Académique Documents
Professionnel Documents
Culture Documents
art ic l e i nf o a b s t r a c t
Article history: This paper describes a new pseudostatic limit equilibrium method for the design of cantilevered
Received 10 May 2012 retaining walls under seismic actions. The method has been applied in a parametric study of the effects of
Received in revised form the geometry of the wall, considering different excavated and embedded depths, of the strength of the
6 February 2013
soil, and of the contact between the soil and the wall. The pseudostatic predictions are in very good
Accepted 18 March 2013
agreement, both in terms of horizontal contact stress and bending moment distributions, with the results
of truly dynamic 2-D nite difference analyses and published experimental data. It is found that for
Keywords: increasing strengths of the soilwall system both the critical acceleration and the maximum bending
Pseudostatic limit equilibrium moment on the wall increase. In other words, a stronger soilwall system will experience smaller
Earthquake
displacements during the earthquake, but this is paid for by increasing internal forces in the wall.
Embedded retaining walls
& 2013 Elsevier Ltd. All rights reserved.
0267-7261/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.soildyn.2013.03.008
144 R. Conti, G.M.B. Viggiani / Soil Dynamics and Earthquake Engineering 50 (2013) 143150
1.0 30
limit equilibrium approaches do not provide reliable nor conser-
vative values of the yield acceleration for a pseudostatic calcula- 0.8 25
D [%]
G/G0
This work is focused on the pseudo-static limit equilibrium design Vucetic & Dobry(1991)
15
of embedded cantilevered retaining walls, for which both the 0.4 numerical
10
internal forces under dynamic conditions and the critical acceleration
0.2 5
are computed customarily with the Blum method. The main draw-
backs of this method are discussed, and a new limit equilibrium 0.0 0
0.0001 0.001 0.01 0.1 1
approach is proposed, based on results from truly dynamic 2-D nite
[%]
difference analyses [6]. Also, pseudostatic predictions are compared
with experimental data obtained from dynamic centrifuge tests on a Fig. 2. Modulus decay and damping ratio curves for the constitutive soil model.
cantilever wall with dry sand backll [12].
In the following, is the soil friction angle, is the soilwall
friction angle, is the unit weight of the soil, h is the excavation t [s]
depth, d is the embedded depth, and d0 is the location of the pivot 0 4 8 12 16 20
0.4
point; the MononobeOkabe method [15,18] is used to compute
0.2 Tolmezzo
the dynamic active (KAE) earth pressure coefcient, while passive
a [g]
0.0
static (KP) and dynamic (KPE) earth pressure coefcients are
-0.2
computed using the Lancellotta [10,11] closed form solutions. -0.4
0.4
0.2 Assisi
a [g]
2. Dynamic behaviour of cantilevered retaining walls 0.0
-0.2
-0.4
A number of studies on the dynamic behaviour of cantilevered
retaining walls are reported in the literature, most of them based 0.4
on the results of centrifuge tests [25,12,13,5]. A general observa- 0.2 Kocaeli
a [g]
(iii) permanent displacements of the wall correspond to an values computed by numerical analyses. This arises from two main
approximately rigid rotation around a pivot point located at factors: in the Blum method (i) the same pseudo-static accelera-
a depth of between 0.8 d and 0.9 d. tion is assumed for the soil in front and behind the wall and (ii) the
pivot point is assumed to be at a depth of 0.8 d from dredge
The mechanism by which the soil passive resistance is mobi- level. Both assumptions lead to an underestimation of the soil
lised below dredge level is the main factor governing the dynamic passive resistance that can be effectively mobilised in front of the
behaviour of cantilevered embedded walls, where the structure wall during an earthquake.
can rotate when a state of limit equilibrium is attained in the
adjacent soil. On the other hand, dynamic active earth pressure
distribution behind the walls can be predicted reasonably with the 3.2. Proposed method
standard MononobeOkabe theory [18,15], provided a suitable
value for the pseudostatic coefcient is adopted, taking into For a given kh okc, the horizontal earth pressure distribution
account amplication phenomena into the soil. Thus, consistently assumed in the proposed method is shown in Fig. 5 (continuous
with other experimental and numerical studies on cantilevered line). On the retained side, the soil is in active limit state down to
walls [1], a triangular distribution for the dynamic active contact d0, and in passive limit state below the rotation point. On the
stresses can always be assumed in pseudostatic calculations, just excavated side, the passive resistance of the soil is fully mobilised
as in the static case. down to a depth d; the soil is in active limit state below the pivot
point, and the horizontal contact stress decrease linearly with
depth between d and d0. To account for the numerical evidence
3. Pseudostatic approaches that the accelerations below dredge level are always much smaller
than those computed on the retained part of the soil, it is assumed
3.1. Blum method that the passive earth pressure coefcient takes its static value, KP.
This assumption is clearly conservative from the point of view of
An extension of the Blum [2] method to dynamic conditions is the structural design of the wall, but the difference between KP
generally adopted to compute both the critical acceleration and and KPE(kh 0.1) is lower than 5%.
the internal forces on the wall for khkc (Callisto and Soccodato The force equilibrium of the wall can be established by
[4]). According to this method, the dynamic active pressure, KAE, is considering the force acting on the right hand side (FRHS) and on
fully mobilised in the retained soil, down to a depth h d0 from the left hand side (FLHS) of the wall:
the surface, where d0 0.8 d (Fig. 4); a constant fraction of the
dynamic passive pressure, KPE/F, is mobilised into the soil below 1 1
F RHS K AE h d0 2 K P 2h d d0 dd0 1
dredge level, while R accounts globally for the distribution of earth 2 2
pressures around and below the pivot point. For a given kh, the
safety factor F may be computed from the moment equilibrium 1 2 1 1
F LHS K P d K P d K AE d0 d0 d K AE d d0 dd0 2
around the pivot point; the maximum value of the pseudostatic 2 2 2
coefcient is kc, which corresponds to F1. The value of 0.8 d Similarly, the moment equilibrium can be established by taking
adopted for d0 is necessary to guarantee the force equilibrium for the moment about the toe of the wall, generated by the forces
the assumed earth pressure distribution, not explicitly taken into acting on the right hand side (MRHS) and on the left hand side
account in the method. (MLHS):
As shown by Callisto and Soccodato [4] and Conti et al. [6], the
Blum method does not provide a conservative estimate of the 1 1 1
M RHS K AE h d0 2 h d0 dd0 K P 3h d 2d0 dd0 2
critical acceleration, in the sense that the pseudo-static values of 2 3 6
Mc(Mmax(kc)), i.e. the maximum bending moment computed 3
with the pseudostatic method, are about 50% lower than the
kh<kc
kh=kc (d=d0)
kh<kc h
h
kh=kc (F=1)
d0 KPd
d0 d
d
KAEd0 KAE(h+d0) KP(h+d0)
R
(KPE/F)d0 KAE(h+d0)
KAEd KP(h+d)
Fig. 4. Distribution of seismic earth pressures in the method by Blum [2]. Fig. 5. Distribution of seismic earth pressures in the proposed method.
146 R. Conti, G.M.B. Viggiani / Soil Dynamics and Earthquake Engineering 50 (2013) 143150
1 2 2 static conditions (kh/kc 0), the ratio d=d depends strongly on h/d
M LHS K P d d d
2 3 " # (and on the mechanical properties of both the soil and the soil
1 d0 d K AE d0 2K P d wall interface), while in critical conditions (kh/kc 1) the passive
K P d K AE d0 d0 d dd0 resistance of the soil below the excavation is mobilised completely
2 3 K AE d0 K P d
down to the pivot point, i.e. d d0 , with d0 about 0.9 d inde-
1
K AE dd0 2 d 2d0 4 pendently on the ratio h/d (and on the mechanical properties of
6 both soil and soilwall interface). No further increase in the
By equating FRHS FLHS and MRHS MLHS one obtains a system of pseudostatic coefcient can take place once critical condition in
two equations in the two unknowns d and d0. the system is attained, as no more passive resistance is available in
Fig. 6(a) shows the ratios d=d and d0 =d as a function of kh/kc, for the soil in front of the wall. In the Blum method (dotted lines) the
three different values of the ratio h/d ( 351, 201, 20 kN/m3, depth of the pivot point is a constant fraction of the embedment
h4 m), while in Fig. 6(b) the same quantities are plotted in a depth and does not evolve with the pseudostatic coefcient
dimensional form. Both d and d0 evolve with increasing kh: in applied. As shown in Fig. 6(b), this assumption leads to values of
the critical acceleration which are substantially lower than those
1.0 predicted by the proposed method.
Fig. 7 shows (a) the horizontal stress and (b) the bending
d0/d (Blummethod) moment distributions computed with the proposed limit equili-
0.8
d0/d d/d brium method for different values of kh (351, 201, 20 kN/
0.6 h/d = 1.3 m3, h 4 m, d 4 m). When kh increases, the dynamic active earth
d/d, d0/d
h/d = 1.0 pressures increase in the retained soil and equilibrium of moments
h/d = 0.8
0.4 requires a larger fraction of the passive earth pressure to be
mobilised in front of the wall. Only in critical conditions (kh kc)
0.2 the passive earth pressure in front of the wall is fully mobilised
down to the pivot point (d d0). Redistribution of earth pressures
0.0 due to inertia forces into the soil results in increasing bending
0.0 0.2 0.4 0.6 0.8 1.0
moments into the wall.
kh/kc For comparison, Fig. 8 shows (a) the horizontal stresses and
(b) the bending moment distributions computed with the Blum
5.0 method for the same case reported in Fig. 7. This time the critical
acceleration of the wall is ac 0.28 g and, accordingly, the max-
h/d = 0.8
4.0 imum internal forces in the wall are much lower than those
h/d = 1.0 predicted by the proposed method. As already discussed, an
3.0 increasing, but constant, fraction of the soil dynamic passive
d, d0 [m]
h[kPa] M [Knm/m]
-500 -400 -300 -200 -100 0 100 200 0 100 200 300 400
0
kh=0.00 (static)
kh=0.10
kh=0.20
-2
kh=0.30
kh=0.40
kh=0.48 (critical)
z [m]
-4
-6
-8
Fig. 7. Proposed method. Distribution of (a) horizontal stresses and (b) bending moment for different values of kh ( 351, 201, 20 kN/m3, h 4 m).
R. Conti, G.M.B. Viggiani / Soil Dynamics and Earthquake Engineering 50 (2013) 143150 147
h[kPa] M [Knm/m]
-500 -400 -300 -200 -100 0 100 200 0 100 200 300 400
0
-4
-6
-8
Fig. 8. Blum method. Distribution of (a) horizontal stresses and (b) bending moment for different values of kh ( 351, 201, 20 kN/m3, h 4 m).
= 1/3
1.0
=35 =30 =25
0.8
Mmax/h3
0.6
h/d = 0.8 h/d = 0.7
0.4 h/d = 0.8 h/d= 0.7
h/d = 1.0 h/d= 0.8
h/d = 1.0
0.2 h/d = 1.3 h/d= 1.0
h/d = 1.3
0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
kh [g] kh [g] kh [g]
Fig. 9. Normalised maximum bending moment, Mmax/h3 as a function of kh, for different values of and h/d ( 1/3 and 2/3 , 20 kN/m3).
angle. These observations are more evident by inspection of In the pseudostatic calculations, kh is assumed to be equal to the
Fig. 10, where the critical acceleration, ac, and the corresponding average value of the accelerations behind the wall, along the rst
normalised maximum bending moment, Mc/h3, are plotted as a 2 m below the soil surface, if kh okc, while kh kc otherwise. The
function of h/d, for different values of ( 1/32/3 , 20 kN/ earth pressure distributions assumed in the Blum method are
m3). The stronger the soilwall system, that is the larger the soil completely different from the numerical ones, and the method
friction angle and the embedded depth, the larger both ac and Mc. provides reliable values for the maximum bending moment in the
In other words, a stronger soilwall system will experience smaller wall only for moderate earthquakes (Fig. 11(bd)). On the contrary,
displacements during the earthquake, but should be designed to both horizontal stresses and bending moments computed with the
sustain larger internal forces. proposed method are in good agreement with the numerical data,
even if some discrepancies can be still observed (see e.g. Fig. 11
3.3. Comparison between pseudo-static calculations, numerical data (d)), mainly due to small non-uniformities in the distribution of
and experimental results the accelerations into the retained soil, not taken into account in
the limit equilibrium calculation.
Fig. 11 shows a comparison between numerical results and The prediction capabilities of the proposed method are
limit equilibrium predictions of the earth pressure and the bend- assessed also with reference to the experimental data presented
ing moment distributions on a cantilevered wall subjected to a real by Madabhushi and Zeng [12,13], obtained from a centrifuge
earthquake scaled at four different values of the maximum dynamic test carried out on a cantilevered wall model embedded
acceleration (analyses no. 14 to 17 in Conti et al. [6], see Table 1). in a uniform layer of dry sand. All the data presented in the
The critical acceleration of the wall is equal to ac 0.28 g, accord- following are at prototype scale.
ing to the Blum method, and ac 0.48 g according to the proposed The excavation and the embedded depths of the wall were equal
method. Numerical data refer to the time instant when the to h7.2 m and d7.2 m respectively. A standard ne silica sand was
accelerations behind the wall reach their maximum value, amax. used (Leighton Buzzard 52/100), reconstituted at a relative density
148 R. Conti, G.M.B. Viggiani / Soil Dynamics and Earthquake Engineering 50 (2013) 143150
= 1/3 = 2/3
1.0
0.8
0.6
ac [g]
0.4
0.2
0.0
1.2
1.0 = 40
= 35
0.8
= 30
Mc/h3
0.6 = 25
= 20
0.4
0.2
0.0
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5
h/d h/d
Fig. 10. Dependence of (a) and (b) critical acceleration, ac, and (c) and (d) normalised maximum bending moment, Mc/h3, on h/d ( 1/3 and 2/3 , 20 kN/m3).
-2
z [m]
-4
-6
-8
0
numerical (static)
proposed (kh=0)
Blum (kh=0)
-2
z [m]
-4
-6
-8
0 100 200 300 400 0 100 200 300 400 0 100 200 300 400 0 100 200 300 400 0 100 200 300 400
bending moment distribution, M [kNm/m]
Fig. 11. Comparison between numerical results and limit equilibrium predictions for a real earthquake scaled at four different values of the maximum acceleration ( 351,
201, 20 kN/m3, h 4 m, d 4 m).
0
0.5
Mmax/h3
0.3
EQ3 Blum method h/d=1.0
h/d= 0.8
-4 proposed 0.2
h/d=1.0
Blum h/d=1.3
0.1 h/d=1.3
-6
0
0.0 0.2 0.4 0.6 0.8 1.0
amax [g], kh
z [m]
-10
seismic actions. The method has been applied in a parametric
study of the effects of the geometry of the wall, considering
different excavated depths, h, and different ratios of the excavated
-12 and embedded depth, h/d, and of the strength of the soil, , and of
EQ3
the contact of the soil and the wall, . The results show that for
increasing strengths of the soilwall system, that is for increasing
EQ1
-14 values of the soil and soilwall friction angles and for increasing
static embedded depths, both the critical acceleration, ac, and the
maximum bending moment on the wall, Mc, increase. In other
words, a stronger soilwall system will experience smaller dis-
-16
0 200 400 600 800 1000 placements during the earthquake, but this is paid for by increas-
M [kNm/m] ing internal forces in the wall.
Fig. 12. Bending moment distribution on the wall: comparison between experi-
The proposed limit equilibrium method has been validated
mental data from centrifuge dynamic tests, numerical results and limit equilibrium against both the results of an extensive numerical study of the
predictions (experimental and numerical data from [12]). behaviour of cantilevered walls under seismic actions [6] and the
experimental data of a centrifuge dynamic test on cantilevered
Table 2 wall embedded in dry sand [13]. In both cases the pseudostatic
Parameters adopted for the pseudostatic analysis of the centrifuge dynamic test. predictions have shown good agreement with the numerical and
experimental ndings, both in terms of magnitude and trend of
Event [1] kh (proposed m.) kh (Blum m.) the internal forces in the wall.
static 40 0.000 0.000
EQ1 34 0.120 0.120
EQ3 34 0.230 0.192 Acknowledgements
[11] Lancellotta R. Lower-bound approach for seismic passive earth resistance. [19] Richards R, Elms DG. Seismic behavior of gravity retaining walls. Journal of
Gotechnique 2007;57(3):31921. Geotechnical Engineering Division, 105; 44964 ASCE.
[12] Madabhushi SPG, Zeng X. Seismic response of exible cantilever retaining [20] Richards R, Elms DG. Seismic passive resistance of tied-back walls. Journal of
walls with dry backll. Geomechanics and Geoengineering 2006;1(4):27589. Geotechnical Engineering 1992;118(7):9961011.
[13] Madabhushi SPG, Zeng X. Simulating seismic response of cantilever retaining [21] Steedman RS. Seismic design of retaining walls. Proceedings of the Institution
walls. Journal of Geotechnical and Geoenvironmental Engineering 2007: of Civil Engineers. Geotechnical Engineering 1998;131:1222.
53949. [22] Towhata I, Islam MS. Prediction of lateral displacement of anchored bulkheads
[14] Masing G. Eigenspannungen und Verfertigung bim Messing. In: Proceedings of induced by seismic liquefaction. Soils and Foundations 1987;27(4):13747.
the 2nd international congress on applied mechanics, Zurich; 1926. [23] Vucetic M, Dobry R. Effect of soil plasticity on cyclic response. Journal of
[15] Mononobe N, Matsuo H. On the determination of earth pressure during Geotechnical and Geoenvironmental Engineering 1991;117(1):89107.
earthquake. Proceedings of the 2nd World Engineering Conference 1929;9: [24] Whitman RV. Seismic design and behaviour of gravity retaining walls. Design
17785. and Performance of Earth Retaining Structures, 25. ASCE, Geotech. Special
[16] Neelakantan G, Budhu M, Richards R. Balanced seismic design of anchored Publication; 81742.
retaining walls. Journal of Geotechnical Engineering 1992;118(6):87388. [25] Zeng X.. Modelling the behaviour of quay walls in earthquakes. PhD thesis.
[17] Newmark NM. Effects of earthquakes on dams and embankments. Gotechni- Cambridge, UK: University of Cambridge; 1990.
que 1965;15(2):13960. [26] Zienkiewicz OC, Bianic N, Shen FQ. Earthquake input denition and the
[18] Okabe S. General theory of earth pressure and seismic stability of retaining transmitting boundary condition. In: St. Doltnis I, editor. Conf: Advances in
wall and dam. Journal of Japanese Society of Civil Engineering 1924;12:1. computational non-linear mechanics; 1988. p. 10938.