ¼ rNi ui þ lNi .
Find u; which pffiffiffiffiffiffiffiffi ð1Þ 4. Repeat steps 1 through 3 until the process converges.
subject to uT u ¼ bt
5. After identifying xi value, the partial factors which
where g (u) = 0 describes the failure surface in the consists of load factors (gi) and resistance factors (Wi)
standard normal space. The design point (uk) corresponding required to provide the given target reliability index
to a given target reliability index (bt) in the standard (bt), can be computed by dividing xi with the mean
normal space (U) can be expressed as (Basha and Babu values of basic variables (li) as shown below:
[13]) x
ð W i ; gi Þ ¼ i ð5Þ
Pn og li
i¼1 oxk ri
uk ¼ bt rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n where
Pn Pn og o2 ð2Þ
j¼1 i¼1 oxk ri External Stability Analysis of RSW
k ¼ 1; 2; . . .. . .n
The reinforced soil zone is assumed to behave as one rigid
Where, ri is the standard deviation of random variable xi. unit in the external stability analysis. Since this zone is
Further, the random variable ‘xi’ can be found as follows: supposed to act as one unit, the failure mechanisms used

123
296 Indian Geotech J (October–December 2013) 43(4):292–302

Fig. 1 Determination of most


probable point of failure for
target reliability index (MPPT)
using target reliability based
approach (TRA) (Basha and
Babu [13])

for conventional gravity retaining walls also apply to the d = interface friction angle between the reinforced and
external stability analysis of reinforced soil retaining walls retained backfill (= 0 for horizontal backfill). For the static
(FHWA [14]). A schematic diagram of a typical reinforced case, the location of the resultant of the soil pressure acting
soil retaining wall, along with the forces acting on the wall on the reinforced block as shown in Fig. 2 is taken as H/3.
used for external stability analysis, is shown in Fig. 2. By The expressions of limit state functions for three failure
using the design methods and equations given in FHWA modes are derived as discussed below.
[14] guidelines and AASHTO [16] specifications, the
expressions for three failure modes have been derived and
presented in this section. The sliding wedge failure Performance Function for Sliding Failure Mode
mechanism is considered for the external stability analysis
as shown in Fig. 2. For the stability against sliding failure along the base of
The composite mass must be stable against sliding P
wall, sum of the horizontal resisting forces ( Fr ) should
failure along the base of the structure at the foundation/ P
be more than sum of the horizontal driving forces ( Fd ).
backfill interface, eccentricity failure of resultant force (or The factor of safety against sliding failure is given by,
overturning failure about the toe), and bearing-capacity P P
failure of the supporting foundation soils. Bearing capacity Fr V tan db þ k1 cL
FSsli ¼ P ¼ ð8Þ
calculations assume that the base of the reinforced zone Fd P a þ Pq
acts as an eccentrically loaded footing with an equivalent P
where, V is the normal force at the base = (WABFE ?
footing width (i.e., Meyerhof approach, FHWA [14]). The DL 9 L), DL = distributed dead load surcharge (DL) on the
measure of relative stability against the external modes of wall, k1 = 2/3, Pa is the static active earth pressure (=
failure is defined by the ratio of resisting forces to driving 0.5 cH2Ka), Pq is the active earth pressure due to dead load
forces (or moments in the case of overturning) as in the and live load surcharge =(DL?LL)HKa, LL is the live load
design of conventional gravity wall structures. A free body traffic surcharge, db = interface friction angle between wall
diagram of the wall showing different forces coming onto it base and foundation soil expressed in terms of /b,
from soil along with their respective points of applications /b = friction angle of the soil below the base slab of the
is shown in Fig. 2. retaining wall, and L = length of the geosynthetic
The external stability of the wall can be analyzed by the reinforcement. Substituting the above definitions in Eq. (8)
following procedure as reported in FHWA [14]. The earth and normalizing with 0.5 cH2, we get
pressure can be calculated from the following equation:
2½L=H þ fDL=cH gL=H  tan db þ 2k1 ðc=cHÞðL=HÞ
Pa ¼ 0:5cH 2 Ka ð6Þ FSsli ¼
Ka ð1 þ QÞ
where c = unit weight of reinforced backfill, H = height ð9Þ
of reinforced soil wall, and Ka is the static active earth
where, Q = 2(DL?LL)/cH, ‘c’ is the cohesion of
pressure coefficient given by
foundation soil. Safety margin against sliding failure of
cos2 ð/Þ GRS wall can be expressed as
Ka ¼ h qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffii2 ð7Þ
cosðdÞ 1 þ sinðdþ/Þ sinð/Þ gsli ðxÞ ¼ FSsli  1 ð10Þ
cosðdÞ

123
Indian Geotech J (October–December 2013) 43(4):292–302 297

Fig. 2 Forces and earth


pressures considered for the
external stability analysis of
GRS wall

Performance Function for Eccentricity Failure Mode L


ge ðxÞ ¼ e ð15Þ
6
FHWA [14] reported that for the stability of wall in terms
Performance Function for Bearing Capacity Failure
of eccentricity of resultant force under static conditions, the
Mode
eccentricity should be less than one sixth of the base width
of the wall (L/6) (as shown in Fig. 2). Eccentricity of the
Meyerhof distribution assumes that eccentric loading
resulting force (e) on the base can be calculated by sum-
results in a uniform redistribution of pressure over a
ming the moments of resisting forces and disturbing forces
reduced area at the base of the wall as shown in Fig. 2. This
about the center line of the wall base as shown in the
area is defined by a width equal to the wall width less twice
following equation:
the eccentricity (L-2e). The factor of safety against bear-
R  e ¼ Pa  H=3 þ Pq  H=2 ð11Þ ing capacity failure can be estimated as the ratio of ultimate
where R is resultant of vertical forces and WABFE is the bearing capacity (qu) of a shallow foundation below the
weight of the reinforced soil block ‘ABFE’can be base slab of wall and vertical stress at the base (rv) cal-
expressed as culated with Meyerhof type distribution (Fig. 2) and is
shown below:
R ¼ WABFE þ DL  L ð12Þ
qu
WABFE ¼ cHL ð13Þ FSb ¼ ð16Þ
rv
After simplification, Eq. (11) reduces to
where qu and rv are defined as follows:
e ½Ka ð1=3Þ þ 0:5QKa 
¼ ð14Þ qu ¼ cNc þ 0:5cb ðL  2eÞNc ð17Þ
H 2ðL=HÞ þ ½2  DL=ðcH ÞðL=HÞ
P
V
The safety margin against eccentricity failure under rv ¼ ð18Þ
static loading can be estimated as the difference of one L  2e
sixth of the base width of the wall and eccentricity of the cb = unit weight of foundation soil, Nc and Nc are bearing
P
resultant force (e) striking the base from middle point of capacity factors (Das [17]), and V ¼ WABFE þ
the base and is given by ðDL þ LLÞL. Normalizing qu and rv with cH, we get

123
298 Indian Geotech J (October–December 2013) 43(4):292–302

   
qu c c L e parameters are reported in Table 1. The COV values are
¼ Nc þ 0:5 b 2 Nc ð19Þ consistent with representative data reported in literature
cH cH c H H
[18–21]. The summary of load and resistance factors
rv 1 þ Q=2 computed for different failure modes are listed in Table 2.
¼ ð20Þ
cH 1  2ðe=HÞðH=LÞ The probabilistic constraints in the form of performance
Safety margin against bearing capacity failure generally functions are given below
can be written as gsli ðxÞ  0; ge ðxÞ  0; gb ðxÞ  0 ð22Þ
gb ðxÞ ¼ FSb  1 ð21Þ The load factors (gk) and resistance factors (Wk) com-
puted for different failure modes are presented in Figs. 3, 4,
5, 6, 7, 8, 9, 10. The load and resistance factors for sliding,
Results and Discussion eccentricity and bearing capacity failure modes are com-
puted using the expressions, W/ = /*/l/, W/b ¼ /b =l/b ,
Selection of Mean, Standard Deviation and Probability Wc = c*/lc and gLL = LL*/lLL. The terms in these
Density Function (PDF) of Random Variables expressions, viz. /*, /b , c* and LL* are the design values
(corresponding to target value of reliability index, bt) of /,
A RSW resting on c-/ soil is considered for the present /b, c and LL respectively and l/, l/b , lc and lLL are the
study. The distributed DL and LL surcharges are assumed mean values of the random variables /, /b, c and LL
to be 12 and 24 kN/m2 in the present study. This study respectively. The load and resistance factors (gk and Wk)
includes variability in the properties of backfill, foundation are presented in Figs. 3, 4, 5, 6, 7, 8, 9, 10 for target values
soil below the base slab of wall, the length of the rein- of reliability indices, bsli, be and bb = 2.0, 2.5, 3.0, 3.5 and
forcement. The random variables involved in three modes 4.0 against sliding, eccentricity and bearing failure modes
of failure are presented in Table 1. For evaluating the respectively.
effect of uncertainty in the performance of reinforced soil
walls, uncertainty associated with material properties has Resistance Factors for Friction Angle of Backfill (W/)
been represented by assigning a mean and standard devi-
ation in terms of coefficient of variation (COV) for each The results presented in Fig. 3 shows the effect of COV
parameter. Table 1 summarizes the assumed PDF types, values of / on resistance factors for / (W/) against sliding,
mean values (l) and COV for the random variables in a eccentricity and bearing failure modes for bsli, be and
typical design. The statistics of material properties have bb = 2.0, 2.5, 3.0, 3.5 and 4.0, and for COV of / ranges
been chosen from the range of values suggested in litera- from 0 to 20 % and for typical values of c = 18 kN/m3, /
ture (Phoon and Kulhawy [18], Duncan [19]). The mean = 32, cb = 18 kN/m3, /b = 30, c = 10 kN/m2, L/
values are selected as representative of typical soils H = 0.7, DL = 12 kN/m2, LL = 24 kN/m2, db//b = 2/3,
encountered on site (i.e., coarse-grained backfill material COV of c, cb, /b, c, db, DL, LL and L/H = 5, 5, 5, 10, 5, 15,
and more fine-grained foundation material). In the absence 20 and 2.5 % respectively. The same typical values of the
of extensive site-specific data base, the design parameters parameters are chosen for the computation of load and
are assumed to be normally distributed. However, this resistance factors which are presented in Figs. 4, 5, 6, 7, 8,
assumption is not appropriate for the random variables, /, 9, 10. The following observations can be made from the
/b, db and c due to the fact that the lower bound values results presented in Fig. 3. The magnitudes of resistance
become negative. It can be noted that the values of shear factors are different for different failure criteria. It can be
parameters of the backfill and interface friction are always noted from Fig. 3 that for constant value of COV of /,
greater than or equal to zero. Therefore the random vari- resistance factors, W/ reduce significantly with increase in
ables, /, /b, db and c are assumed to be log normally values of bsli, be and bb from 2.0 to 4.0 in all the failure
distributed in the present study. modes. It can also be noted from Fig. 3 that for constant
Ellingwood [20] and Nowak [21] reported that the COV values of bsli, be and bb, significant reduction in the mag-
of DL and LL can be taken as 15 and 25 % respectively. nitude of W/ from 1.0 to 0.4 can be observed with increase
Further stated that dead loads are normally distributed with in COV of / from 0 to 20 % in all the failure modes.
a bias factor of 1.05, while the LL exhibits lognormal Therefore, the influence of COV of / for the stability of
distribution with a bias factor of 1.15. Since the dead loads wall may not be neglected as it is a significant random
can be estimated more accurately than live loads, the COV variable.
for live load is usually higher than that used for dead loads. The results presented in Fig. 4 shows the effect of COV
Therefore the maximum value of COV of LL is considered values of /b on resistance factors for / (W/) against sliding
as 30 % in the present study. The range of all input and bearing failure modes for bsli and bb = 2.0, 2.5, 3.0,

123
Indian Geotech J (October–December 2013) 43(4):292–302 299

Table 1 Mean and COV values


Random variable Statistics
of random parameters
considered in the study Mean (l) COV (%) Distribution Source

c 18 kN/m3 5 Normal Duncan [19]


cb 18 kN/m3 5 Normal Duncan [19]
/ 32 5–20 Log-Normal Phoon and Kulhawy [18]
/b 30 5–20 Log-Normal Phoon and Kulhawy [18]
db//b 2/3 5–20 Log-Normal Phoon and Kulhawy [18]
c 10 kN/m2 0–40 Log-Normal Duncan [19]
DL 12 kN/m2 15 Normal Ellingwood [20] and Nowak [21]
LL 24 kN/m2 0–30 Log-Normal Ellingwood [20] and Nowak [21]
L/H 0.7 2.5 Uniform Chalermyanont and Benson [1]

Table 2 Summary of load and resistance factors computed for dif-


ferent failure modes
Load and resistance factors Failure modes
gsli (x) ge (x) gb (x)

W/ X X X
W/b X – X
Wc X – X
gLL X X X

Fig. 4 Influence of COV values of /b on resistance factors for the


friction angle of the backfill (W/) against sliding and bearing failure
modes for bsli, be and bb = 2, 2.5, 3, 3.5 and 4

Eq. (4)). Consequently there is an increase in the magni-


P P
tude of sum of the resisting forces ( Fr ). Therefore Fd
should be increased to maintain the same value of target
reliability index (bsli). It can be possible, when magnitude
of earth pressure coefficient, Ka increases with increase in
/ value. Hence, resistance factors (W/) should be increased
with increase in COV of /b.
Fig. 3 Influence of COV values of / on resistance factors for the
friction angle of the backfill (W/) against sliding, eccentricity and Resistance Factors for Friction Angle of Foundation
bearing failure modes for bsli, be and bb = 2, 2.5, 3, 3.5 and 4
Soil (W/b ) and Cohesion (Wc)

3.5 and 4.0, and for COV of /b ranges from 0 to 20 %. It Figure 5 shows the effect of COV values of /b on resis-
can be observed from Fig. 4 that for constant mean value tance factors for /b (W/b ) against sliding and bearing
of bsli and bb, the resistance factors (W/) for sliding and failure modes for bsli and bb = 2.0, 2.5, 3.0, 3.5 and 4.0,
bearing stability modes, should be increased when COV of and for COV of /b ranges from 0 to 20 %. It may be
/b increases from 0 to 20 %. This is because, as the observed from Fig. 5 that for constant value of COV of /b,
magnitude of COV of /b increases, the magnitude of resistance factors, W/b reduce significantly with increase in
design value of /b (/b ) required to provide the target value values of bsli and bb from 2.0 to 4.0 in sliding and bearing
of reliability index (bsli = 3.0) is also increases (refer to failure modes. Further, for constant values of bsli and bb,

123
300 Indian Geotech J (October–December 2013) 43(4):292–302

Fig. 5 Influence of COV values of /b on resistance factors for the Fig. 7 Influence of COV values of live load (LL) on load factors for
friction angle of the foundation soil (W/b ) against sliding and bearing LL against sliding, eccentricity and bearing failure modes for bsli, be
failure modes for bsli, be and bb = 2, 2.5, 3, 3.5 and 4 and bb = 2, 2.5, 3, 3.5 and 4 for COV of / = 10 %

Fig. 6 Influence of COV values of cohesion of the foundation soil (c) Fig. 8 Influence of COV values of live load (LL) on load factors for
on resistance factors for the cohesion (Wc) against sliding and bearing LL against sliding, eccentricity and bearing failure modes for bsli, be
failure modes for bsli, be and bb = 2, 2.5, 3, 3.5 and 4 and bb = 2, 2.5, 3, 3.5 and 4 for COV of / = 20 %

significant reduction in the magnitude of W/b from 1.0 to Influence of COV of Live Load Surcharge (LL)
0.4 can be observed with increase in COV of /b from 0 to on Load Factor (gLL)
20 % in sliding and bearing failure modes. It can be noted
from Fig. 6 that COV of c is an another significant variable If a wall is expected to support a surcharge load, a designer
for sliding and bearing capacity failure modes as there is a should consider the influence of distributed dead load
drastic reduction in the magnitude of Wc with the increase surcharge, traffic load and variability associated with it
in COV of c from 5 to 40 %. This can be attributed to the during the design calculations. An appropriate load factors
P
increase of sum of resisting forces ( Fr ) in sliding limit for live load surcharge intensity (LL) should be applied in
state and ultimate bearing capacity (qu) in bearing failure the wall design in order to maintain the desired safety
mode, due to increase in cohesion of foundation soil (c). levels in relation to sliding, eccentricity and bearing sta-
The magnitude of Wc reduces from 1.0 to 0.6 for different bility modes. This aspect is discussed in the framework of
values of bsli, bb and COV of c. LRFD in this section. For this purpose, the computed load

123
Indian Geotech J (October–December 2013) 43(4):292–302 301

P
the value of V and decreases the eccentricity of resultant
force (e),
P which further increase the horizontal resisting
forces ( Fr ) and ultimate bearing capacity (qu). Therefore
the magnitude of the active earth pressure due to surcharge
loading (i.e. LL 9 HKae) and eccentricity (e) should be
increased to maintain the same value of target reliability
index. This can be possible, if the magnitude of LL
increases. Hence, load factors (gLL) should be increased
with increase in COV of LL.
One more observation can be made from Figs. 7 and 8 is
that for constant values of bsli, be and bb, the magnitude of
load factor, gLL increases considerably with increase in
COV of LL from 0 to 30 %. Therefore, it can be observed
from Fig. 8 that the load factor (gLL) ranges from 1.0 to 1.8
for sliding stability mode, 1.0–2.2 for eccentricity stability
mode and 1.0–2.0 for bearing stability mode when COV of
Fig. 9 Influence of COV of / on load factors for LL against sliding, LL changes from 0 to 30 % and COV of / = 10 %. Sim-
eccentricity and bearing failure modes for bsli, be and bb = 2, 2.5, 3, ilarly, it may be noted from Fig. 8 that the load factor (gLL)
3.5 and 4
ranges from 1.0 to 1.35 for sliding stability mode, 1.0–1.5
for eccentricity stability mode and 1.0–1.55 for bearing
stability mode when COV of LL changes from 0 to 30 %
and COV of / = 20 %.

Influence of Mean and COV Values of / and /b


on Load Factor, gLL

To maintain the desired safety level of the RSW, adequate


load factors for LL should be applied by considering the
variability associated with /. For this purpose, load factors
for LL (gLL) are computed bsli and be = 2.0, 2.5, 3.0, 3.5
and 4.0, COV of / ranges from 2 to 20 % against sliding
and eccentricity failure modes. The values of load factors
(gLL) are presented in Fig. 9 for COV of DL = 15 % and
COV of LL = 20 %. It can be observed from Fig. 9 that for
constant mean value of COV of /, the load factor (gLL) for
sliding and eccentricity stability modes, should be
Fig. 10 Influence of COV of /b on load factors for LL against increased significantly when bsli and be increases from 2 to
sliding, eccentricity and bearing failure modes for bsli, be and bb = 2, 4. Further, it can be observed from Fig. 9 that for constant
2.5, 3, 3.5 and 4 values of bsli and be, the magnitude of gLL reduces sig-
nificantly with the increase in COV of / from 2 to 20 %.
Similarly, Fig. 10 is plotted to show the effect of COV of
factors are presented in Figs. 7 and 8 bsli, be and bb = 2.0, /b on load factors for LL against sliding and bearing
2.5, 3.0, 3.5 and 4.0, COV of / = 10, 20 % and for COV of failure modes for bsli, be and bb = 2.0, 2.5, 3.0, 3.5 and 4.0
LL ranges from 0 to 30 % to illustrate the influence of COV and COV of /b ranges from 0 to 20 %.
of LL on the stability of wall for typical values adopted in
the earlier sections.
The results presented in Figs. 7 and 8 show the influence Conclusions
of COV of LL on the load factors, gLL for sliding, eccen-
tricity and bearing stability modes. It can be observed from LRFD procedure that considers uncertainties associated
these figures that for constant value of COV of LL, the with the design of reinforced soil walls is illustrated in the
magnitudes of load factor, gLL show significant increase paper. The load and resistance factors developed in the
with increase in magnitudes of bsli, be and bb from 2 to 4. present study considering the variability associated with
This is because, the parameter LL significantly increases static loading, may be used to support adaptation of the

123
302 Indian Geotech J (October–December 2013) 43(4):292–302

partial factors to enable inclusion in AASHTO [12] and 4. Basha BM, Babu GLS (2009) Reliability assessment of internal
FHWA [11] specifications. However the resistance and stability of reinforced soil structures: a pseudo-dynamic
approach. Soil Dyn Earth Eng 30(5):336–353
load factors computed for the external stability must be 5. Basha BM, Babu GLS (2009) Optimum design for external
checked for the safety against internal failure modes, i.e. seismic stability of geosynthetic reinforced soil walls: a reliability
tensile and pullout of the reinforcements. The analyses based approach. J Geotech Geoenviron Eng ASCE 136(6):797–
presented in this paper lead to the following major 812
6. FHWA (2001) Load and resistance factor design (LRFD) for
conclusions: highway bridge substructures: reference manual and participant
workbook. Publication No. FHWA HI-98-032. Federal Highway
1. The magnitudes of resistance factors are different for
Administration and National Highway Institute, Washington, DC
different failure criteria. 7. AASHTO (1996) Standard specifications for highway bridges,
2. It can be noted that the COV’s of /, /b, c and LL has 16th edn. American Association of State Highway and Trans-
significant effect on load and resistance factors for all portation Officials, Washington, DC
8. Allen TM, Nowak AS, Bathurst RJ (2005) Calibration to deter-
the failure modes.
mine load and resistance factors for geotechnical and structural
3. It is found that for various magnitudes of target design. Issue Number: E-C079. Transportation Research Board,
reliability index, COVs of / and /b, the resistance Washington, DC
factors, W/ and W/b range from 0.4 to 1.0 for all the 9. Bathurst RJ, Allen TM, Nowak AS (2008) Calibration concepts
for load and resistance factor design (LRFD) of reinforced soil
failure modes, and the resistance factor for cohesion of
walls. Can Geotech J 45:1377–1392
the foundation soil (Wc) varies from 0.70 to 1.0 for 10. Kulhawy FH, Phoon KK (2009) Geo-RBD for foundations—let’s
sliding mode and 0.6–1.0 for bearing capacity mode. do it right!. In: Magued Iskander, Debra F. Laefer, and Mohamad
4. Since the dead loads can be estimated more accurately H. Hussein (eds) Contemporary topics in situ testing, analysis,
and reliability of foundations. ASCE Geotechnical Special Pub-
than live loads, the COV for live load is usually higher
lication 186, p 442–449
than that used for dead loads. Therefore due consid- 11. Basha BM, Babu GLS (2010) Load and resistance factor design
eration must be paid to determine the COV of LL as it (LRFD) approach for the reliability-based seismic design of
influences the RSW design significantly. bridge abutments. Georisk: Assess Manag Risk Eng Syst Geo-
hazards 4(3):127–139
5. It is observed from the study that the load factors (gLL)
12. Kim D, Salgado R (2012) Load and resistance factors for external
ranges from 1.0 to 1.8 for sliding stability mode, 1.0– stability checks of mechanically stabilized earth walls. J Geotech
2.2 for eccentricity stability mode and 1.0–2.0 for Geoenviron Eng ASCE 138(3):241–251
bearing stability mode when COV of LL changes from 13. Basha BM, Babu GLS (2008) Target reliability based design
optimization of anchored cantilever sheet pile walls. Can Geotech
0 to 30 % and COV of / = 10 %. Similarly when COV
J 45(4):535–548
of / = 20 %, it can be noted that load factor (gLL) 14. FHWA (2001) Mechanically stabilized earth walls and reinforced
ranges from 1.0 to 1.35 for sliding stability mode, 1.0– soil slopes: design and construction guidelines. Publication
1.5 for eccentricity stability mode and 1.0–1.55 for FHWA NHI-00-43. Federal Highway Administration and
National Highway Institute, Washington, DC
bearing stability mode.
15. Ang HS, Tang WH (1984) Probability concepts in engineering
planning and design. Wiley, New York
16. AASHTO (2007) LRFD bridge design specifications, 4th edn.
American Association of State Highway and Transportation
Officials, Washington, DC
References 17. Das BM (1999) Principles of foundation engineering, 4th edn.
Publishing Workflow System Publication, Pacific Grove
1. Chalermyanont T, Benson C (2005) Reliability-based design for 18. Phoon KK, Kulhawy FH (1999) Evaluation of geotechnical
external stability of mechanically stabilized earth walls. Int J property variability. Can Geotech J 36(4):625–639
Geomech ASCE 5(3):196–205 19. Duncan J (2000) Factors of safety and reliability in geotechnical
2. Basha BM (2009) Optimum design of retaining structures under engineering. J Geotech Geoenviron Eng ASCE 126(4):307–316
static and seismic loading: a reliability based approach. Ph.D 20. Ellingwood BR (1999) Wind load statistics for probability-based
thesis, Department of Civil Engineering, Indian Institute of Sci- structural design. J Struct Eng ASCE 125(4):453–463
ence Bangalore 21. Nowak AS (1994) Load model for bridge design code. Can J
3. Basha BM, Babu GLS (2009) Seismic reliability assessment of Civil Eng 21:36–49
external stability of reinforced soil walls using pseudo-dynamic
method. Geosynth Int 16(3):197–215

123

Você também pode gostar