You are on page 1of 9

Computers and Geotechnics 118 (2020) 103297

Contents lists available at ScienceDirect

Computers and Geotechnics


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

Research Paper

Seismic rotational stability analysis of reinforced soil retaining walls T


a b,⁎ c
Peng Xu , Kianoosh Hatami , Guanlu Jiang
a
Key Laboratory of High-Speed Railway Engineering of Ministry of Education, School of Civil Engineering, Southwest Jiaotong University, Chengdu 610031, China
b
School of Civil Engineering and Environmental Science, University of Oklahoma, 202 W. Boyd St., Room 334, Norman, OK 73019, USA
c
Key Laboratory of High-Speed Railway Engineering of Ministry of Education, School of Civil Engineering, Southwest Jiaotong University, Chengdu 610031, China

ARTICLE INFO ABSTRACT

Keywords: Reinforced soil retaining walls have been widely constructed all over the world. Although the limit equilibrium
Geosynthetics (LE) method is common practice in design, a growing body of experimental and analytical evidence has de-
Reinforced soil retaining walls monstrated that it could lead to overly conservative (and hence, uneconomical) design. In this paper, a kinematic
Limit analysis limit analysis method, combined with a pseudo-dynamic method of analysis, is developed for rotational stability
Rotational stability
analysis of reinforced soil retaining walls based on the log-spiral failure mechanism. The proposed analysis
Pseudo-dynamic seismic analysis
method is validated against the results from different pseudo-static methods. However, the rotational stability
Log-spiral slip plane
and the corresponding yield acceleration coefficients from the proposed method are shown to vary with time,
and with factors such as the backfill shear modulus and the frequency of ground motion. Parametric analyses are
also conducted to study the influences that pseudo-dynamic analysis parameters, reinforcement properties, wall
height, and soil strength have on rotational stability and failure geometry of reinforced soil retaining walls.
Results indicate that optimum value of ultimate reinforcement strength is greater when the reinforcement is
more widely spaced. Additionally, the cross-sectional area of the rotational failure mass is smaller for walls with
lower-strength reinforcement and larger vertical spacing.

1. Introduction static seismic analysis of reinforced soil walls, the earthquake load is
typically assumed as an equivalent static load with a constant magni-
In recent decades, several numerical and physical studies have been tude. Huang and Wang [7] calculated the seismic displacement of the
carried out to develop a better understanding of the seismic perfor- Tanata wall in Hyogo-ken Nambu earthquake using a pseudo-static
mance of reinforced soil retaining walls [1]. Zarnani et al. [2] compared method and determined that the influence of vertical acceleration on
their numerical simulation results with those reported by El-Emam and the wall displacement was insignificant. Nouri et al. [8] used the hor-
Bathurst [3]. Both studies indicated that reinforcement loads can be izontal slice method (HSM) by considering the sliding wedge as an
significantly influenced by the toe boundary conditions. Krishna and assembly of horizontal slices to evaluate seismic stability of reinforced
Latha [4] carried out numerical simulations on wrap-faced reinforced soil structures. They found that the influence of vertical ground accel-
soil retaining walls. They found that soil-reinforcement interface eration had a negligible influence on the minimum required re-
parameters had a more significant influence on the wall dynamic re- inforcement length at small horizontal seismic accelerations, which was
sponse than the reinforcement tensile modulus. Yazdandoust [5] re- also concluded by Ling and Leshchinsky [9]. Also using the HSM,
ported shaking table test results on steel-strip reinforced soil retaining Shekarian et al. [10] analyzed the resultant earth pressure and slip
walls and concluded that the deformation mode under harmonic ex- plane geometry in reinforced soil walls. They found that the slip plane
citations were highly dependent on the length of the strip reinforcement was shallower at larger horizontal ground acceleration amplitudes.
used in the models. Xu el al. [6] compared shaking table results with Vahedifard et al. [11] proposed a method to determine the required
those from pseudo-static methods, and found that the M-O method reinforcement strength for reinforced soil structures under seismic
could significantly overestimate the horizontal earth pressure relative loading. Their results indicated that backfill surcharge inclination angle
to measured values. had insignificant influence on the required reinforcement strength for
Even though model tests and numerical simulations can help de- walls with high-quality (i.e. high friction angle) backfills. Aside from
velop a more in-depth understanding of structural behavior, limit pseudo-static methods, pseudo-dynamic methods are also used to study
equilibrium (LE)-based methods are still common practice. In pseudo- seismic stability of reinforced soil structures, which consider the


Corresponding author.
E-mail addresses: bk20090201@my.swjtu.edu.cn (P. Xu), kianoosh@ou.edu (K. Hatami).

https://doi.org/10.1016/j.compgeo.2019.103297
Received 9 May 2019; Received in revised form 5 October 2019; Accepted 8 October 2019
Available online 04 November 2019
0266-352X/ © 2019 Elsevier Ltd. All rights reserved.
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

Nomenclature r, ro, rh radii of the log-spiral with respect to angles θ, θo, and θh
SOAD, SOAG, SOGD, SAGD areas of wedge OAD, OAG, OGD, and AGD
c backfill cohesion S, V cross-sectional area and volume of the sliding wedge
DI internal energy dissipation rate sv reinforcement vertical spacing
Dg, Ds internal energy dissipation rates induced by the re- t time
inforcements and the soil Ti* boundary load vector
di distance from point O to point C Tr-i mobilized reinforcement force
f pullout resistance factor Tu-i ultimate tensile strength
fa acceleration amplification factor in reinforced soil wall vi * velocity vetetor
G backfill shear modulus vp , v s primary wave velocity and shear wave velocity
H total structure height WE external work rate
h wall height We, Wg external work rates due to earthquake and gravity
h1 elevation of the bottom reinforcement layer relative to α, β inclinations of GD and AG
foundation γs unit weight of backfill
hi elevation of reinforcement layer i ϕ backfill friction angle
kh, kv horizontal and vertical seismic acceleration coefficient Ai* body force vector
kh-h’, kv-h’ horizontal and vertical seismic acceleration coefficients at εij* strain rate
time t and a height of h’ υ Poisson’s ratio of backfill
ky rotational yield acceleration coefficient ζi inclination of line OC′
kmy minimum rotational yield acceleration coefficient ρ backfill density
L length of GD σij* stress tensor
lr-i anchorage length of reinforcement layer σvi overburden stress on reinforcement layer i
li reinforcement length Ω angular frequency of base acceleration
n total number of reinforcement layers

influences of ground motion duration and frequency in the analysis 2. Methodology


[12–15].
In contrast to LE methods, limit analysis (LA) methods include the 2.1. Assumptions
mechanical performance (i.e. stress-strain relationships) of constituent
materials in the analysis, and the methodology is rigorous in the sense The following assumptions are made to calculate the rotational
that the velocity field associated with an upper-bound solution is yield acceleration coefficient, ky:
compatible with the imposed displacement [16–18]. Michalowski [19]
presented admissible velocity fields for different failure mechanisms of 1. The geometry of the reinforced soil retaining wall is represented
reinforced soil retaining walls. Porbaha et al. [20] reported centrifuge using the two-dimensional model shown in Fig. 1.
test results and used LA method to predict the failure surfaces in re- 2. The backfill is assumed as a homogeneous material, and therefore its
inforced soil wall models. Ausilio et al. [21] calculated yield accelera- seismic inertial force acts at the centroid of the sliding wedge.
tion using LA method and compared their results with those reported by 3. The homogeneous model satisfies associative flow rule associated
Ling et al. [22]. Mojallal et al. [23] proposed a method based on the with the Coulomb failure criterion.
upper-bound theorem of limit analysis to calculate the sliding stability 4. The rotational stability is analyzed assuming that the sliding wedge
and displacement of reinforced soil retaining walls. Their calculations has a log-spiral geometry, which has also been used in several past
confirmed that the wall stability increased with backfill friction angle studies [8,26–27]. As shown in Fig. 1, the wedge rotates as a rigid
and reinforcement length. Leshchinsky [24] presented a series of charts body about the center O with an angular velocity ω, and the failure
for the design of reinforced soil bridge abutments using the LA method. surface in the polar coordinates system is described using the fol-
His results confirmed earlier findings that the reinforcement vertical lowing equation:
spacing has a significant influence on the bearing capacity and stability
of reinforced soil structures. The above survey indicates that the LA
method has been used to analyze the stability of reinforced soil struc-
tures, primarily under static loading. However, it has not been ex-
tensively used for seismic analysis of these structures, and specifically,
the important influences of ground motion duration and frequency have
rarely been included in the LA method.
In this paper, the LA method has been used in combination with the
pseudo-dynamic method of analysis to study the rotational stability of
reinforced soil retaining walls subjected to earthquakes. In contrast to
the traditional LE method, calculated values of yield accelerations from
the proposed method vary with time, and the influences of ground
motion duration, frequency, and backfill modulus could also be con-
sidered in the analysis. Calculated results from the proposed method in
this study are compared with those from other methods [21–22,25].
Parametric analyses are also conducted to study the influences that
pseudo-dynamic analysis parameters, reinforcement properties, wall
height, and soil strength have on the rotational stability and failure
geometry of vertical reinforced soil retaining walls.
Fig. 1. The log spiral failure mechanism.

2
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

r = ro e ( o ) tan (1) vs = G

where ro and r are the radii of the log-spiral curve with respect to angles vp = 2G (1 )
θo and θ, respectively, and ϕ is the backfill friction angle. Eq. (1) yields (1 2 ) (7)
the following radius value for point A at the toe of the sliding wedge:
where G, ρ, and υ are shear modulus, density, and Poisson’s ratio of the
r h = ro e ( h o) tan (2) backfill, respectively. It is worth noting that Eq. (6) merely represents a
simplified representation of ground acceleration and is not intended to
where rh is the length of the radius OA, and θh is its corresponding reflect a specific recorded motion so that it would not violate zero-stress
inclination angle relative to horizontal. boundary conditions in the analysis [30].
The following geometric relationships can also be obtained from In the analysis presented here, only half cycle of kh in Eq. (6),
Fig. 1: corresponding to its positive values, is considered due to the fact that
the rotational failure of the wall shown in Fig. 1 is caused by the kh in
h + L sin + ro sin = r h sin
o h
positive (i.e. outward) direction.
h
+ L cos = ro cos r h cos
tan o h
(3)

where h is the wall height; L is the length of GD along the inclined 2.3. Stability calculations
surcharge; and α and β are inclination angles of faces GD and AG, re-
spectively. The external work rate, WE due to gravity and ground acceleration
Substituting Eq. (1) into Eq. (3), we obtain the following equations acting on the sliding wedge AGD in Fig. 1 can be expressed as:
to determine the location of slip plane intercept at the top of the re-
WE = Wg + We=
inforced soil wall, D:
s SAGD XAGD s SAGD XAGD k v - h' + s SAGD YAGD kh - h' (8)
h
tan H
tan
ro =
cos o tan e( h ) tan cos e( h ) tan sin where Wg and We are external work rates induced by the weight of
h tan h + sin o
h wedge AGD and its inertial load due to base acceleration, respectively;
ro ( cos o e( h ) tan cos
h)
L=
tan
γs is the backfill unit weight; XAGD and YAGD are the centroidal co-
cos (4)
ordinates of wedge AGD, respectively; and SAGD is its cross-sectional
area.
Also in Fig. 1, the areas of Sections OAD, OAG, and OGD (i.e. SOAD,
2.2. Analysis SOAG, and SOGD) can be calculated as:

In this paper, limit analysis method is used to analyze the rotational SOAD =
h
r 2/2d
stability of reinforced soil retaining walls subjected to ground accel- o (9)
eration. Additionally, a pseudo-dynamic approach is used in the ana-
lysis to include the frequency and duration of ground acceleration, hr h
SOAG = sin( h + )
which is not accounted for in conventional pseudo-static methods of 2 sin (10)
analysis that are typically adopted in current design guidelines [28,29].
Using the upper-bound theorem of limit analysis in Fig. 1, the re- Lro
lationship between the external work rate, WE by the gravity and SOGD = sin( + o)
2 (11)
ground acceleration, and the dissipation rate of internal energy, DI by
the reinforcement layers and the backfill soil along a kinematically and their corresponding centroidal coordinates in the rectangular co-
admissible slip plane can be expressed as [17,18]: ordinate system shown are:

DI = ij ij dV WE = Ti*vi*dS + Ai* vi*dV (5)


3
h r sin d
V s V 0 3
XOAD = SOAD
where εij* and vi* are the strain rate and velocity in the kinematically 3
h r cos d
admissible velocity field, respectively; σij* is the stress tensor associated
0 3
YOAD = SOAD (12)
with εij*; V and S are the volume and boundary of the failure wedge,
respectively; and Ti* and Ai* are the boundary load vector and body
h
force vector, respectively. 2r h cos h +
tan
XOAG =
In the pseudo-dynamic method of seismic analysis, the horizontal 3

and vertical seismic coefficients, kh-h, and kv-h, of earthquake load acting
2r h sin h h
YOAG = 3 (13)
at time t and elevation h’ from the foundation level are calculated as
[12–13,15]:
h
2 r h cos h + + L cos
tan

( ( ))
h h XOGD = 3
kh - h' = 1 + H
(fa 1) kh sin t vs 2(r h sin h h) L sin
YOGD = 3 (14)
t ( )
h h
k v - h' = 1 + (fa 1) k v sin
H vp
(6) Once the areas of Sections OAD, OAG, and OGD and their corre-
sponding centroidal coordinates are calculated, the area of the rota-
where H is the height of the structure; kh and kv are the horizontal tional soil wedge AGD, SAGD and its centroidal coordinates can be de-
and vertical seismic coefficients, respectively, due to base acceleration; termined as:
fa is the acceleration amplification factor within the earth structure; Ω is
the angular frequency of base acceleration; and vs and vp are shear wave SOAD SOAG SOGD ( h + )
and primary wave velocities, respectively, which can be calculated SAGD =
SOAD + SOAG SOGD ( h + > ) (15)
using the following equations:

3
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

(1) h + method, which partly explains the difference in geometry observed in


SOAD XOAD SOAG XOAG SOGD XOGD the figure. Nevertheless, the predicted slip plane from the proposed
XAGD = SAGD method is within the range of geometries resulting from all limit
equilibrium methods examined, albeit with better agreements with the
SOAD YOAD SOAG YOAG SOGD YOGD
YAGD = SAGD
Sarma and Spencer methods, consistent with the corresponding ky va-
(2) h + > lues reported in Table 1.
SOAD XOAD + SOAG XOAG SOGD XOGD
XAGD = SAGD
SOAD YOAD + SOAG YOAG SOGD YOGD 3.2. Comparisons with other methods
YAGD = SAGD (16)
Predicted values of ky using the method in this study were compared
Subsequently, the elevation of the centroid of Wedge AGD from the with the results reported by Ausilio et al. [21]. Ausilio et al. [21] used
base can be calculated as: an upper bound method to analyze rotational stability of reinforced soil
h = r h cos h YAGD (17) retaining walls and validated their results against those reported by
Ling et al. [22] using a limit equilibrium-based method. Comparisons
which can be substituted in Eq. (6) to determine the horizontal and were made for the case of a reinforced soil retaining wall with H = 5 m,
vertical seismic coefficients acting at the centroid of Wedge AGD. li = 5 m, Tu-i = 20 kN/m, β = 60°, n = 10, and γs = 18 kN/m3. Results
Based on the assumptions made in Section 2.1, the internal energy in Fig. 3 indicate that ky increases with backfill friction angle, and va-
dissipation rate, DI due to reinforcement layers and the backfill soil lues of ky from the proposed method are closer to those from the Ausilio
along the slip plane can be expressed in the form: et al. [21] compared to those from the Ling et al. [22], especially in the
DI = Dg + Ds= case of backfills with smaller friction angle values.
i=n
cro2
Tr - i sin i di + (e 2( h o ) tan 1) 4. Parametric study
i=1
2 tan
(18)

where Dg and Ds are the internal energy dissipation rates induced by the In this section, results of a parametric study are reported to study
reinforcement and the soil, respectively; di is distance from point O to the effects of pseudo-dynamic analysis parameters, reinforcement de-
point C where reinforcement layer i intercepts the slip plane; c is the sign, soil strength, and wall height on predicted values of km, kmy, and
backfill cohesion; n is the total number of reinforcement layers; and ζi is slip plane geometries of vertical wall (i.e. β = 90°), with the following
the inclination angle of line OC, which can be obtained using the fol- control values: h = 6 m, γs = 18 kN/m3, υ = 0.3, G = 50 MPa, ϕ = 32°,
lowing equation: c = 0 kPa, α = 0°, li = 4.2 m, Tu-i = 90 kN/m, sv = 0.3 m, f = 0.33,
kv = 0, fa = 1.0, and T = 2π/Ω = 1.0 s. Differences in slip plane geo-
hi metries are expressed in terms of L/h and L (L is the length of GD in
lr - i = l i ro e ( i o ) tan cos r h cos h
i
tan (19) Fig. 1) in the following section.

where li is the length of reinforcement layer i; hi is its elevation relative 4.1. Influence of pseudo-dynamic analysis parameters
to base; and lr-i is its anchorage length (CC′).
The mobilized reinforcement force, Tr-i in Eq. (18) can be calculated Fig. 4 shows predicted values of ky as a function of time, t, for
using the equation: various values of fa, and G as calculated using the proposed method as
Tr - i = min[2f vi l r - i (1 k v - h'), Tu - i] (20) compared to that from a pseudo-static (i.e. Spencer) method. Results
show that:
where f is the pullout resistance factor; σvi is the overburden pressure on
reinforcement layer i; and Tu-i is its ultimate tensile strength. 1. Pseudo-static methods can only yield a constant ky value regardless
Since seismic coefficients from the pseudo-dynamic method of of the factors related to ground motion characteristics and backfill
analysis used in this study vary with time, t, as per Eq. (6), yield ac- quality (i.e. T and G). In the case of the Spencer method shown in
celeration coefficients calculated from the proposed method also vary Fig. 4, the predicted value of ky is constant and equal to 0.455 at any
with time. The rotational horizontal yield acceleration coefficient at moment.
time t, ky can be calculated by substituting Eqs. (8) and (18) into Eq. 2. In contrast to pseudo-static methods, seismic coefficients from the
(5), and the minimum value of ky in domain t is defined as kym. pseudo-dynamic method of analysis used in this study vary with
time, t, as per Eq. (6). This means that ky values from the proposed
3. Validation of the proposed method method and hence the rotational stability of the reinforced soil wall
also vary with time. Results in Fig. 4 also show that the values of ky
3.1. Comparisons with SLOE/W [25] from the proposed method initially decrease within the half cycle to
a minimum value equal to kmy at t = T/4 but then increase with t
The analytical method proposed and described in this paper was toward the end of the half cycle. The kmy value for the case of no
validated against results from SLOPE/W [25] using different limit- acceleration amplification in the backfill (i.e. fa = 1) is consistent
equilibrium analysis methods for the case of a wall with h = 6 m, with the value from the pseudo-static method regardless of T or G
γs = 18 kN/m3, ϕ = 32°, c = 0 kPa, β = 90°, α = 0°, li = 4.2 m, values, which means that the existing design approaches using
Tu-i = 90 kN/m, sv = 0.3 m, f = 0.33, h1 = 0, kv = 0, and fa = 1.0.
Values of rotational yield acceleration coefficient, ky from different Table 1
analysis methods are presented in Table 1, and the corresponding Comparison of ky values from different methods.
failure surfaces are shown in Fig. 2. Results in Table 1 indicate that Proposed Analysis method used in SLOPE/W [25] in this study
except for some difference, with the Corps of Engineers value, the value method
of ky predicted from the proposed method is otherwise very close to, Morgenstern- Spencer Janbu Bishop Corps of Sarma
and well within the range of, those from all other limit equilibrium Price Engineers

methods examined. In contrast to the circular failure mechanism that ky 0.454 0.451 0.455 0.438 0.447 0.384 0.454
was assumed for the other limit equilibrium methods in Fig. 2, a log-
spiral failure mechanism was assumed in the proposed limit analysis

4
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

Fig. 2. Predicted slip planes corresponding to the calculated values of the ro-
tational yield acceleration coefficient, ky from different methods.

Fig. 3. Comparison of ky values as a function of backfill friction angle from the


proposed method with those reported by Ausilio et al. [21] and Ling et al. [22]

Fig. 4. Variation of ky with t.

pseudo-static methods could be overly conservative. In the case of


T = 2 s, the ky-t curve shifts to the right to reach its same minimum Fig. 5. Variation of kmy for walls with: (a) h1 = 0.6 m; (b) h1 = 0.3 m; (c)
value at the corresponding t = T/4. h1 = 0 m.
3. Acceleration amplification in the backfill, characterized by the value
of fa, influences the value of kmy in the sense that greater amplifi-
of G = 50 MPa and 100 MPa in Fig. 4. However, the kmy values in
cation of base acceleration in the backfill leads to a smaller kmy
both cases are essentially the same. It follows that the influence of fa
value and vice versa. However, the characteristic shape of the ky-t
on kmy is greater than those of T and G for the range of values ex-
curve remains unchanged for different fa values.
amined in this study. These observations indicate that even though
4. As can be deduced from Eq. (6), a change in G (and hence, in vs) will
the proposed method accounts for the influences of T and G on ky in
change the magnitude of phase difference in the horizontal seismic
contrast to pseudo-static methods of seismic analysis, those influ-
coefficient, kh-h,. This explains a slight difference that is observed
ences may not be fully identified due to the limitations of pseudo-
between the ky-t curves corresponding to otherwise identical cases

5
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

dynamic methods relative to true (e.g. time-history) dynamic with smaller h1 can be assumed to act as a rigid body (as is the case in
methods of analysis using numerical simulation. current design guidelines [28,29]), the rotational slip plane shifts far-
ther away from the reinforced zone for smaller h1 values, leading to
4.2. Influence of reinforcement design larger values of L/h. This means that the value of kmy and the size of its
corresponding failure zone are greatest when h1 = 0.
Fig. 5 shows variations of kmy with the reinforcement tensile Fig. 7 shows the influence of vertical spacing, sv, on the calculated
strength and length for different reinforcement arrangements char- value of kmy for different values of reinforcement length. As shown in
acterized by the elevation of the first (bottom) layer, h1 (Fig. 1). The Fig. 7a, the results from the proposed method are in close agreement
results indicate that: with those from the Spencer method, and both sets of results indicate
that the influence of sv on kmy is greater at smaller Tu-i values, which
1. The values of kmy from the proposed method in Fig. 5a increase non- means that rotational stability can be improved significantly using
linearly with Tu-i up to 60 kN/m, beyond which increasing Tu-i has tightly spaced, low-strength (and hence, less expensive) reinforcement.
no influence on the rotational stability of the reinforced wall, which Alternatively, reinforcement spacing can be increased to sv = 1 m,
is consistent with the results from the Spencer method. In other to achieve the same degree of rotational stability at reduced cost when
words, the value of kmy is sensitive to reinforcement strength over Tu-i ≥ 150 kN/m. Results in Fig. 7a also suggest optimum values of
the lower range of strength values, and a minimum amount of reinforcement strength from the proposed method equal to 60 kN/m,
strength is necessary (i.e. Tu-i/opt ≈ 60 kN/m) to optimize the ro- 90 kN/m, 100 kN/m, and 150 kN/m for walls with sv = 0.3 m, 0.4 m,
tational stability of the reinforced soil wall, which is essentially 0.6 m, and 1.0 m, respectively, which means that the optimum re-
independent of reinforcement length, li/h. This can be explained inforcement strength for rotational stability increases with sv. It is
with the magnitude of mobilized reinforcement force Tr-i in Eq. (20) worth noting that the values of optimum reinforcement strength in this
in which the second term (i.e. reinforcement tensile failure) governs paper are based on the parametric values assumed in the examples
the failure mechanism for lower Tu-i values, whereas the first term provided and could vary for different wall sizes and material properties.
becomes dominant in the case of strong reinforcement (i.e. high Tu-i Comparison of the results in Fig. 7a with those in Fig. 7b and 7c in-
values), and the governing failure mechanism changes from tensile dicates that sv has a greater influence on kmy when shorter and lower
failure to pullout. strength reinforcement is used.
2. The values of kmy from the proposed method in Fig. 5a and 5b in- Fig. 8 shows variations of L/h with sv for different values of re-
crease with li/h, albeit with a diminished return (e.g. smaller in- inforcement strength, Tu-i. Results indicate that the predicted size of
creases are observed for li/h > 0.7), which means that rotational rotational failure zone (i.e. predicted value of L/h) from the proposed
stability of the reinforced soil wall can be improved by increasing method is smaller when reinforcement layers are placed at wider spa-
reinforcement length. cing (larger sv) in the reinforced soil wall, leading to a more deformable
3. The results in Fig. 5b are similar to those in Fig. 5a. However, the and lower-strength reinforced mass. It can be inferred that the extent of
values in case of h1 = 0.3 m in Fig. 5b are slightly larger than those the failure wedge could even be contained within the reinforced zone if
when h1 = 0.6 m (i.e. Fig. 5a) due to a slight increase in the col- low-strength reinforcement is placed at too wide a spacing in the re-
lective overburden pressure on, and the anchorage length of re- inforced soil wall.
inforcement layers when they are placed at lower elevations relative
to the log-spiral slip plane shown in Fig. 1. As mentioned earlier, 4.3. Influences of backfill strength and wall height
results in Fig. 5a and 5b both indicate that the optimum reinforce-
ment strength for rotational stability in the parametric study pre- Fig. 9 shows the influence of backfill shear strength (i.e. friction
sented here is 60 kN/m. It is worth noting that the term “optimum angle, ϕ, and cohesion, c) on the predicted values of kmy. Results show
reinforcement strength” used here means that greater reinforcement that the predicted values of kmy from the proposed method are in close
strength values do not change (i.e. increase) the yield acceleration agreement with those from the Spencer method (see the case of c = 0 in
for the range of parameters examined in this study. Since re- Fig. 9), and both methods predict that kmy increases essentially linearly
inforcement cost is typically higher for higher tensile strength, the with ϕ. Results using the proposed method also predict greater values of
term “optimum reinforcement strength” in this paper indicates a kmy for backfills with larger cohesion values, even at the nominal values
strength value which is adequate for seismic stability at a minimum presented in the figure (e.g. an apparent cohesion of c = 5 kPa that
cost. could exist due to some moisture in an otherwise granular backfill).
4. In the case of h1 = 0 shown in Fig. 5c, since the entire length of the However, since fines content in current design guidelines for reinforced
bottom reinforcement layer is practically located in the anchorage
zone, the predicted values of kmy in Fig. 5c are larger than those in
Fig. 5a and 5b. Additionally, in contrast to the results for h1 > 0,
kmy values in Fig. 5c continue to increase with Tu-i beyond its op-
timum value, albeit at a slower rate, for the entire range of re-
inforcement strength values examined. This indicates that the ten-
sile strength of reinforcement products can be better utilized by
placing the first reinforcement layer as close to the foundation as
possible. Results in Fig. 5c also indicate that reinforcement strength
has a greater influence on kmy in the case of shorter reinforcement
length.

Fig. 6 shows predicted values of L/h in rotational stability calcula-


tions as a function of Tu-i using the proposed method for different values
of h1. Recall that L represents the extent of failure zone at the top of the
reinforced soil mass as shown in Fig. 1.
Results in Fig. 6 show that the values of L/h increase non-linearly
with Tu-i up to 60 kN/m, which is the same optimum value obtained
from Fig. 5. Additionally, since a well-designed reinforced zone in walls Fig. 6. Variation of L/h with Tu-i.

6
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

Fig. 8. Variation of L/h with sv for different values of reinforcement strength.

Fig. 9. Variation of kmy with backfill friction angle, ϕ and cohesion, c.

Fig. 7. Variation of kmy for walls with: (a) li/h = 1.1; (b) li/h = 0.7; (c) li/ Fig. 10. Variation of L/h with ϕ.
h = 0.5.

Results in Fig. 9 indicate that the values of kmy are larger in the case of
soil walls [28–29] are limited to small percentages to avoid water- and kv/kh = -0.5 relative to those for kv/kh ≥ 0. This means that upward
creep-related problems, the use of higher quality fill with higher fric- inertial forces are more crucial for the stability of reinforced soil re-
tion angle is deemed more reliable to increase the rotational stability of taining walls, which is consistent with the conclusions reported by
reinforced soil walls. Huang [31] and Ling and Leshchnisky [9].
Vertical acceleration also could affect the stability of reinforced soil Fig. 10 shows the relationship between L/h and ϕ for different va-
retaining wall. Ling and Leshchnisky [9] found that a larger upward lues of c. Results show that in the parametric study presented here, L/h
vertical inertial force could mobilize a larger reinforcement length to values from the proposed method generally increase with ϕ .
resist compound failure in the backfill and control displacements.

7
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

1. Compared with the pseudo-static method, the proposed method can


consider the influence of duration and frequency of earthquakes. ky
calculated using the proposed method vary with time t, and the
minimum values of ky, kmy occur when t = T/4.
2. Calculated values of kmy using the proposed method increase non-
linearly with Tu-i, whereas the influence of reinforcement length on
kmy decreases with li/h. Optimum values of reinforcement strength
from the proposed method increase with reinforcement spacing, sv,
within the range of parametric values examined in this study.
3. Different from the linear relationship between kmy and φ, the re-
lationship between kmy and h is non-linear, and the maximum values
of kmy are obtained when h = 9 m for walls in this study.
4. The rotating failure surface is relatively farther away from the wall
facing for walls with smaller value of h1, sv, and larger values of φ,
Tu-i.

Fig. 11. Variation of kmy with h. Acknowledgments

Funding and support from the China Railway Corporation (No.


2014G003-C) and the China Scholarship Council (CSC) are gratefully
acknowledged.

References

[1] Vieira CS. Review of seismic performance of geosynthetic reinforced soil retaining
walls. 6th ECCOMAS thematic conference on computational methods in structural
dynamics and earthquake engineering, Rhodes Island, Greece. 2017. p. 1058–70.
[2] Zarnani S, El-Emam MM, Bathurst RJ. Comparison of numerical and analytical
solutions for reinforced soil wall shaking table tests. Geomech Eng
2011;3(4):291–321.
[3] El-Emam MM, Bathurst RJ. Experimental design, instrumentation and interpreta-
tion of reinforced soil wall response using a shaking table. Int J Phys Model Geotech
2004;4(4):13–32.
[4] Krishna AM, Latha GM. Modeling the dynamic response of wrap-faced reinforced
soil retaining walls. Int J Geomech 2011;12(4):439–50.
[5] Yazdandoust M. Investigation on the seismic performance of steel-strip reinforced-
soil retaining walls using shaking table test. Soil Dyn Earthq Eng 2017;97:216–32.
[6] Xu P, Hatami K, Jiang G. Study on seismic stability and performance of reinforced
soil walls using shaking table tests. Geotext Geomembr 2019;103507.
Fig. 12. Variation of L (size of failure wedge) with reinforcement length for [7] Huang CC, Wang WC. Seismic displacement of a geosynthetic-reinforced wall in the
different wall heights. 1995 Hyogo-Ken Nambu earthquake. Soils Found 2005;45(5):1–10.
[8] Nouri H, Fakher A, Jones CJFP. Evaluating the effects of the magnitude and am-
plification of pseudo-static acceleration on reinforced soil slopes and walls using the
Meanwhile, the predicted size of failure zone, represented by the value limit equilibrium Horizontal Slices Method. Geotext Geomembr 2008;26(3):263–78.
[9] Ling HI, Leshchinsky D. Effects of vertical acceleration on seismic design of geo-
of L/h, also increases with c, especially for smaller ϕ values. synthetic-reinforced soil structures. Geotechnique 1998;48(3):347–73.
Fig. 11 shows predicted values of kmy as a function of wall height, h. [10] Shekarian S, Ghanbari A, Farhadi A. New seismic parameters in the analysis of
Results from the proposed method are in good agreement with those retaining walls with reinforced backfill. Geotext Geomembr 2008;26(4):350–6.
[11] Vahedifard F, Leshchinsky D, Meehan CL. Relationship between the seismic coef-
from the Spencer method, and both methods show that the relationship ficient and the unfactored geosynthetic force in reinforced earth structures. J
between kmy and h is non-linear. Geotech Geoenviron Eng 2012;138(10):1209–21.
Even though the inertial force increases with h, so does the re- [12] Steedman RS, Zeng X. The influence of phase on the calculation of pseudo-static
earth pressure on a retaining wall. Geotechnique 1990;40(1):103–12.
inforcement pullout resistance (i.e. the first term in Eq. (20)) due to [13] Nimbalkar SS, Choudhury D, Mandal JN. Seismic stability of reinforced-soil wall by
greater overburden pressure, leading to larger kmy values as h increases pseudo-dynamic method. Geosynth Int 2006;13(3):111–9.
within a lower range of wall heights. However, in taller walls, re- [14] Bathurst RJ, Hatami K, Alfaro M. Geosynthetic-reinforced structures – seismic as-
pects. In: Shukla SK, editor. Handbook of geosynthetic engineering. London: ICE
inforcement rupture, as opposed to pullout, governs stability, and the
Publishing (Institution of Civil Engineers); 2011 [Chapter 16].
mobilized force in the reinforcement is governed by the second term in [15] Zhou Y, Chen F, Wang X. Seismic active earth pressure for inclined rigid retaining
Eq. (20). In this case, kmy decreases with h for taller walls. For the set of walls considering rotation of the principal stresses with pseudo-dynamic method.
Int J Geomech 2018;18(7):04018083.
parameters used to obtain the results in Fig. 11, maximum values of kmy
[16] Donald IB, Chen Z. Slope stability analysis by the upper bound approach: funda-
correspond to h = 9 m. mentals and methods. Can Geotech J 1997;34(6):853–62.
Fig. 12 shows the values of L from the proposed method as a [17] Kim J, Salgado R, Yu HS. Limit Analysis of Soil Slopes Subjected to Pore-Water
function of reinforcement length for walls with different wall height, h. Pressures. J Geotech Geoenviron Eng 1999;125(1):49–58.
[18] Chen WF, editor. Limit analysis and soil plasticity, 10. Elsevier; 2013.
Results indicate that overall, L increases with both h and li/h, which [19] Michalowski RL. Limit analysis in stability calculations of reinforced soil structures.
means that the slip plane forms farther away from the wall facing in the Geotext Geomembr 1998;16(6):311–31.
case of longer reinforcement and/or taller walls. [20] Porbaha A, Zhao A, Kobayashi M, Kishida T. Upper bound estimate of scaled re-
inforced soil retaining walls. Geotext Geomembr 2000;18(6):403–13.
[21] Ausilio E, Conte E, Dente G. Seismic stability analysis of reinforced slopes. Soil Dyn
Earthq Eng 2000;19(3):159–72.
5. Conclusions [22] Ling HI, Leshchinsky D, Perry EB. Seismic design and performance of geosynthetic-
reinforced soil structures. Geotechnique 1997;47(5):933–52.
[23] Mojallal M, Ghanbari A, Askari F. A new analytical method for calculating seismic
The LA method combined with the pseudo-dynamic method is used displacements in reinforced retaining walls. Geosynth Int 2012;19(3):212–31.
in this paper to determine the rotating yield acceleration coefficient, ky. [24] Leshchinsky B. Limit analysis optimization of design factors for mechanically sta-
bilized earth wall-supported footings. Transp Infrastruct Geotech
The following conclusions are drawn from the proposed method:

8
P. Xu, et al. Computers and Geotechnics 118 (2020) 103297

2014;1(2):111–28. reinforced soil slopes design and construction guidelines. FHWA-NHI-10-025, Tech.
[25] GEOSLOPE, Ltd.. Slope/W for slope stability analysis: user’s guide. Calgary Rep. prepared for the Federal Highway Administration (FHWA). 2009.
(Canada): GEOSLOPE Ltd.; 2019. [29] AASHTO. LRFD bridge design specifications. 7th ed.n Washington, DC (USA):
[26] Ling HI, Leshchinsky D. Failure analysis of modular-block reinforced-soil walls American Association of State Highway and Transportation Officials; 2014.
during earthquakes. J Perform Constr Facil 2005;19(2):117–23. [30] Qin CB, Chian SC, Gazetas G. Kinematic analysis of seismic slope stability with
[27] Xie Y, Leshchinsky B, Yang S. Evaluating reinforcement loading within surcharged discretisation technique and pseudo-dynamic approach: a new perspective.
segmental block reinforced soil walls using a limit state framework. Geotext Geotechnique 2019:1–3.
Geomembr 2016;44(6):832–44. [31] Huang CC. Vertical acceleration response of horizontally excited reinforced walls.
[28] Berg RR, Christopher BR, Samtani NC. Mechanically stabilized earth walls and Geosynth Int 2013;20(1):1–12.

You might also like