Sie sind auf Seite 1von 10

Soil Dynamics and Earthquake Engineering 31 (2011) 391400

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering


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

Seismic lateral movement prediction for gravity retaining walls


on granular soils
J.C. Tiznado a, F. Rodrguez-Roa b,n
a
Department of Structural and Geotechnical Engineering, Ponticia Universidad Catolica de Chile, Chile
b
Department of Structural and Geotechnical Engineering, Ponticia Universidad Catolica de Chile, Avenida Vicuna Mackenna 4860, Codigo Postal 782-0436, Santiago, Chile

a r t i c l e in fo abstract

Article history: At present, methods based on allowable displacements are frequently used in the seismic design of
Received 11 March 2010 earth retaining structures. However, these procedures ignore both the foundation soil deformability and
Received in revised form the seismic amplication of the soil placed behind the retaining wall. Thus, they are not able to predict
15 September 2010
neither a rotational failure mechanism nor seismic induced lateral displacements with an acceptable
Accepted 27 September 2010
degree of accuracy for the most general case. In this paper, a series of 2D nite-element analyses
were carried out to study the seismic behavior of gravity retaining walls on normally consolidated
Keywords: granular soils. Chilean strong-motion records were applied at the bedrock level. An advanced non-linear
Retaining walls constitutive model was used to represent both the backll and foundation soil behavior.
Finite-element method
This elastoplastic model takes into account both the stress dependency of soil stiffness and coupling
Granular soils
between shear and volumetric strains. In unloadingreloading cycles, the non-linear shear-modulus
Wall seismic-displacements
Soilstructure interaction reduction with shear strain amplitude is considered. Interface elements were used to model
soilstructure interaction. Routine-design charts were derived from the numerical analyses to predict
the lateral movements at the base and top of gravity retaining walls located at sites with similar seismic
characteristics to the Chilean subduction zone. Thus, wall seismic rotation can also be obtained. The
developed charts consider wall dimensions, granular soil properties, bedrock depth, and seismic input
motion characteristics. As shown, the proposed charts match well with available experimental data.
& 2010 Elsevier Ltd. All rights reserved.

1. Introduction not able to predict neither a rotational failure mechanism nor


seismic induced lateral displacements with an acceptable degree
On the basis of the Newmark sliding block procedure [1], of accuracy for the most general case.
several simplied methods have been proposed for predicting In the present study, a simple and realistic method is proposed
seismic permanent displacements of earth retaining structures. to estimate both lateral displacements and rotations of gravity
A quantitative comparison of these methods was carried out by retaining walls subjected to strong earthquakes.
Cai and Bathurst [2]. They assume a rigid retaining structure The hardening soil model with small-strain stiffness (HSsmall
supported on a rigid horizontal surface, and a rigid perfectly model) proposed by Benz [9] was selected to represent the soil
plastic behavior of the backll. Seismic amplication and phase stressstrain behavior. This is an elastoplastic constitutive model
effects, not considered in these methods, can signicantly modify that takes into account both the stress dependency of soil stiffness
ground surface motions characteristics, and this point should be and coupling between shear and volumetric strains. In unloading
considered mainly for design of tall walls [3]. reloading cycles, the non-linear shear-modulus reduction with
Some extensions of those methods have considered changes in shear strain amplitude is considered. Most of the twelve
the inclination of failure surface [4], kinematic compatibility [5], geotechnical parameters required for this model can be derived
phase effects [6], and rotational displacements [7,8]. However, from results of conventional drained triaxial tests.
key aspects in the seismic performance of gravity walls such as The HSsmall model is available in the standard library of the
the deformability of both backll and foundation soil, and the commercial nite element (FE) code PLAXIS. This program was
dynamic response of the whole system have not been taken into used to perform a series of plane strain FE analyses to study the
account properly. Therefore, the current available procedures are seismic response of gravity retaining structures supported on
normally consolidated granular deposits.
A linear-elastic stressstrain behavior was assumed for the
retaining walls using a high enough Youngs modulus to simulate
n
Corresponding author. Tel.: + 56 2 3544207. a rigid structure. Interface nite elements were used to model
E-mail address: frroa@ing.puc.cl (F. Rodrguez-Roa). soilstructure interaction.

0267-7261/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2010.09.008
392 J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400

A certain construction sequence of both wall and backll Asymptotic line to the
was assumed in the analyses for obtaining the initial effective hyperbolic curve
stress eld in the medium previous to the seismic solicitation. qa
Chilean strong-motion records observed in rock sites during the qf
March 3, 1985, in central Chile earthquake, were applied at the E50

13
bedrock level. Failure line
1
Different wall heights and sand deposits with different
thicknesses and geotechnical properties were considered. Rou- 0.5qf
Eur
tine-design charts were derived from the numerical analyses to
predict the lateral movements at the base and top of gravity
retaining walls located at sites with similar seismic characteristics 1
to the Chilean subduction zone. Thus, wall seismic rotation can
also be obtained. The developed charts consider wall dimensions, Axial strain
granular soil properties, bedrock depth, and seismic input motion Fig. 2. Hyperbolic stressstrain relation for standard compression triaxial test.
characteristics. As shown, the proposed charts match well with
available experimental data. 8
For sin fm o 3=4 sin f cm 0
>
>  
>
< sin fm sin fcv
For sin fm Z 3=4 sin f and c 4 0 sin cm max ,0
>
> 1sin fm sin fcv
>
: For sin f Z 3=4 sin f and c r0
2. The HSsmall model m cm c
3
The HSsmall model is an elastoplastic constitutive model, in
which shear hardening due to plastic strains during primary su1 su3
sin fm 4
deviatoric loading is modeled by a yield surface that can su1 su3 2c cot f
isotropically expand up to the MohrCoulomb failure criterion
[9,10]. A yield cap is introduced to model compression hardening,
i.e., plastic strains due to primary compression in oedometer The angle of dilatancy at failure, c, is expressed in terms of the
loading and isotropic loading (Fig. 1). This model uses a non- peak friction angle, f, by the following relationship:
associated ow rule for shear hardening and an associated ow sin fsin fcv
rule for compression hardening. Negative values are used for sin c 5
1sin fsin fcv
compressive stresses, and positive numbers for tensile stresses.
In the particular case of standard triaxial-test loading condi-
tions, the model leads to the well known Duncan and Chang The critical state angle, fcv, is automatically computed from
hyperbolic stressstrain relationship [11,12]. The soil stiffness for Eq. (5). Therefore, PLAXIS users do not need to specify this angle.
primary loading is modeled by using the secant E50-modulus In unloadingreloading cycles, the non-linear shear-modulus
dened at 50% ultimate strength (Fig. 2). This modulus is given by reduction with shearstrain amplitude is considered by using the
 m following modied Hardin and Drnevich hyperbolic relationship
c cos fsu3 sin f [12,13]:
E50 Eref
50 1
c cos f pref sin f
G 1
6
where s3u effective conning pressure, ccohesion, f peak G0 1 0:3859g=g0:7 9
friction angle, Eref
50 secant modulus for the reference conning
pressure pref (usually equal to 100 kN/m2), and m constant where G0 shear modulus at very small shear strains (g o10  6),
dimensionless number. G secant shear modulus corresponding to a shear strain
The shear hardening ow rule is expressed as amplitude g, and g0.7 shear strain value at which the secant
shear modulus G is reduced to about 70% of G0.
e_ pv sin cm g_ p 2 In the HSsmall model, the shear-modulus reduction curve is
bounded by a certain lower limit as shown in Fig. 3. Masings rule
[14] is considered for modeling the soil hysteretic behavior in
unloadingreloading cycles.
The formulation of this model has been developed mainly for
geotechnical applications in which small cycles of shear strains
take place, i.e., when shear strains are lower than 0.001,
approximately. For a more detailed description of the HSsmall
model the reader is referred to PLAXIS Manual [12] and Benz et al.
[15]. The following parameters are required by the model:

c: effective cohesion
f: effective friction angle at peak strength
c: angle of dilatancy at failure
nur: Poissons ratio for unloadingreloading (the default value
is 0.2)
K0: coefcient of earth pressure at rest (for normally
consolidated sands, K0 1  sin f)
m: dimensionless E50-modulus exponent
Rf: failure ratio dened as qf/qa (Fig. 2) (the default value is 0.9)
ref
Fig. 1. Yield surface in the principal stress space for cohesionless soil. Eref
50 : E50-modulus at the reference conning pressure p
J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400 393

Eref
oed : tangent oedometer modulus at the reference pressure 3. FE analysis performed
(the default relation is: Eref
oed
Eref
50 )
Eref 3.1. Simulation of the retaining-wall and backll construction
ur : Youngs modulus for unloadingreloading conditions, at
the reference conning pressure (the default relation is process
Eref ref
ur 3E50 )
Two-dimensional plane-strain FE analyses were carried out by
Gref
0 : dynamic shear modulus at very small strains, at the using six-node triangular elements. After preliminary analyses, a
reference pressure
value of n5 was selected for the dimensions of the mesh (Fig. 4).
g0.7: shear strain value at which the secant shear modulus G is The discretization of the continua used for the analyses is shown
reduced to about 70% of G0.
in Fig. 5.
The initial effective stress eld in the medium, previous to the
40000 seismic solicitation, was obtained simulating a certain construc-
tion sequence of both wall and backll. Three different construc-
tion stages were considered as shown in Fig. 6.
Stage I corresponds to the assumed initial situation in which
Shear modulus G [kN/m2]

30000
the toe of an existing slope is located at a distance approximately
equal to B from the wall (B is the bottom width of the wall). Initial
stresses were obtained in all elements by applying the gravity
20000 loading in an incremental way. After this FE analysis, both nodal
displacements and strains were set equal to zero.
Stage II corresponds to the incremental construction of the
HSsmall wall. Placement of 1 m-thick layers in successive steps was
10000 Hardin-Drnevich
cut-off considered.
Stage III corresponds to the incremental construction of the
backll. Also, placement of 1 m-thick layers in successive steps
0 was herein considered.
1E-6 1E-5 0.0001 0.001 0.01 Once the construction process was completed, both nodal
displacements and strains were set equal to zero. In these FE
analyses, the nodes located at the lateral boundaries of the mesh
Fig. 3. Cut-off of the shear-modulus degradation curve as used in the HSsmall
were restrained against horizontal displacements. At the bottom
model (adapted from [12]).
boundary, the nodes were restrained in both horizontal and
vertical direction.
Backfill top
3.2. Dynamic analysis
Ground surface level H

The seismic response analysis was based on input accelero-


grams applied at the bottom boundary of the mesh, i.e., at the
T
bedrock level.
Bedrock level Absorbent boundaries were used at the lateral sides of the FE
mesh, according to the formulation proposed by Lysmer and
nT n (H+T) Kuhlemeyer [16].
To avoid loss of accuracy during the propagating wave
Fig. 4. Dimensions of the FE mesh. phenomenon, the height of the elements was limited to a
maximum value given by [17]
To 3 (H+T) lmin Vs
hmax 7
To 3T 5 5fmax
H
where lmin is the wavelength associated with the highest
frequency of the input signal fmax and Vs is the shear wave
T velocity.
The HSsmall model is almost linear at very small strains with
practically no hysteretic material damping, and this can cause
unrealistic resonance during numerical modeling of wave propa-
Fig. 5. Discretization of the continua for retaining-wall analyses. gation. For this reason, viscous damping was considered by means

Stage I: Initial stresses Stage II: Wall construction Stage III: Backfill placement

Fig. 6. Retaining-wall construction sequence.


394 J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400

of the damping matrix, [C], expressed as [18] whereas d 2f/3, or d f, depending on the analyzed case, was
used for the wallsubsoil interface. To prevent overlap between
C auM buK 8
adjacent elements, a sufciently high value for the normal
where M and K are the mass and stiffness matrices of the system, stiffness was selected. More details about the formulation of
respectively. The coefcients a0 and b0 can be obtained from the this type of element can be found in the PLAXIS reference manual
relationship [12].
( )  
au 2x on om
9
bu on om 1 4. Seismic input motion
where on and om are natural circular frequencies corresponding
4.1. Chilean subduction zone
to signicant vibration modes of the system response, e.g., rst
and second modes, and x is the viscous damping ratio. Some
authors such as Matasovic [19] and Lanzo and Vucetic [20] The seismicity and tectonic of central Chile is mainly
recommend the use of a constant viscous damping ratio of 1.54%. characterized by the subduction of the oceanic Nazca plate
In the present research a value of x 3% was used. beneath the continental South American lithosphere. The contact
between these two plates produces very large earthquakes, with
epicenters located along the Chilean coast, and depths ranging
3.3. Material properties between 15 and 50 km [23]. The Chilean subduction zone shares
many characteristics with subduction zones in Cascadia,
The present analysis involved three different clean sands southwestern Japan, and Colombia, where comparably young
composed of subrounded particles: a sand of medium consistency oceanic lithosphere is also subducting [24].
(M sand) with a relative density (Dr) equal to 60%, a medium-
dense sand (MD sand) with Dr 70%, and a dense sand (D sand)
4.2. Earthquake records used
with Dr 80%. The properties used for these sands are listed in
Table 1.
The recorded bedrock motions of the March 3, 1985, in central
When shear wave velocity measurements are not available, G0
Chile earthquake (Ms 7.8) were input into the FE model. Their
can be estimated from empirical relationships. Hardin and Black
maximum accelerations, velocities, and displacements are pre-
[21] proposed the following expressions:
sented in Table 2. The average elastic response spectrum for 5% of
2:17e2 p damping is shown in Fig.7, where it can be observed a short
G0 kPa 6908 su0 kPa Rounded particles 10
1 e predominant period of about 0.15 s. This is a clear evidence of the
signicant high frequency content of the Chilean rock motion
2:97e2 q

records.
G0 kPa 3230 s00 kPa Angular particles 11
1 e Each one of the recorded horizontal seismic components
where evoid ratio and su0 mean effective conning pressure. (Table 2) was scaled to get both 0.2 and 0.3 g of maximum
The mean values of Gref acceleration (amax). Then, time history analyses could be
0 computed from Eqs. (10) and (11) were
used as input data for the study sands. performed using 14 different accelerograms corresponding to
In addition, the non-linear empirical relation between shear strong earthquake motions. The used records, with an equal
modulus and shear strain amplitude proposed for sands by Seed spacing of 200 points/s, were baseline corrected.
and Idriss [22] was used for estimating the g0.7 values included in
Table 1.
5. Gravity retaining walls analyzed and results obtained
A linear elastic behavior of the concrete material of the wall
was assumed, using a Poissons ratio of 0.2 and a high enough
Different walls were analyzed in order to involve usual cases
Youngs modulus to simulate a rigid structure.
encountered in practice. Their geometry and dimensions are
Soilstructure interaction was modeled by means of six-node
illustrated in Fig. 8 and Table 3, respectively. In addition, different
interface elements. The interface mechanical behavior was
homogeneous and normally consolidated granular deposits were
represented using a rigid perfectly plastic constitutive law. A
considered. These soil deposits were composed of different sands
friction angle d f/2 was used for the wallbackll interface,
with thicknesses ranging from 8 to 20 m. A summary of the
characteristics assumed for the four wall-models analyzed is
Table 1 listed in Table 4. Thus, a total of 56 numerical analyses was
Index properties and HSsmall-model parameters of study sands.
carried out (each wall model was subjected to fourteen different
Parameter M sand MD sand D sand Unit accelerograms).
Ground motion amplication plays an important role in the
Gs 2.7 2.6 2.7 dynamic analysis of retaining walls. In Fig. 9, the input UTFSM-N70E
g 16 17 18 kN/m3
c 0.0 0.0 0.0
f 351 371 401 Table 2
c 51 71 101 Recorded bedrock-motions of the 1985 Chile earthquake.
nur 0.2 0.2 0.2
K0 0.426 0.398 0.357 Station Component amax (g) vmax (m/s) dmax (m)
m 0.5 0.5 0.5
Rf 0.9 0.9 0.9 Zapallar NS 0.270 0.1122 0.0115
Eref 3.50E+ 04 4.00E +04 4.50E +04 kPa Zapallar EW 0.304 0.1346 0.0169
50
UTFSM S20E 0.165 0.0631 0.0125
Eref
oed
3.50E+ 04 4.00E +04 4.50E +04 kPa
UTFSM N70E 0.176 0.1460 0.0311
Eref 1.05E+ 05 1.20E + 05 1.35E + 05 kPa
ur Quintay NS 0.236 0.1251 0.0282
Gref
0
7.45E+ 04 9.57E + 04 9.82E + 04 kPa Quintay EW 0.260 0.1934 0.0348
g0.7 2.00E  04 2.00E  04 2.00E  04 Pichilemu NS 0.259 0.1168 0.0373
J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400 395

3.5 increased up to 2.14 times the value of amax of the bedrock motion,
whereas the value of amax computed in the backll zone (at 0.7H
3 above the wall base) increased up to 1.19 times the value of amax
computed at the wall base. An increase of the accelerations induced
on the soil behind the wall will cause an increase of the seismic
2.5
Amplification factor

lateral pressures acting on the retaining structure. This seismic


amplication phenomenon is not considered by the traditional
2
methods based on the simple sliding-block procedure, i.e., they can
lead to wall designs on the unsafe side.
1.5 On the other hand, due to subsoil deformability, typical wall
movements combine rotation and lateral displacements. If the
1 seismic lateral movement of the wall is characterized by both
the absolute horizontal displacement at the wall base, xB, and at
0.5 the wall top, xT, then the seismic permanent rotation, ar, can be
obtained as follows:
x x 
0
ar tan1 T B 12
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 H
Period (s)
The wall-model-III deformed mesh obtained for the UTFSM-
Fig. 7. Average elastic response spectrum for the Chilean recorded bedrock
N70E observed record, scaled to amax 0.3 g, is shown in Fig. 10.
motions (x 5%).
The permanent horizontal displacements were 0.55 and 0.33 m at
the top and base of this medium-high wall, respectively, and a
permanent rotation of 2.51 was obtained. A comparison between
B' these horizontal displacements and those given by the well-
known RichardsElms [25] and WhitmanLiao [26] procedures
was performed. For using these procedures, the maximum values
of both acceleration and velocity were derived from the motion
characteristics at the wall base given by the FE analysis. The yield
horizontal acceleration, ay, for the wallbackll system was
H
obtained on the basis of the MononobeOkabe relationship
[27,28] by following the iterative method described in detail by
Kramer [29]. As shown in Table 5, the displacements given by
these simplied procedures fall within the range of values 0.33
0.55 m of the FE analysis.
B

Fig. 8. Wall geometry considered.


6. Proposed charts to predict permanent displacements

As mentioned above, the seismic behavior of gravity retaining


Table 3 walls is quite complex. It depends on many variables such as the
Dimensions of study walls. dynamic response of the soil foundation and backll, the inertial
response of the wall, soilstructure interaction, and character-
W1 W2 W3
istics of the seismic motion. Therefore, if the horizontal displace-
Wall height, H (m) 3 5 8 ments for a given earthquake are expressed as a function of a
Bottom width, B (m) 1.8 3 4.5 unique design factor Fd for the sake of simplicity, such a factor
Top width, B (m) 0.3 0.5 0.5 should depend on those variables. Thus, different expressions for
Fd were tried for including the effects of most signicant variables
such as seismic pressures, earthquake characteristics, ground
motion amplication, and dynamic properties of the involved
Table 4
Characteristics of the wall models considered. soils. From this trial-and-error process based on the 56 numerical
analyses performed, high values of the correlation coefcient R2
Model I Model II Model III Model IV were obtained by expressing Fd as
Wall geometry W1 W2 W2 W3 EMO =H=pa 0:7 FT 0:7 dmax =dref
Backll M sand M sand M sand M sand Fd
0:2  1000 13
Foundation soil MD sand D sand M sand MD sand GB =pa 0:7 0:642 tan d1 =f
Thickness of foundation-soil (m) 8 20 20 12
d1/f for interface at wall base 1 2/3 2/3 1
where Hwall height, dmax peak bedrock displacement (input
d2/f for wallbackll interface 0.5 0.5 0.5 0.5 seismic motion characteristic), dref reference displacement
( 0.1 m), GB mean dynamic shear modulus (at very small shear
strains) of the foundation soil between the level of the wall base
and a depth B below that level, d1 frictional angle at the wall
record scaled to amax 0.3 g is compared to the horizontal subsoil interface, pa the atmospheric pressure, EM  O the hor-
accelerations computed for the analyzed wall-models II and IV. As izontal component of total active thrust obtained from the
expected, the bedrock motion is not only modied by the dynamic MononobeOkabe formula [27,28], and FT dimensionless factor
response of the soil underlying the wall, but also by the seismic that relates the input earthquake predominant-period to the
amplication of the backll. The value of amax at the wall base predominant period of the involved soils. The factor FT is obtained
396 J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400

0.3

0.2

0.1

aH (g)
0

-0.1

-0.2

-0.3
0 10 20 30 40 50 60 70
Time (s)

0.7 0.7
aH at wall base (g)

0.5 0.5

aH at wall base (g)


0.3 0.3
0.1 0.1
-0.1 -0.1
-0.3 -0.3
-0.5 -0.5
-0.7 -0.7
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time (s) Time (s)

0.6 0.6
aH in backfill at 0.7-H (g)
aH in backfill at 0.7-H (g)

0.4 0.4
0.2 0.2
0 0
-0.2 -0.2
-0.4 -0.4
-0.6 -0.6
-0.8 -0.8
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time (s) Time (s)

Fig. 9. Observed and computed horizontal acceleration (aH) time histories: (a) UTFSM-N70E observed record scaled to a maximum acceleration of 0.3 g. (b) Computed
accelerations at wall-model-II base. (c) Computed accelerations at wall-model-IV base. (d) Computed accelerations in the wall-model-II backll-zone at 0.7H above the wall
base. (e) Computed accelerations in the wall-model-IV backll-zone at 0.7H above the wall base.

Table 5
Horizontal displacements predicted by simplied procedures for wall-model III
(UTFSM-N70E record scaled to 0.3 g).

amax (g) vmax (m/s) ay (g) dR  E (m)a dW  L (m)b

0.52 0.48 0.159 0.45 0.33

a
Predicted value by Richards and Elms method [25].
b
Predicted value by Whitman and Liao method [26].

where TE is the predominant period of the input bedrock motion


Fig. 10. Wall-model-III deformed mesh for the UTFSM-N70E observed record obtained from the elastic response spectrum for a damping ratio
scaled to amax 0.3 g. of 5%, and TS is the fundamental vibration period of the backll
subsoil system estimated from the expression [29]

4Hbackfill Hsubsoil
TS 15
as follows: Vs
1 where Hbackll and Hsubsoil are the thicknesses of backll and
FT 14
1:5TE =TS 0:5 foundation soil, respectively, and V s is the shear wave velocity
J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400 397

(at very small shear strains) representative of the backllsubsoil where


system. This velocity V s is computed as the weighted average
value of existing layers in both subsoil and backll. The inclusion cos2 fyb
KAE h pi2
of factor FT in Eq. (13) was for incorporating seismic amplication cos y cos2 b cosd2 b y 1 sinf d2 sinfy=cosd2 b ycos b
effects in a simple approximate way.
17
If the ground surface behind the wall is assumed to be
horizontal, EM  O is given by the following expression:
 
1 kh
EMO gH2 1kv KAE cosd2 b 16 y tan1 18
2 1kv

Table 6
Variables involved in the calculation of Fd and displacements obtained from FE analyses for wall-model I.

Record amax (g) EMO (kN/m) FT dmax (m) GB (kPa) d1/f Fd XT (m) XB (m)

Zapallar NS 0.2 20.58 1.732 0.0085 7.193E +04 1.00 0.191 0.058 0.021
Zapallar NS 0.3 22.93 1.732 0.0128 7.193E +04 1.00 0.311 0.108 0.048
Zapallar EW 0.2 20.58 1.732 0.0111 7.193E +04 1.00 0.250 0.053 0.025
Zapallar EW 0.3 22.93 1.732 0.0167 7.193E +04 1.00 0.406 0.094 0.045
UTFSM S20E 0.2 20.58 1.520 0.0152 7.193E +04 1.00 0.312 0.070 0.030
UTFSM S20E 0.3 22.93 1.520 0.0227 7.193E +04 1.00 0.503 0.141 0.067
UTFSM N70E 0.2 20.58 1.732 0.0353 7.193E +04 1.00 0.795 0.100 0.046
UTFSM N70E 0.3 22.93 1.732 0.0530 7.193E +04 1.00 1.287 0.224 0.106
Quintay NS 0.2 20.58 1.130 0.0239 7.193E +04 1.00 0.399 0.034 0.010
Quintay NS 0.3 22.93 1.130 0.0358 7.193E +04 1.00 0.645 0.071 0.025
Quintay EW 0.2 20.58 1.339 0.0268 7.193E +04 1.00 0.504 0.043 0.015
Quintay EW 0.3 22.93 1.339 0.0402 7.193E +04 1.00 0.816 0.096 0.038
Pichilemu NS 0.2 20.58 1.457 0.0288 7.193E +04 1.00 0.575 0.060 0.027
Pichilemu NS 0.3 22.93 1.457 0.0432 7.193E +04 1.00 0.930 0.113 0.053

Table 7
Variables involved in the calculation of Fd and displacements obtained from FE analyses for wall-model II.

Record amax (g) EMO (kN/m) FT dmax (m) GB (kPa) d1/f Fd XT (m) XB (m)

Zapallar NS 0.2 57.16 1.219 0.0085 9.604E + 04 0.67 0.200 0.040 0.019
Zapallar NS 0.3 63.70 1.219 0.0128 9.604E + 04 0.67 0.325 0.077 0.041
Zapallar EW 0.2 57.16 1.219 0.0111 9.604E + 04 0.67 0.262 0.034 0.017
Zapallar EW 0.3 63.70 1.219 0.0167 9.604E + 04 0.67 0.425 0.060 0.032
UTFSM S20E 0.2 57.16 1.137 0.0152 9.604E + 04 0.67 0.341 0.050 0.023
UTFSM S20E 0.3 63.70 1.137 0.0227 9.604E + 04 0.67 0.550 0.099 0.049
UTFSM N70E 0.2 57.16 1.219 0.0353 9.604E + 04 0.67 0.832 0.146 0.090
UTFSM N70E 0.3 63.70 1.219 0.0530 9.604E + 04 0.67 1.348 0.365 0.221
Quintay NS 0.2 57.16 0.955 0.0239 9.604E + 04 0.67 0.475 0.033 0.016
Quintay NS 0.3 63.70 0.955 0.0358 9.604E + 04 0.67 0.767 0.072 0.039
Quintay EW 0.2 57.16 1.058 0.0268 9.604E + 04 0.67 0.572 0.061 0.036
Quintay EW 0.3 63.70 1.058 0.0402 9.604E + 04 0.67 0.926 0.153 0.096
Pichilemu NS 0.2 57.16 1.110 0.0288 9.604E + 04 0.67 0.636 0.050 0.031
Pichilemu NS 0.3 63.70 1.110 0.0432 9.604E + 04 0.67 1.029 0.104 0.066

Table 8
Variables involved in the calculation of Fd and displacements obtained from FE analyses for wall-model III.

Record amax (g) EMO (kN/m) FT dmax (m) GB (kPa) d1/f Fd XT (m) XB (m)

Zapallar NS 0.2 57.16 1.185 0.0085 7.174E+ 04 0.67 0.241 0.059 0.031
Zapallar NS 0.3 63.70 1.185 0.0128 7.174E+ 04 0.67 0.391 0.112 0.063
Zapallar EW 0.2 57.16 1.185 0.0111 7.174E+ 04 0.67 0.315 0.045 0.021
Zapallar EW 0.3 63.70 1.185 0.0167 7.174E+ 04 0.67 0.511 0.088 0.044
UTFSM S20E 0.2 57.16 1.110 0.0152 7.174E+ 04 0.67 0.412 0.087 0.043
UTFSM S20E 0.3 63.70 1.110 0.0227 7.174E+ 04 0.67 0.663 0.173 0.086
UTFSM N70E 0.2 57.16 1.185 0.0353 7.174E+ 04 0.67 1.001 0.243 0.145
UTFSM N70E 0.3 63.70 1.185 0.0530 7.174E+ 04 0.67 1.621 0.546 0.330
Quintay NS 0.2 57.16 0.941 0.0239 7.174E+ 04 0.67 0.577 0.069 0.034
Quintay NS 0.3 63.70 0.941 0.0358 7.174E+ 04 0.67 0.932 0.150 0.076
Quintay EW 0.2 57.16 1.037 0.0268 7.174E+ 04 0.67 0.692 0.098 0.052
Quintay EW 0.3 63.70 1.037 0.0402 7.174E+ 04 0.67 1.120 0.230 0.129
Pichilemu NS 0.2 57.16 1.085 0.0288 7.174E+ 04 0.67 0.768 0.078 0.047
Pichilemu NS 0.3 63.70 1.085 0.0432 7.174E+ 04 0.67 1.242 0.150 0.091
398 J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400

Table 9
Variables involved in the calculation of Fd and displacements obtained from FE analyses for wall-model IV.

Record amax (g) EM  O (kN/m) FT dmax (m) GB (kPa) d1/f Fd XT (m) XB (m)

Zapallar NS 0.2 146.33 1.307 0.0085 1.144E + 05 1.00 0.226 0.078 0.030
Zapallar NS 0.3 163.07 1.307 0.0128 1.144E + 05 1.00 0.367 0.136 0.054
Zapallar EW 0.2 146.33 1.307 0.0111 1.144E + 05 1.00 0.295 0.048 0.020
Zapallar EW 0.3 163.07 1.307 0.0167 1.144E + 05 1.00 0.478 0.088 0.039
UTFSM S20E 0.2 146.33 1.206 0.0152 1.144E + 05 1.00 0.382 0.078 0.027
UTFSM S20E 0.3 163.07 1.206 0.0227 1.144E + 05 1.00 0.615 0.153 0.059
UTFSM N70E 0.2 146.33 1.307 0.0353 1.144E + 05 1.00 0.937 0.222 0.102
UTFSM N70E 0.3 163.07 1.307 0.0530 1.144E + 05 1.00 1.518 0.516 0.249
Quintay NS 0.2 146.33 0.990 0.0239 1.144E + 05 1.00 0.522 0.066 0.022
Quintay NS 0.3 163.07 0.990 0.0358 1.144E + 05 1.00 0.844 0.138 0.053
Quintay EW 0.2 146.33 1.111 0.0268 1.144E + 05 1.00 0.635 0.099 0.039
Quintay EW 0.3 163.07 1.111 0.0402 1.144E + 05 1.00 1.028 0.236 0.104
Pichilemu NS 0.2 146.33 1.174 0.0288 1.144E + 05 1.00 0.709 0.075 0.036
Pichilemu NS 0.3 163.07 1.174 0.0432 1.144E + 05 1.00 1.148 0.150 0.077

+3.20
Sand fill (SP-SM) (Layer 1, N60 = 22)
0.00

Mixture of sand and rock fill


(Layer 2, N60 = 43)

-11.80
-13.80 Rock fill (Layer 3)
-15.80 Sand fill (SP-SM) (Layer 4, N60 = 66)

Dense sand stratum


Fig. 11. Chart for estimating seismic lateral displacement at wall base.
(Layer 5, N60 40)

-26.80

4.0 m
Bedrock
Scale

Fig. 13. Quay-wall geometry and soil prole characteristics at Valparaiso port
(Site #3).

Table 10
Geotechnical parameters considered in the back analysis of the quay wall.

Parameter Layer 1 Layer 2 Layer 3 Layer 4 Layer 5

g (kN/m3) 18 20 22 20 20
f 401 411 501 421 371
N60 22 43 66 40
G0 (kPa) 5.89E + 04 1.20E+ 05 3.12E+ 05a 1.67E+ 05 1.61E +05
Fig. 12. Chart for estimating seismic lateral displacement at wall top.
a
It was assumed K2,max 150 to estimate modulus G0 in the rock ll layer [33].

b angle formed by the wall inside-face and the vertical direction, 7. Validation of the proposed charts
d2 frictional angle at the wallbackll interface, and khg and kvg
are the horizontal and vertical pseudostatic accelerations, respec- To verify the proposed charts on the basis of the experi-
tively. For design purposes, kvg is not usually considered, and khg mental evidence, it was carried out a back analysis of the seismic
ranges from one-third to one-half the peak acceleration of the performance of a quay wall located at Valparaiso port
input motion [29]. A value of kh 0.5amax/g was used for (Site #3) during the Chile earthquake of March 3, 1985. This
computing EM  O in the present research. 15 m-high wall, is composed of pre-cast concrete blocks [30].
Values of the main variables involved in the calculation of Fd, A permanent horizontal displacement of 0.15 m was observed
and the displacements obtained from the FE analyses for the wall at the top of this quay wall, towards the sea, after the earth-
models studied, are given in Tables 69. Then, design charts were quake.
developed for predicting seismic horizontal absolute displace- In Fig. 13, the soil prole, quay wall geometry, water level,
ments in terms of Fd at the base and top of the wall, as shown in and available SPT data are shown. To estimate the dynamic
Figs. 11 and 12, respectively. shear modulus G0, the expression proposed by Seed et al. [31]
J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400 399

Table 11
Variables involved in the calculation of the factor Fd for the back analysis of the quay wall.

H (m) kh kh0 geq (kN/m3) d1/f d2/f Pw (kN/m) EM  O (kN/m)a FT GB (kPa) Fd

15 0.088 0.13 12.8 0.6 0.5 70.1 528 1.15 1.67E+ 05 1.08

a
The value of the hydrodynamic force, Pw, is herein included.

was used: soils under strong seismic motions observed in the Chilean
q subductive environment.
2 2
G0 lb=ft 1000K2,max su0 lb=ft 19 It was found that seismic amplication effects in both soil
1/3 foundation and backll play a very important role on the
where K2,max 20[(N1)60] , and (N1)60 CNN60 is the normalized
permanent displacements of the wall. The accelerations induced
SPT N-value (1 lb/ft2 0.049 kPa). To compute the overburden
on the soil behind the wall can differ substantially from the
pressure correction factor, CN, the expression proposed by Liao
accelerations applied at the bedrock level. The induced accelera-
and Whitman was used [32]:
 0:5 tions cannot be known beforehand because they depend on
pa several variables such as dynamic properties of the foundation
CN 20
suv soil and backll, earthquake characteristics, and the inertial
response of the wall. Moreover, it was found that due to soil
where suv is the effective vertical overburden pressure.
deformability the typical wall movements combine rotations and
Then, the existing soil layers at Site #3 were characterized by
lateral displacements. Thus, an approximate method was devel-
means of the parameters given in Table 10.
oped on the basis of a unique design factor Fd, for the sake of
The international code for design of port structures PIANC [34]
simplicity, to include all those variables.
was used to evaluate the earth and water pressures against the
Design charts were derived from the numerical analyses to
quay wall. According to this reference, the total active thrust
predict both absolute lateral displacements at the base and top of
due to a partially submerged backll is calculated by using the
gravity retaining walls located at sites with similar seismic
following modied seismic coefcient for an equivalent
characteristics to the Chilean subduction zone. Thus, wall seismic
homogeneous soil in the MononobeOkabe formula:
rotation can also be obtained. The proposed charts consider the
gsat Hw2 gHsur
2
2gHw Hsur most relevant factors in the system response: wall geometry,
kuh kh 21
gb Hw2 gHsur
2 2gH H
w sur granular soil properties, bedrock depth, and seismic input motion
And, the equivalent unit weight of soil is calculated as characteristics. As shown, the proposed charts match well with
 2 "   # available experimental data.
Hw Hw 2
geq gb g 1 22
H H
Acknowledgements
In these equations, gb is the backll buoyant unit weight, gsat is the
backll saturated unit weight, g is the unit weight of the soil
above the water level, Hw and Hsur are the backll heights below The authors thank the School of Engineering of the Ponticia
and above the water level, respectively, and HHw + Hsur is the Universidad Catolica de Chile, Santiago, Chile, for the support
height of the wall. given to the rst author during the development of his MSc thesis.
The Westergaard relationship [35] was used to calculate the The authors also thank the anonymous reviewers for their
hydrodynamic force acting on the seaward side of the wall. The valuable comments. The license for using the program PLAXIS
magnitude of this force per unit length of the wall is given by was purchased from PLAXIS B.V. Corporation.

7
Pw k g H2 23
12 h w w References
where gw is the unit weight of water.
The UTFSM University located at a distance of about 3 km from [1] Newmark NM. Effect of earthquakes on dams and embankments. Geotechni-
que 1965;15(2):13960.
the Valparaiso port, is the nearest seismograph station. From the [2] Cai Z, Bathurst RJ. Deterministic sliding block methods for estimating seismic
two horizontal components measured at this site, the UTFSM- displacements of earth structures. Soil Dyn Earthquake Eng 1996;15:25568.
N70E record was selected as seismic input motion for the back [3] Nadim F, Whitman RV. Seismically induced movement of retaining walls.
J Geotech Eng ASCE 1983;109(7):91531.
analysis because the direction of this component is approximately [4] Zarrabi-Kashani K. Sliding of gravity retaining walls during earthquakes
parallel to the vertical plane in which the seismic movement of considering vertical accelerations and changing inclination of failure surface.
the quay wall took place. In Table 11 the values of the main SM thesis, Department of Civil Engineering, MIT; 1979.
[5] Stamatopoulos C, Velgaki E, Modaressi A, Lopez-Caballero F. Seismic
variables involved in the calculation of the factor Fd are given. For displacement of gravity retaining walls by a two-body model. Bull Earth-
the computed value of Fd 1.08, it was obtained directly from the quake Eng 2006;4:295318.
proposed chart (Fig. 12) a horizontal displacement of 0.18 m at [6] Steedman RS, Zeng X. The inuence of phase on the calculation of pseudo-
static earth pressure on a retaining wall. Geotechnique 1990;40(1):10312.
the top of the quay wall. As expected, the effect of hydrodynamic
[7] Zeng X, Steedman RS. Rotating block method for seismic displacement of
action led to a slight overestimation of the measured gravity walls. J Geotech Geoenviron Eng 2000;126(8):70917.
displacement because dynamic water pressures were not [8] Choudhury D, Nimbalkar S. Seismic rotational displacement of gravity walls
by pseudodynamic method. Int J Geomech ASCE 2008;8(3):16975.
included in the database used for the charts.
[9] Benz T. Small-strain stiffness of soils and its numerical consequences. PhD
thesis. Universitat Stuttgart; 2007.
[10] Schanz T, Vermeer PA, Bonnier PG. Formulation and verication of the
8. Conclusions hardening-soil model. In: Brinkgreve RBJ, editor. Beyond 2000 in computa-
tional geotechnics. Rotterdam: Balkema; 1999. p. 28190.
[11] Duncan JM, Chang CY. Nonlinear analysis of stress and strain in soils. J Soil
This paper presents the results obtained from a series of FE Mech Found Div ASCE 1970;96(SM5):162953.
analyses about the behavior of gravity retaining walls on granular [12] Plaxis 2D. Reference Manual, version 9.0; 2008.
400 J.C. Tiznado, F. Rodrguez-Roa / Soil Dynamics and Earthquake Engineering 31 (2011) 391400

[13] Hardin BO, Drnevich VP. Shear modulus and damping in soils: design [24] Heaton TH, Hartzell SH. Earthquake hazards on the Cascadia subduction zone.
equations and curves. J Soil Mech Found Div ASCE 1972;98(SM7):66792. Science 1987;236(4798):1628.
[14] Masing G. Eignespannungen und verfestigung beim messing. In: Proceedings [25] Richards R, Elms DG. Seismic behavior of gravity retaining walls. J Geotech
of the second international congress on applied mechanics, Zurich, Switzer- Eng ASCE 1979;105:44964.
land; 1926, p. 3325. [26] Whitman RV, Liao S. Seismic design of retaining walls. In: Proceedings of the
[15] Benz T, Vermeer PA, Schwab R. A small-strain overlay model. International eighth world conference on earthquake engineering, San Francisco; 1984.
Journal for Numerical and Analytical Methods in Geomechanics 2009;33: p. 53340.
2544. [27] Mononobe N, Matsuo H. On determination of earth pressure during
[16] Lysmer J, Kuhlemeyer RL. Finite dynamic model for innite media. J Eng Mech earthquakes. In: Proceedings of the world engineering congress, vol. 9,
ASCE 1969;95(4):85977. Tokyo; 1929.
[17] Kuhlemeyer RL, Lysmer J. Finite element method accuracy for wave [28] Okabe S. General theory on earth pressure and seismic stability of retaining
propagation problems. J Soil Mech Found Div ASCE 1973;99(5):4217. walls and dams. J Jpn Soc Civ Eng 1924;10(6):1277323.
[18] Chopra AK. Dynamics of structures: theory and applications to earthquake [29] Kramer SL. Geotechnical earthquake engineering. Upper Saddle River, NJ:
engineering. Prentice-Hall international series in civil engineering and Prentice Hall; 1996.
engineering mechanics. Englewood Cliffs, NJ: Prentice Hall; 1995. [30] Ortigosa P. Some geotechnical effects. In: Monge J, editor. El sismo de marzo
[19] Matasovic N. Seismic response of composite horizontally-layered soil 1985. Chile. Santiago: Acero Comercial SA; 1985. p. 17996. in Spanish.
deposits. PhD thesis, University of California, Los Angeles; 1993. [31] Seed HB, Wong RT, Idriss IM, Tokimatsu K. Moduli and damping factors for
[20] Lanzo G, Vucetic M. Effect of soil plasticity on damping ratio at small cyclic dynamic analysis of cohesionless soils. J Geotech Eng ASCE 1986;112(2):
strains. Soils Found 1999;39(4):12141. 101632.
[21] Hardin BO, Black WL. Vibration modulus of normally consolidated clays. J Soil [32] Liao S, Whitman RV. Overburden correction factor for SPT in sand. J Geotech
Mech Found Div ASCE 1968;94:35368. Eng ASCE 1986;112(3):3737.
[22] Seed HB, Idriss IM. Soil moduli and damping factors for dynamic response [33] Uddin N, Gazetas G. Dynamic response of concrete-faced rockll dams to
analyses. College of Engineering University of California, Berkeley; 1970. strong seismic excitation. J Geotech Eng ASCE 1995;121(2):18597.
[23] Leyton F, Ruiz J, Campos J, Kausel E. Intraplate and interplate earthquakes in [34] PIANC. Seismic design guidelines for port structures. Rotterdam: Balkema; 2001.
Chilean subduction zone: a theoretical and observational comparison. Phys [35] Westergaard HM. Water pressures on dams during earthquakes. Trans ASCE
Earth Planet Inter 2009;175(2):3746. 1933;98:41832.

Das könnte Ihnen auch gefallen