Você está na página 1de 9

Soil Dynamics and Earthquake Engineering 25 (2005) 451459

www.elsevier.com/locate/soildyn

Evaluation of seismic displacements of quay walls


Sung-Ryul Kima,1, In-Sung Jangb,2, Choong-Ki Chungc,3, Myoung-Mo Kimc,*
a
Department of Civil Engineering, Dong-A University, 840 Hadan2-dong, Saha-gu, Busan 604-714, South Korea
Korea Ocean Research and Development Institute, Sa2-dong, Sanglok-gu, Ansan-si, Kyoungi-do 426-744, South Korea
c
School of Civil, Urban and Geosystem Engineering, Seoul National University, San 56-1, Shinlim-dong, Gwanak-gu, Seoul 151-742, South Korea
b

Accepted 29 March 2005

Abstract
A new simplified dynamic analysis method is proposed to predict the seismic sliding displacement of quay walls by considering the
variation of wall thrust, which is influenced by the excess pore pressure developed in backfill during earthquakes. The method uses the
Newmark sliding block concept and the variable yield acceleration, which varies according to the wall thrust, to calculate the quay wall
displacement.
A series of 1 g shaking table tests were executed to verify the applicability of the proposed method, and a parametric study was performed.
The shaking table tests verified that the proposed method properly predicts the wall displacement, and the parametric study showed that the
evaluation of a realistic wall displacement is as important as the analysis of liquefaction potential for judging the stability of quay walls.
q 2005 Elsevier Ltd. All rights reserved.
Keywords: Earthquakes; Newmark sliding block method; Seismic design; Shake table tests; Quay walls

1. Introduction
Recent design codes for quay walls demand performance-based designs. To obtain a safe and economical
aseismic design, these codes require that the seismic
performance of the walls be evaluated based on the
permanent wall displacement after an earthquake. The
seismic displacement of the quay walls can be evaluated by
dynamic analyses or simplified dynamic analyses based on
the Newmark sliding block concept. Comprehensive results
for the dynamic behavior of soil and walls can be obtained
from dynamic analyses. However, dynamic analyses not
only require much effort and time but also the proper values
for the various input parameters, which are difficult to
obtain.
* Corresponding author. Tel.: C82 2 880 7348; fax: C82 2 875 6933.
E-mail addresses: sungryul@daunet.donga.ac.kr (S.-R. Kim), isjang@
kordi.re.kr (I.-S. Jang), geolabs@snu.ac.kr (C.-K. Chung), geotech@snu.
ac.kr (M.-M. Kim).
1
Tel.: C82 51 200 7622; fax: C82 51 207 4654.
2
Tel.: C82 31 400 6325; fax: C82 31 408 5823.
3
Tel.: C82 2 880 7347; fax: C82 2 875 6933.

0267-7261/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2005.03.002

The simplified dynamic analyses based on the Newmark


sliding block concept are widely used for preliminary
designs because they can easily evaluate the wall displacement by using basic design parameters such as the weight of
the wall, the internal friction angle of the backfill, and the
frictional coefficient at the bottom of the wall. Richard and
Elms and Whitman and Liao [1,2] proposed simplified
dynamic analyses based on the Newmark sliding block
concept to evaluate the seismic displacement of the wall. In
their methods, yield acceleration is defined as the wall
acceleration, when the factor of safety of the wall for sliding
becomes 1.0, and the wall displacement is presumed to
occur if the ground acceleration exceeds the yield acceleration. However, these analyses cannot consider the
variation of wall thrust due to the development of excess
pore pressure in the backfill when they determine the yield
acceleration; therefore, previous analyses are inappropriate
for the design of quay walls with saturated backfill soils
where high excess pore pressure can develop during
earthquakes.
Several researchers suggested degrading yield acceleration models, in which yield acceleration decreased as a
function of shear deformation for geosynthetic cover
analyses [3] or as a function of the magnitude of excess
pore pressure for saturated slope analyses [4].

452

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

Notation
amax
ay
cB
CSR
Dr
FD
FDF
FDI
FDY
FED
FFWD
FI
FR
FS
FST
FTH

maximum amplitude of ground acceleration


yield acceleration
cohesive stress between bottom of wall and
foundation
cyclic stress ratio required to initiate
liquefaction
relative density of backfill
fluctuating component of dynamic thrust
fluctuating component of dynamic thrust after
liquefaction
fluctuating component of dynamic thrust without excess pore pressure
dynamic thrust
fluctuating component of dynamic earth force
fluctuating component of dynamic water force
acting on front of wall
inertia force of wall
resisting force between bottom of wall and the
foundation soil
non-fluctuating component of dynamic thrust
static thrust
wall thrust acting on wall

To account for this excess pore pressure in the


calculation of wall sliding displacement, a new simplified
dynamic analysis method, which still utilizes the Newmark
concept but varies the yield acceleration according to the
varying wall thrust, is proposed.
A parametric study is performed to analyze the effect of
the input parameters of this new method on the seismic wall
displacement, and the proposed method is verified by
comparing the predicted displacements with the results of a
series of shaking table tests.

FWD
H
kh
L
M
N
NL
rd
ru
W
fB
gsat
gsub
q
sv
sv 0
tcyc

fluctuating component of dynamic water force


acting on back of wall
wall height
horizontal seismic coefficient
length of contact surface between bottom of wall
and foundation
mass of wall
number of loading cycles
number of loading cycles required to produce
initial liquefaction
stress reduction factor
excess pore pressure ratio
weight of wall
interface friction angle between bottom of wall
and foundation
saturated unit weight of soil
submerged unit weight of soil
constant representing soil properties and test
condition
total stress in backfill
effective stress in backfill
cyclic shear stress

the factor of safety for sliding becomes 1.0, and evaluates


the block displacement by double integration of the ground
acceleration, which exceeds the yield acceleration.
Several methods to calculate the block displacement by
integration of the ground acceleration have been proposed.
Among these, the integration method by Wilson and Keefer
[5] (Fig. 1) has been widely used. Fig. 1(a) shows the time
history of the ground acceleration, which is compared to the
yield acceleration ay. The wall displacement begins to occur

2. Development of new displacement calculation method


2.1. Assumptions
The following assumptions were used in the proposed
method.
Quay walls always fail in a sliding mode. This
assumption is valid only for dense foundation soils.
Wall displacement occurs in a forward direction only.
This should be a reasonable assumption for most cases
since the wall can hardly move towards the backfill soils
during shaking.
2.2. Newmark sliding block method
The Newmark sliding block method defines the yield
acceleration as the amplitude of the block acceleration when

Fig. 1. Development of displacement for actual earthquake ground motion


[5].

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

453

calculated by Eq. (2).

Fig. 2. Orientation and location of force components [6].

at point X where the ground acceleration becomes larger


than ay (Fig. 1(c)). The cumulative permanent displacement
of the block is obtained by the double integration of the
difference between the ground and yield accelerations until
the velocity of the wall becomes zero (Fig. 1(c)).
2.3. Determination of yield acceleration
The method proposed in this study evaluates the yield
acceleration according to the varying wall thrust and
therefore, this method is distinct from the previous ones.
The wall thrust on a quay wall (FTH) is a resultant of diverse
force components as shown in Fig. 2 [6]: static water forces
acting on the back and front of the wall, inertia force of the
wall (FI), dynamic water force on the front of the wall
(FFWD), static thrust on the back of the wall before shaking
(FST), and dynamic thrust on the back of the wall, which
develops during shaking (FDY). In this research, we
assumed that the water levels on both sides of the wall
were the same, and thus, the static water forces acting on the
both sides of the wall were not considered. The wall thrust
FTH can be obtained by summing the various force
components as shown in Eq. (1). The resisting force (FR)
of the wall, which comes from the frictional force between
the bottom of the wall and the foundation soil, can be

FTH Z FI C FFWD C FST C FDY

(1)

FR Z cB L C W tan fB

(2)

where LZlength of contact surface between bottom of wall


and foundation, WZweight of wall per running unit length,
cBZcohesive stress between bottom of wall and foundation,
and fBZinterface friction angle between bottom of wall
and foundation.
The wall displacement begins to occur when the wall
thrust FTH exceeds the resisting force FR (Eq. (3a)). The
inertia force of the wall (FI) at this point can be obtained by
multiplying the mass of the wall (M) by the yield
acceleration of the wall (ay) (Eq. (3b)). Finally, ay is
obtained by Eq. (3c). However, ay is also needed to
calculate FFWD and FDY on the right-side of Eq. (3c).
Therefore, the yield acceleration can only be determined by
an iterative calculation
FTH Z FI C FFWD C FST C FDY R FR

(3a)

FI Z FR FFWD C FST C FDY Z M !ay

(3b)

ay Z

FR K FFWD C FST C FDY


M

(3c)

where MZmass of wall


2.4. Determination of force components acting on wall
The time histories of force components acting on walls
have to be evaluated to determine the time history of the
yield acceleration, as is shown in Eq. (3). The force
component evaluating methods are summarized in Table 1
as suggested by Kim et al. [6]. Kims method requires

Table 1
Evaluation of force components acting on quay walls ([6])
Force component

Evaluation method

Inertia force of wall (FI)


Dynamic water force acting on front side of wall (FFWD)
Static thrust (FST)
Dynamic thrust (FDYZFDCFS)
Fluctuating component (FD)

Mass of wall!time history of input acceleration


Westergaard equation
Coulomb method
FDZFDI!(1-ru)CFDF!ru
FDIZFWDCFEDK(FICFFWD)
7
FDF Z 12
kh gsat H 2
where FDI: fluctuating component of dynamic thrust without excess pore
pressure
FDF: fluctuating component of dynamic thrust after liquefaction
FED: fluctuating component of dynamic earth force acting on back side of wall
FWD: fluctuating component of dynamic water force acting on back side of wall
ru: excess pore pressure ratio in the backfill
H: wall height
kh: horizontal seismic coefficient
gsat: the saturated unit weight of soil
FS ZK 12 gsub KAS H 2 ru C 12 gsub H 2 ru
where gsub: submerged unit weight of backfill soil
KAS: static active earth pressure coefficient calculated by Coulomb method

Non-fluctuating component (FS)

454

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

the time histories of the wall acceleration and the excess


pore pressure ratio, ru in the backfill as input for the
evaluation of the force components. The latter can be
evaluated by either a simple empirical formula like Eq. (4)
or a dynamic analyses or a laboratory test such as a shaking
table test.
Eq. (4), which was suggested by De Alba et al. [7],
relates ru with the number of loading cycles of the
earthquake under consideration (N).
"  
#
1 1 K1
N 1=q
ru Z C sin
2
K1
(4)
2 p
NL
where NLZnumber of loading cycles required to produce
initial liquefaction corresponding to the cyclic stress ratio,
CSR (Zcyclic stress required to initiate liquefaction/vertical effective stress in soil) and qZconstant representing
soil properties and test condition (commonly, qZ0.7)
The CSR can be calculated by Eq. (5) [8]
CSR Z

tcyc
a
s
Z max v0 rd
sv0
g sv

(5)

where tcycZcyclic shear stress, amaxZmaximum amplitude


of ground acceleration, sv, sv 0 Ztotal stress and effective
stress in backfill, respectively, and rdZstress reduction
factor.

3. Parametric study
3.1. Input parameters
A parametric study was performed using the previously
described model to analyze the wall behavior under various
combinations of input parameters, using the proposed
method. Fig. 3 shows an example quay wall used for the
parametric study. The backfill was made with sand, whose
effective grain size was 0.10 mm. The coefficient of
permeability was 1.0!10K4 m/s. Relative densities of the
backfill (Dr) were varied to be 40, 60, and 70%. The
interface friction angles between the bottom of the wall and
the foundation soil (fB) were varied to 25, 30, 35, and 408.
Input acceleration was a 1 Hz sine wave. The number of

Fig. 3. Example quay wall for parametric study.

Fig. 4. Cyclic strength curve of backfill sand.

loading cycles of the input acceleration was set to 10, which


corresponds to the design earthquake magnitude of 6.5 in
Korea. The amplitude of the input acceleration (amax) was
varied to 0.072, 0.10, 0.12, 0.14, 0.15, and 0.20 g. The
0.072 g was obtained by converting the amplitude of
irregular earthquake wave of 0.11 g into the amplitude of
regular sine wave. The amplitude of 0.11 g is the amplitude
of the bedrock outcrop motion suggested in the Seismic
Design Code of Korea [9]. The 0.20 g is the maximum
probable amplitude after the bedrock acceleration is
amplified through the backfill soils.
To obtain the excess pore pressure ratio, ru which is one
of the input parameters, Eq. (4) was used. Fig. 4 shows the
cyclic strength curves obtained by the cyclic triaxial tests for
the backfill sands of various relative densities. The CSR
(cyclic stress ratio) at mid-depth of the backfill (depthZ
7.5 m) was calculated by Eq. (5) by inputting 1.0 for rd
because the acceleration amplitude in backfill was assumed
to be constant. Fig. 4 also shows how NL is obtained after
CSR is calculated by Eq. (5).
Fig. 5 shows the time histories of the calculated excess
pore pressure ratios, using Eq. (5) and Fig. 4, for various
relative densities at amaxZ0.10 g. Liquefaction of the
backfill occurred at 3.3 s for DrZ40% and 7.3 s for DrZ
60%. Meanwhile, in the backfill with DrZ70%, liquefaction
did not occur until the end of shaking.

Fig. 5. Time histories of excess pore pressure ratios (amaxZ0.10 g).

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

455

Fig. 6. Time histories of force components acting on wall (amaxZ0.10 g, DrZ40%, fBZ408).

3.2. Results of parametric study


Fig. 6 shows the time history of each force component for
the example wall with DrZ40% of backfill soil. In the
comparison of Fig. 6(a) and (d), the backfill thrust shown in
Fig. 6(d) follows the pattern of increase of ru in Fig. 6(a) in
the first part up to the time of liquefaction, after which the
backfill thrust oscillates and maintains a constant neutral
value.
Fig. 7 shows the time histories of the input acceleration
and the yield acceleration at amaxZ0.10 g. At the beginning
of shaking, the calculated yield acceleration is much larger
than the input acceleration for any relative density of

backfill. As cyclic loading continues, the yield acceleration


decreases and finally becomes smaller than the input
acceleration at 3.3 s for DrZ40% and at 7.3 s for DrZ
60% after the backfill soil liquefies. From this result,
permanent wall displacements are anticipated for the cases
of DrZ40 and 60%.
If the effect of excess pore water pressure is not
considered in the determination of the yield acceleration
as is in the existing methods, the yield acceleration does not
change from the initial value as is shown in the same figure
for DrZ40% and ruZ0. Thus, previous methods will
predict zero wall displacement for the example quay wall,
which is obviously an erroneous result.
Fig. 8 shows the cumulative wall displacements for the
various densities of backfills at amaxZ0.10 g and fBZ408.

Fig. 7. Time histories of input and yield accelerations (amaxZ0.10 g,


fBZ408).

Fig. 8. Cumulative wall displacements with time calculated by new


displacement calculation method (amaxZ0.10 g, fBZ408).

456

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

Fig. 9. Final displacements of wall with respect to amplitude of input acceleration, relative density of backfill and interface friction angle at bottom of wall.

Fig. 9 shows the final displacement of the wall for


various combinations of input parameters. Even if the
backfill liquefied at amaxZ0.072 g with DrZ40%, the final
displacement of the wall was only 9.6 cm with fBZ358 and
0.7 cm with fBZ408. On the other hand, the wall
displacement of 59 cm occurred at amaxZ0.12 g with
DrZ70% and fBZ258, where liquefaction did not occur
in the backfill. In addition, it can be observed that the
displacements calculated by the proposed model are very
sensitive to the interface friction angle. Therefore, it is
important to properly evaluate the frictional resistance
between a wall and foundation.
In general, port structures are designed to prevent the
liquefaction in backfill. However, the parametric study
showed that the wall may keep its stability even if the
liquefaction of the backfill occurs, and the wall can be
unstable when liquefaction does not occur in the backfill.
Therefore, wall displacement must be evaluated with the
consideration of the excess pore pressure development in
the backfill in the design stage of quay walls.

4. Verification of proposed method


4.1. Test set-up and procedure
The proposed method was verified by comparing the
calculated wall displacements with the results of 1 g shaking

table tests. The dimension of the soil box was 194 cm long,
44 cm wide, and 60 cm high, and the model wall was
17.5 cm long, 42.0 cm wide, and 26.4 cm high. Fig. 10(a)
shows the test section and the instrumentation and Fig. 10(b)
shows the input acceleration. Eight water pressure transducers, seven accelerometers, three LVDTs, and three load
cells were installed. The three load cells were installed in
shape of a triangle between the back plate and the main body
of the model wall to measure the thrust acting on the backside of the wall. The time histories of the horizontal
displacement and the rotation angle of the wall were
calculated by measuring three relative distance variations
between the wall and the fixed tape measure type LVDTs.
The amplitude of the sinusoidal input motion at 5 Hz was
increased linearly up to 0.2 g during the initial 5 s, and the
final amplitude was maintained for the next 5 s, as shown in
Fig. 10(b).
The model soil was Joomoonjin sand, whose average
particle size was 0.55 mm and uniformity coefficient was
1.37. The maximum and minimum dry unit weights of the
sand were 16.7 and 13.9 kN/m3, respectively. The loose
backfill of 20% relative density was prepared by the water
sedimentation method. The internal friction angle and the
saturated unit weight of the backfill soil were about 308 and
18.9 kN/m3, respectively. The permeability coefficient of the
backfill soil was measured to be 4.1!10K4 m/s by the
constant head permeability test. A dense foundation layer
was made by preshaking the foundation soil. The relative

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

457

Fig. 10. Test section and input acceleration for shaking table test.

density and the internal friction angle of the foundation layer


were about 90% and 408, respectively. The interface friction
angle between the foundation soil and the bottom of the wall
was estimated by pulling tests. The interface friction angle
increased with the wall movement velocity. The average
value of the interface friction angle for the velocity range of
the wall movement in the shaking table tests was about 288.
The shaking table tests were performed for two walls of
identical geometry but of different unit weights, 23.0 and
25.7 kN/m3. In the latter, water was situated in the space
where load cells were installed (Fig. 10(a)).

the shaking continued because the volume of backfill soil


near the wall tended to increase during the outward
movement of the wall.
Fig. 12 shows the time histories of the calculated yield
acceleration and the measured input and wall accelerations.
Only the upper and the lower limit values are plotted in the
figure. When the yield acceleration is greater than the input
acceleration, the wall acceleration is the same as the input
acceleration. But when the yield acceleration becomes less
than the input acceleration, the wall acceleration fluctuates
due to the dynamic interaction between the backfill soil with
pore water and the wall.

4.2. Comparison with results of shaking table tests


The time history of the horizontal seismic coefficient kh
was obtained by non-dimensionalizing the time histories of
the input acceleration. The average excess pore pressure,
which is the average value of the excess pore pressures
measured from the two water pressure transducers (P2 and
P3) installed on the back side of the wall, was used to obtain
the time history of the excess pore pressure ratio in the
backfill ru. The time histories of kh and ru measured during
the tests were used to predict the displacements of the wall.
Fig. 11 shows the time histories of the excess pore
pressure ratio, in which it is seen that the excess pore
pressure reached its maximum value at around 4.55.0 s.
The excess pore pressure decreased after about 5 s although

Fig. 11. Measured excess pore pressure ratios in backfill soil.

458

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

Fig. 12. Comparison between yield acceleration and measured values of input acceleration and wall acceleration.

Fig. 13. Comparisons between measured and predicted wall displacements.

Fig. 13 shows the comparisons between the measured


and the predicted displacements of the walls of two
different unit weights. The measured wall displacement
started to occur at around 4 s which is the time when the
excess pore pressure ratio increased rapidly and reached
almost its maximum value (Fig. 11). The final differences
of the horizontal displacements at the top and the bottom of
the wall were about 0.4 cm (0.98 to vertical) for the wall of
unit weight of 25.7 kN/m3 and 1.0 cm (28 to vertical) for
the wall of unit weight of 23.0 kN/m3, which are small
compared with the final horizontal displacements of 5.5
and 8.3 cm, respectively. Therefore, this observation
satisfies the assumption of the proposed model in that
only the sliding failure of walls occurs. The calculated final
displacements of the walls compared very well with the
measured values: 5.2 cm for the wall of unit weight of
25.7 kN/m3 and 7.6 cm for the wall of unit weight of
23.0 kN/m3.
Thus, it is believed that the proposed method predicted
the quay wall behavior properly in terms of its cumulative
displacement with time.

5. Conclusions
The conclusions of this study are as follows.
1. A new displacement calculation method was proposed
which considers the effect of the excess pore pressure
developed in backfill. This method basically uses the
Newmark sliding block concept but varies the yield
acceleration according to the varying wall thrust.
2. The parametric study showed that the evaluation of
realistic wall displacements under earthquakes is as
important as the analysis of liquefaction potential for
judging the stability of quay walls.
3. It was verified from a series of 1 g shaking table tests that
the proposed method properly predicts the wall displacement.

Acknowledgements
The financial support from the Ministry of Maritime
Affairs and Fisheries (MOMAF) in support of this work is
gratefully acknowledged.

S.-R. Kim et al. / Soil Dynamics and Earthquake Engineering 25 (2005) 451459

References
[1] Richards R, Elms D. Seismic behavior of gravity retaining walls.
J Geotech Eng Div 1979;105(4):44964.
[2] Whitman RV, Liao S. Seismic design of retaining walls. Miscellaneous
paper GL-85-1, US Army engineer waterways experiment station,
Vicksburg, MS 1985.
[3] Matasovic N, Kavazanjian Jr E, Giroud JP. Newmark seismic
deformation analysis for geosynthetic interfaces. Geosynthetics Int
1998, Special Issue Geosynth Earthq Eng 1998;5(12):23764.
[4] Giovanni B, Ernesto C, Michele M. Seismic response of submerged
cohesionless slopes. In: Proceedings of the fourth international
conference on recent advances in geotechnical earthquake engineering,
San Diego, CA. Paper no. 7.07; 2001.

459

[5] Wilson RC, Keefer DK. Dynamic analysis of a slope failure from the 6
August 1979 Coyote Lake, CA, earthquake. Bull Seismol Soc Am
1983;73(3):86377.
[6] Kim SR, Kwon OS, Kim MM. Evaluation of force components acting
on gravity type quay walls during earthquakes. Soil Dyn Earthq Eng
2004;24(11):85366.
[7] De Alba P, Chan CK, Seed HB. Determination of soil liquefaction
characteristics by large-scale laboratory tests. In: UCB/EERC-75/14,
Berkeley. Earthquake Engineering Research Center, University of
California; 1975.
[8] Seed HB, Idriss IM. Simplified procedure for evaluating soil
liquefaction potential. J Soil Mech Found Div 1971;97(9):124973.
[9] Ministry of Construction and Transportation. Seismic design code of
Korea, Seoul, Korea; 1997 (in Korean).

Você também pode gostar