Sei sulla pagina 1di 14

See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/222700017

An alternative to the MononobeOkabe


equations for seismic earth pressures
ARTICLE in SOIL DYNAMICS AND EARTHQUAKE ENGINEERING OCTOBER 2007
Impact Factor: 1.22 DOI: 10.1016/j.soildyn.2007.01.004

CITATIONS

READS

34

1,103

3 AUTHORS, INCLUDING:
George Mylonakis

Panos Kloukinas

University of Bristol

University of Bristol

154 PUBLICATIONS 780 CITATIONS

26 PUBLICATIONS 44 CITATIONS

SEE PROFILE

SEE PROFILE

Available from: George Mylonakis


Retrieved on: 02 October 2015

ARTICLE IN PRESS

Soil Dynamics and Earthquake Engineering 27 (2007) 957969


www.elsevier.com/locate/soildyn

An alternative to the MononobeOkabe equations for


seismic earth pressures
George Mylonakis, Panos Kloukinas, Costas Papantonopoulos
Department of Civil Engineering, University of Patras, Rio 26500, Greece
Received 23 July 2006; received in revised form 23 January 2007; accepted 25 January 2007

Abstract
A closed-form stress plasticity solution is presented for gravitational and earthquake-induced earth pressures on retaining walls. The
proposed solution is essentially an approximate yield-line approach, based on the theory of discontinuous stress elds, and takes into
account the following parameters: (1) weight and friction angle of the soil material, (2) wall inclination, (3) backll inclination, (4) wall
roughness, (5) surcharge at soil surface, and (6) horizontal and vertical seismic acceleration. Both active and passive conditions are
considered by means of different inclinations of the stress characteristics in the backll. Results are presented in the form of
dimensionless graphs and charts that elucidate the salient features of the problem. Comparisons with established numerical solutions,
such as those of Chen and Sokolovskii, show satisfactory agreement (maximum error for active pressures about 10%). It is shown that
the solution does not perfectly satisfy equilibrium at certain points in the medium, and hence cannot be classied in the context of limit
analysis theorems. Nevertheless, extensive comparisons with rigorous numerical results indicate that the solution consistently
overestimates active pressures and under-predicts the passive. Accordingly, it can be viewed as an approximate lower-bound solution,
than a mere predictor of soil thrust. Compared to the Coulomb and MononobeOkabe equations, the proposed solution is simpler, more
accurate (especially for passive pressures) and safe, as it overestimates active pressures and underestimates the passive. Contrary to the
aforementioned solutions, the proposed solution is symmetric, as it can be expressed by a single equationdescribing both active and
passive pressuresusing appropriate signs for friction angle and wall roughness.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Retaining wall; Seismic earth pressure; Limit analysis; Lower bound; Stress plasticity; MononobeOkabe; Numerical analysis

1. Introduction
The classical equations of Coulomb [14,10] and
MononobeOkabe [511] are being widely used for
determining earth pressures due to gravitational and
earthquake loads, respectively. The MononobeOkabe
solution treats earthquake loads as pseudo-dynamic,
generated by uniform acceleration in the backll. The
retained soil is considered a perfectly plastic material,
which fails along a planar surface, thereby exerting a limit
thrust on the wall. The theoretical limitations of such
an approach are well known and need not be repeated
herein [1113,1618]. Given their practical nature and
reasonable predictions of actual dynamic pressures (e.g.
Corresponding author. Tel.: +30 2610 996542; fax: +30 2610 996576.

E-mail address: mylo@upatras.gr (G. Mylonakis).


0267-7261/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.soildyn.2007.01.004

Refs. [9,14,1618]), solutions of this type are expected to


continue being used by engineers for a long time to come.
This expectation does not seem to diminish by the advent
of displacement-based design approaches, as the limit
thrusts provided by the classical methods can be used to
predict the threshold (yield) acceleration beyond which
permanent dynamic displacements start to accumulate
[11,15,1921,43].
Owing to the translational and statically determined
failure mechanisms employed, the limit-equilibrium MononobeOkabe solutions can be interpreted as kinematic
solutions of limit analysis [22]. The latter solutions are
based on kinematically admissible failure mechanisms in
conjunction with a yield criterion and a ow rule for the
soil material, both of which are enforced along prespecied failure surfaces [10,19,23,24,40,42]. Stresses outside the failure surfaces are not examined and, thereby,

ARTICLE IN PRESS
958

G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

equilibrium in the medium is generally not satised. In the


realm of associative and convex materials, solutions of this
type are inherently unsafe that is, they underestimate active
pressures and overestimate the passive [10,24,25,40].
A second group of limit-analysis methods, the stress
solutions, make use of pertinent stress elds that satisfy the
equilibrium equations and the stress boundary conditions,
without violating the failure criterion anywhere in the
medium [2527]. On the other hand, the kinematics of the
problem is not examined and, therefore, compatibility
of deformations is generally not satised. For convex
materials, formulations of this type are inherently safe that
is, they overestimate active pressures and underestimate the
passive [10,25,26]. The best known such solution is that of
Rankine, the applicability of which is severely restricted by
the assumptions of horizontal backll, vertical wall and
smooth soilwall interface. In addition, the solution may
be applied only if the surface surcharge is uniform or nonexisting. Owing to difculties in deriving pertinent stress
elds for general geometries, the vast majority of limitanalysis solutions in geotechnical design are of the
kinematic type [811,26]. To the best of the authors
knowledge, no simple closed-form solution of the stress
type has been derived for seismic earth pressures.
Notwithstanding the theoretical signicance and practical appeal of the Coulomb and MononobeOkabe
solutions, these formulations can be criticized on the
following important aspects: (1) in the context of limit
analysis, their predictions are unsafe; (2) their accuracy
(and safety) diminishes in the case of passive pressures
on rough walls, (3) the mathematical expressions are
complicated and difcult to verify,1 (4) the distribution
of tractions on the wall are not predicted (typically
assumed linear with depth following Rankines solution),
(5) optimization of the failure mechanism is required in the
presence of multiple loads, to determine a stationary
(optimum) value of soil thrust, and (6) in the context of
limit-equilibrium analysis, stress boundary conditions are
not satised, as the yield surface does not generally emerge
at the soil surface at the required angles of 45  f=2.
In light of the above arguments, it appears that the
development of a closed-form solution of the stress type for
assessing seismically-induced earth pressures would be
desirable. It will be shown that the proposed solution,
although approximate, is mathematically simpler than the
existing kinematic solutions, offers satisfactory accuracy
(maximum deviation for active pressures against rigorous
numerical solutions less than 10%), yields results on the
safe side, satises stress boundary conditions, and predicts
the point of application of soil thrust. Last but least, the
solution will be shown to be symmetric with respect to
active and passive conditions, as it can be expressed by a

single equation with opposite signs for friction angle and


wall roughness. Apart from its intrinsic theoretical interest,
the proposed analysis can be used for the assessment and
improvement of other related methods.
2. Problem denition and model development
The problem under investigation is depicted in Fig. 1: a
slope of dry cohesionless soil retained by an inclined
gravity wall, is subjected to plane strain conditions under
the combined action of gravity (g and seismic body
forces ah  g and av  g in the horizontal and vertical
direction, respectively. The problem parameters are: height
(H) and inclination o of the wall, inclination (b) of the
backll; roughness (d) of the wallsoil interface; friction
angle (f) and unit weight (g) of the soil material, and
surface surcharge (q). Since backlls typically consist of
granular materials, cohesion in the soil and cohesion at the
soilwall interface are not studied here. In addition, since
the vibrational characteristics of the soil are neglected, the
seismic force is assumed to be uniform in the backll. Also,
the wall can translate away from, or towards to, the
backll, under zero rotation. Both assumptions have
important implications in the distribution of earth pressures on the wall, as explained below.
The resultant body force in the soil is acting under an
angle ce from vertical
ah
,
(1)
tan ce
1  av
which is independent of the unit weight of the material.
Positive ah (i.e., ce 40) denotes inertial action towards the
wall (ground acceleration towards the backll), which
maximizes active thrust. Conversely, negative ah (i.e.,
ce o0) denotes inertial action towards the backll, which
minimizes passive resistance. In accordance with the rest of
the literature, positive av is upward (downward ground
acceleration). However, its inuence on earth pressures,
although included in the analysis, is not studied numerically here, as it is usually minor and often neglected in
design [9,21].

inclined
backfill

+ e
cohesionless soil
(, )
H

+
z

The story of a typographical error in the MononobeOkabe formula


that appeared in a seminal article of the early 1970s and subsequently
propagated in a large portion of the literature, is indicative of the difculty
in checking the mathematics of these expressions (Davies et al. [41]).

inclined wall,
roughness ()

Fig. 1. The problem under consideration.

+ah
+av

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

In the absence of surcharge, the MononobeOkabe


solution to the above problem is given by the well-known
formula [11]:
KE

2PE

gH 2

To prevent slope failure when inertial action is pointing


towards the wall, the seismic angle ce should not exceed
the difference between the friction angle and the slope

cos2 f  ce o
s#2 ,
"
sind f sinf  ce b
cos ce cos2 o cosd  ce o 1 
cosd  ce o cosb  o

where PE denotes the limit of seismic thrust on the wall


units F=L and K E is the corresponding earth pressure
coefcient. In the above representation (and hereafter),
the upper sign refers to active conditions (PE PAE ;
K E K AE ), and the lower sign to passive (PE PPE ;
K K PE ).
A drawback of the above equation lies in the difculty in
interpreting the physical meaningespecially signsof the
various terms (Ref. [25, footnote in p. 4]). As will be shown
below, the proposed solution is free of this problem.

unit length

soil
surface

ZONE A

(w, w)
passive

(w, w)
active

wall length
L = H / cos

ZONE B

Fig. 2. Stress elds close to soil surface (zone A) and the wall (zone B).

959

(2)

inclination. Therefore, the following constraint applies [9]:


ce of  b.

(3)

A similar relation can be written for the case where inertial


action is pointing towards the backll, but it is of limited
practical interest and will not be discussed here.
To analyze the problem, the backll is divided into two
main regions subjected to different stress elds, as shown in
Fig. 2: the rst region (zone A) is located close to the soil
surface, whereas the second (zone B) close to the wall. In
both regions the soil is assumed to be in a condition of
impeding yielding under the combined action of gravity
and earthquake body forces. The same assumption is
adopted for the soilwall interface, which, however, is
subjected exclusively to contact stresses. A transition zone
between regions A and B is introduced later on.
Fundamental to the proposed analysis is the assumption
that stresses close to the soil surface can be well
approximated by those in an infinite slope, as shown in
Fig. 2. In this region (A), the inclined soil element shown is
subjected to canceling actions along its vertical sides. Thus
equilibrium is achieved solely under body forces and
contact stress acting at its bottom face.
Based on this physically motivated hypothesis, the
stresses sb and tb at the base of the inclined element are
determined from the following expressions [34]:


q
sb gz
(4a)
cos2 b,
cos b


q
tb gz
sin b cos b,
(4b)
cos b
which are valid for static conditions (ah av 0) and
satisfy the stress boundary conditions at the surface.
Eqs. (4) suggest that the ratio of shear to normal stresses
is constant tan b at all depths, and that points at the same
depth are subjected to equal stresses. Note that due to
static determinacy and anti-symmetry, the above relations
are independent of material properties and asymptotically
exact at large distances from the wall.
Considering the material to be in a condition of
impeding yielding, the Mohr circle of stresses in region
A is depicted in Fig. 3. The different locations of the stress
point (sb ; tb ) for active and passive conditions and the
different inclinations of the major principal plane (indicated by heavy lines) are apparent in the graph.
From the geometry of Fig. 3, the normal stress sb is
related to mean stress S A through the proportionality

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

960

solutions based on equilibrium totally ignore the displacement eld [29].


From the geometry of Fig. 3, normal traction sw is
related to mean stress SB through the expression

ZONE A
(,)
passive
case

(,)
active
case
soil

1
1

sw S B 1  sin f cosD2  d,


e
surfac

1A

1+

active

passive

ZONE B
(w, w)
passive
wall
plane
2

passive

2+

1B

SB

active

(w,w)
active
wall
plane
Fig. 3. Mohr circles of effective stresses and inclination of the major
principal planes in zones A and B.

relation
sb S A 1  sin f cosD1  b,

(5)

where D1 denotes the Caquot angle [23,28] given by


sin D1

sin b
.
sin f

(6)

For points in region B, it is assumed that stresses are


functions exclusively of the vertical coordinate and obey
the strength criterion of the frictional soilwall interface, as
shown in Fig. 2. Accordingly, at orientations inclined at an
angle o from vertical,
tw sw tan d,

where D2 is the corresponding Caquot angle given by


sin D2

SA

(7)

where sw and tw are the normal and shear tractions


on the wall, at depth z. The above equation is asymptotically exact for points in the vicinity of the wall. The
corresponding Mohr circle of stresses is depicted in Fig. 3.
The different signs of shear tractions for active and passive
conditions follow the directions shown in Fig. 2 (passive
wall tractions pointing upward, active tractions pointing
downward), which comply with the kinematics of the
problem. This is in contrast with the widespread view that

(8)

sin d
.
sin f

(9)

In light of the foregoing, it becomes evident that the


orientation of stress characteristics in the two regions is
different and varies for active and passive conditions. In
addition, the mean stresses SA and S B generally do not
coincide, which suggests that a Rankine-type solution
based on a single stress eld is not possible.
To determine the separation of mean stresses S A and S B
and ensure a smooth transition in the orientation of
principal planes in the two zones, a logarithmic stress fan2
is adopted in this study, centered at the top of the wall. In
the interior of the fan, principal stresses are gradually
rotated by the angle y separating the major principal planes
in the two regions, as shown in Fig. 4. This additional
condition is written as [10]
SB S A exp2y tan f.

(10)

The negative sign in the above equation pertains to the case


where S B oS A (e.g., active case) and vice versa. The above
equation is an exact solution of the governing Kotter
equations for a weightless frictional material. For a
material with weight, the solution is only approximate as
Kotters equations are not perfectly satised [2527]. In
other words, the log spiral fan accurately transmits stresses
applied at its boundaries, but handles only approximately
body forces imposed within its volume. The error is
expected to be small for active conditions (which are of
key importance in design), because of the small opening
angle of the fan, and bigger for passive conditions. As a
result, the above solution cannot be interpreted in the
context of limit analysis theorems. Nevertheless, it will be
shown that these violations are of minor importance from a
practical viewpoint.
2.1. Solution without earthquake loading
The total thrust on the wall due to surcharge and gravity
loading is obtained by the well-known expression [10]
P K q qH 12 K g gH 2 ,

(11)

which is reminiscent (though not equivalent) of the bearing


capacity equation of a strip surface footing on cohesionless
soil. In the above equation, K q and K g denote the earth
pressure coefcients due to surcharge and self-weight,
respectively.
2
This should not be confused with log-spiral shaped failure surfaces
used in kinematic solutions of related problems.

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

which coincides with the kinematic solution of Chen and


Liu [31], established using a Coulomb mechanism. Note
that for a horizontal backll (b 0), coefcients K q and K g
coincide regardless of wall inclination and material properties. Eq. (14) represents an exact solution for a weightless
material with surcharge. A simplied version of the above
solutions, restricted to the special case of a vertical wall
with horizontal backll and no surcharge (o b 0;
q 0), has been derived by Lancelotta [30]. Another
simplied solution, which, however, contains some algebraic mistakes (see application example in the Appendix)
and is restricted to active conditions and no surcharge, has
been presented by Powrie [35].

1 +
2

AB

961

zone A

zone B
2
2
2

2.2. Solution including earthquake loading


ACTIVE CONDITIONS

Recognizing that earthquake action imposes a resultant


thrust in the backll inclined by a constant angle ce from
vertical (Fig. 1), it becomes apparent that the pseudodynamic problem does not differ fundamentally from the
corresponding static problem, as the former can be obtained
from the latter through a rotation of the reference axes by
the seismic angle ce , as shown in Fig. 5. In other words,
considering ce does not add an extra physical parameter to
the problem, but simply alters the values of the other
variables. This property of similarity was apparently rst
employed by Briske [32] and later by Arango [8,9] in the
analysis of related problems. Application of the concept to
the present analysis yields the following algebraic transformations, according to the notation of Fig. 5:

1
2
2
2 +
2

AB
PASSIVE CONDITIONS

z
zone A

zone B

b b ce ,

(15)

o o ce ,

(16)

Fig. 4. Rotation of major principal planes between zones A and B for


active and passive conditions.

Combining Eqs. (5), (8) and (10), and integrating over


the height of the wall, it is straightforward to show that the
earth pressure coefcient K g is given by [39]


coso  b cos b 1  sin f cosD2  d
Kg
cos d cos2 o
1  sin f cosD1  b
 exp2y tan f,

12
H*
H

where
2y D2  D1 d b  2o

cos o
,
coso  b

e
*

(13)

is twice the angle separating the major principal planes in


zones A and B (Fig. 4). The convention regarding double
signs in the above equations is as before.
It is also straightforward to show that the surcharge
coefcient K q is related to K g through the simple expression
Kq Kg

(14)

Fig. 5. Similarity transformation for analyzing the pseudo-dynamic


seismic problem as a gravitational problem. Note the modied wall
height H  , backll slope b , and wall inclination o in the
transformed geometry. Also note that the rotation should be performed
in the opposite sense (i.e., clockwise) for passive pressures ce o0.

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

962

H  H coso ce = cos o,

(17)

g g1  av = cos ce ,

(18)

q q1  av = cos ce .

(19)

which encompasses seismic action and can be used in the


context of Eq. (11). In the above equation,
2yE D2  D1 d b  2o  ce

(22)

is twice the revolution angle of principal stresses in


the two regions under seismic conditions; D1 equals
Arcsinsinb ce = sin f, following Eqs. (6) and (15).
The seismic earth pressure coefcient K Eq is obtained as

The modication in g and q is due to the change in length of


the corresponding vectors (Fig. 1) as a result of inertial
action. To obtain Eq. (19), it has been tacitly assumed that
the surcharge responds to the earthquake motion in the
same manner as the backll and, thereby, the transformed
surcharge remains vertical. Note that this is not an essential
hypothesisjust a convenient (reasonable) assumption from
an analysis viewpoint. Understandably, the strength parameters f and d are invariant to the transformation.
In the light of the above developments, the soil thrust
including earthquake action can be determined from the
modied expression:

which coincides with the static solution in Eq. (14).


The horizontal component of soil thrust is determined
from the actual geometry, as in the gravitational
problem

PE K q q H  12 K g g H 2 ,

PEH PE coso  d.

(20a)

in which parameters b, o, H, g, and q have been replaced


by their transformed counterparts. The symbols K q and K g
denote the surcharge and self-weight coefcients in the
modied geometry, respectively.
Substituting Eqs. (15) through (19) in Eq. (20a) yields the
modied earth pressure expressions
PE 1  av K Eq qH 1=2 K Eg gH 2 ,

(20b)

where
K Eg

coso  b cosb ce
cos ce cos d cos2 o


1  sin f cosD2  d

exp2yE tan f,
1  sin f cosD1  b ce 
21

K Eq K Eg

cos o
,
coso  b

(23)

(24)

2.3. Seismic component of soil thrust


Following Seed and Whitman [8], the seismic component
of soil thrust is dened from the difference:
DPE PE  P,

(25)

which is mathematically valid, as the associated vectors PE


and P are coaxial. Nevertheless, the physical meaning of
DPE is limited given that the stress elds (and the
corresponding failure mechanisms) in the gravitational
and seismic problems are different. In addition, DPE
cannot be interpreted in the context of limit analysis
theorems, as the difference of PE and P is neither an upper
nor a lower bound to the true value.

Table 1
Comparison of results for active and passive earth pressures predicted by various methods
o

0

20

0

10

0

15

0

(a) K Ag valuesa
Coulomb
Kinematic limit analysis [31]
Zero extension [33]
Slip line [28]
Proposed stress limit analysis

0.490
0.490
0.49
0.490
0.490

0.447
0.448
0.41
0.450
0.451

0.333
0.333
0.33
0.330
0.333

0.301
0.303
0.27
0.300
0.305

0.217
0.217
0.22
0.220
0.217

(b) K Pg valuesb
Coulomb
Kinematic limit analysis [31]
Zero extension [33]
Slip line [28]
Proposed stress limit analysis

2.04
2.04
2.04
2.04
2.04

2.64
2.58
2.55
2.55
2.52

3.00
3.00
3.00
3.00
3.00

4.98
4.70
4.65
4.62
4.44

4.60
4.60
4.60
4.60
4.60

30

40
20

0.199
0.200
0.17
0.200
0.201
11.77
10.07
9.95
9.69
8.92

20

20

30

30

0

15

0

15

0.498
0.498

0.521
0.531

0.476
0.476

0.487
0.485

0.212
0.218

0.229
0.237

0.180
0.189

0.206
0.217

2.27
2.27

2.16
2.13

3.162
3.160

3.16
3.157

5.34
5.09

5.06
4.78

12.91
8.92

8.45
7.07

The results for d o 0 are identical for all methods. Note the decrease in K Pg values as we move from top to bottom in each column, and the
corresponding increase in K Ag ; b 0 (modied from Chen and Liu [31]).
a
K Ag PA =12 gH 2 .
b
K Pg PP =12 gH 2 .

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

3. Model verication and results

0.6
1
KA=PA/ ( H2)
2

Coefficient of Active Earth Pressure, KA

0.5

 = 45,  = 2/ 3

PA
0.4
= 20
0.3

Chen & Liu(1990)


Caquot & Kerisel (1948)
Proposed Stress Limit Analysis

0.2

= 0

0.1

= 20

0.0
0

10

15

20

25

Slope Angle of Backfill, 


Fig. 6. Comparison of results for active earth pressures predicted by
different methods (modied from Chen [10]).

1
KP=PP /(  H2 )
2

 = 30,  = 20

PP

4
Coefficient of Passive Earth Pressure, KP

Presented in Table 1 are numerical results for gravitational active and passive pressures K Ag ; K Pg from the
present solution and established solutions from the
literature. The predictions are in good agreement (largest
discrepancy about 10%), with the exception of Coulombs
method which signicantly overestimates passive pressures.
Moving from the top to the bottom of each column, an
increase in K Ag values and a decrease in K Pg values can be
observed. This is easily understood given the nonconservative nature of the rst two solutions (Coulomb,
Chen), and the conservative nature of the last two
(Sokolovskii [28], proposed). This observation does not
hold for the zero extension line solution of Habibagahi
and Ghahramani [33], which cannot be classied in the
context of limit analysis theorems.
Results for gravitational active pressures on a rough
inclined wall obtained according to three different methods
as a function of the slope angle b, are shown in Fig. 6. The
performance of the proposed solution is good (maximum
deviation from Chens solution about 10%despite the
high friction angle of 45 ) and elucidates the accuracy of
the predictions. The performance of the simplied solution
of Caquot and Kerisel [23] versus that of Chen and Liu [31]
is as expected.
Corresponding predictions for passive pressures are
given in Fig. 7, for a wall with negative backll slope

963

3
 = 0

 = 10

 = 20
1
Lee & Herington (1972)
Chen & Liu (1990)
Sokolovskii (1965)
Proposed Stress Limit Analysis
0
0

10
20
Angle of Wall Friction, 

30

Fig. 7. Comparison of results for passive earth pressures by predicted by


different methods (modied from Chen and Liu [31]).

inclination, as a function of the wall roughness d. The


agreement of the various solutions, given the sensitivity of
passive pressure analyses, is very satisfactory. Of particular
interest are the predictions of Sokolovskiis [28] and
Lee and Heringtons [36] methods, which, surprisingly,
exceed those of Chen for rough walls. This trend is
particularly pronounced for horizontal backll and values
of d above approximately 10 and has been discussed by
Chen and Liu [31].
Results for active seismic earth pressures are given in
Fig. 8, referring to cases examined in the seminal study of
Seed and Whitman [8], for a reference friction angle of 35 .
Naturally, active pressures increase with increasing levels
of seismic acceleration and slope inclination and decrease
with increasing friction angle and wall roughness. The
conservative nature of the proposed analysis versus the
MononobeOkabe (MO) solution is evident in the graphs.
The trend is more pronounced for high levels of horizontal
seismic coefcient ah 40:25, smooth walls, level backlls,
and high friction angles. Conversely, the trend becomes
weaker with steep backlls, rough walls, and low friction
angles.
A similar set of results is shown in Fig. 9, for a reference
friction angle of 40 . The following interesting observations
can be made: First: the predictions of the proposed analysis
are in good agreement with the results from the kinematic
analysis of Chen and Liu [31], over a wide range of material

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

0.7
0.6

0.7

 = = 0
 = 35

0.6

0.5

 = = 0
 = 35

0.5

= 0

= 0
=/ 2

0.4
0.3

1
KAE=PAE/ ( H2)
2

0.2

ah

PAE

KE cos

Coefficient of Seismic Active Earth Pressure, KE

964

0.4
=/ 2

0.3

KAE=PAE/(

0.2

ah

PAE

0.1

0.1
M - O Analysis
Proposed Stress Limit Analysis

0.0
0.0

M - O Analysis
Proposed Stress Limit Analysis

0.0
0.0

0.1
0.2
0.3
0.4
0.5
Horizontal Seismic Coefficient, ah

0.7
0.6

1
H2)
2

0.7

=  = 0
=/ 2

= 30

0.6

0.1
0.2
0.3
0.4
0.5
Horizontal Seismic Coefficient, ah

=  = 0
= 35 ; = / 2
 = 20

35
40

0.5

0.4
0.3
KAE=PAE/ (1 H2)
2

0.2

H
0.1

PAE

ah

KE cos

KE cos

0.5

 = 0

0.4
0.3
0.2
0.1

M- O Analysis
Proposed Stress Limit Analysis

0.0
0.0

0.1
0.2
0.3
0.4
0.5
Horizontal Seismic Coefficient, ah

KAE=PAE/( 1 H2)
2

ah

H
PAE

M- O Analysis
Proposed Stress Limit Analysis

0.0
0.0

0.1
0.2
0.3
0.4
0.5
Horizontal Seismic Coefficient, ah

Fig. 8. Comparison of active seismic earth pressures predicted by the proposed solution and from conventional MO analysis, for different geometries,
material properties and acceleration levels; av 0 (modied from Seed and Whitman [8]).

and geometric parameters. Second, the present analysis is


conservative in all cases. Third, close to the slope stability
limit (Fig. 9d), or for high accelerations and large wall
inclinations (Fig. 9c), Chens predictions are less accurate
than those of the elementary MO solution. In the same
extreme conditions, the proposed solution becomes exceedingly conservative, exceeding MO predictions by
about 35%. Note that whereas the MO and the proposed
solution break down in the slope stability limit, Chens
solution allows for spurious mathematical predictions of
active thrust beyond the limit, as evident in Fig. 9d. Fourth,
with the exception of the aforementioned extreme cases,
Chens and MO predictions remain close over the whole
range of parameters examined. The improvement in the
predictions of the former over the latter is marginal.
Results for seismic passive pressures (resistances) are
shown in Fig. 10 for the common case of a rough vertical
wall with horizontal backll. Comparisons of the proposed

solution with results from the MO and Chens kinematic


methods are provided on the left graph (Fig. 10a). The
predictions of the stress solutions are, understandably,
lower than those of Chen and Liu, whereas MO
predictions are very high (i.e., unconservative)especially
for friction angles above 37 . Given the sensitivity of
passive pressure analyses, the performance of the proposed
method is deemed satisfactory.
An interesting comparison is presented in Fig. 10b:
average predictions from the two closed-form solutions
(MO solution and proposed stress solution) are plotted
against the rigorous numerical results of Chen and
Liu [31]. Evidently, in the range of most practical interest
30 ofo40 , the discrepancies in the results have been
drastically reduced. This suggests that the limit equilibrium
(kinematic) MO solution and the proposed static solution
overestimate and underestimate, respectively, passive
resistances by the same amount in the specic range of

ARTICLE IN PRESS

KAE=PAE/ (

0.7

1 2
H )
2

KAE=PAE/ (

0.6

 = = 0 ; = 2/3

0.6

ah

PAE

0.5

0.5

0.30

0.20

0.4

ah

PAE

0.4

=0

ah = 0

/2
= =

0.2

0.2

M - O Analysis
Kinematic Limit Analysis (Chen & Liu 1990)
Proposed Stress Limit Analysis

M - O Analysis
Kinematic Limit Analysis (Chen & Liu 1990)
Proposed Stress Limit Analysis

0.1
0.0

0.1
25

30

35
40
Friction Angle, 

45

Coefficient of Seismic Active Earth Pressure, KAE

= 40;ah = 0.20 ;  = / 2

0.8
ah

PAE
15

0.4

= 0

15

0.2
M - O Analysis
Kinematic Limit Analysis(Chen & Liu 1990)
Proposed Stress Limit Analysis

0.0
-20

0.2

0.3

0.4

1.6
1
KAE= PAE/ ( H2)
2

0.6

0.1

Horizontal Seismic Coefficient, ah

1.0
Coefficient of Seismic Active Earth Pressure, KAE

1 2
H )
2

0.3

0.10

0.3

965

 = = 0 ; = 40

KAE cos

Coefficient of Seismic Active Earth Pressure, KAE

G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

-10

10

20

Slope Angle of Backfill, 

KAE=PAE/ (

1.4

slope
stability
limit

= 40; = 0 ;  = / 2

1.2
1.0

1 2
H )
2

ah

PAE

0.8
0.6

= / 2

/3

0.4
0.2
0.0
0.0

M - O Analysis
Kinematic Limit Analysis (Chen & Liu 1990)
Proposed Stress Limit Analysis

0.1
0.2
0.3
Horizontal Seismic Coefficient, ah

0.4

Fig. 9. Comparison of active seismic earth pressures predicted by different methods, for different geometries, material properties, and acceleration levels;
f 40 , av 0 (modied from Chen and Liu [31]).

properties. Accordingly, this averaging might be warranted


for design applications involving passive pressures.
Results for the earth pressure coefcient due to
surcharge K qE (Eq. (23)) are presented in Fig. 11, for both
active and passive conditions involving seismic action. The
agreement between the stress solution and the numerical
results of Chen and Liu [31] is excellent in the whole range
of parameters examined (except perhaps for active
pressures, where ah 0:3). As expected, MO solution
performs well for active pressures, but severely overestimates the passive.

3.1. Distribution of earth pressures on the wall: analytical


findings
Mention has already been made that in the realm
of pseudo-dynamic analysis, there is no fundamental
physical difference between gravitational and seismic earth
pressures. Eqs. (4) indicate that stresses in the soil vary
linearly with depth (stress fan does not alter this
dependence), which implies that both gravitational and
seismic earth pressures vary linearly along the back of wall.
In the absence of surcharge, the distribution becomes

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

Coefficient of Seismic Passive Earth Pressure, K PE

966

25

25

a
ah

H PPE

20

H PPE

20

ah = 0
-0.1
-0.2
-0.3

1
KPE=PPE/ ( H2)
2

15

ah = 0
-0.1
-0.2
-0.3

= 0, = 0

Mononobe -Okabe
(ah=0)

 = 2 / 3

10

1
KPE=PPE/ ( H2)
2

15

10

ah

Kinematic Limit Analysis (Chen & Liu1990)


Average of M-O & Proposed Stress Limit Analysis

Kinematic Limit Analysis (Chen & Liu1990)


Proposed Stress Limit Analysis

0
25

30

35

40

25

45

30

Angle of Internal Friction, 

35

40

45

Angle of Internal Friction, 

Fig. 10. Comparison of results for passive seismic resistance on a rough wall predicted by various methods (modied from Chen and Liu [31]).

20

q
0.7

aq

= = 0o

K PE = PPE / q H
q

=2/3

0.5

0.3

0.2

0.4

0.1
0.3

15

K AE = PAE / q H

0.6

aq
P

10

= = 0o
=2/3

ah = 0
0.1
0.2
0.3

Mononobe - Okabe
(ah = 0)

ah = 0
5

0.2
Kinematic Limit Analysis (Chen & Liu 1990)
Proposed Stress Limit Analysis

0.1
25

30

35

40

Kinematic Limit Analysis (Chen & Liu 1990)


Proposed Stress Limit Analysis

45

Friction Angle, o

0
25

30

35

40

45

Friction Angle, o

Fig. 11. Variation of K AEq and K PEq values with fangle for different acceleration levels.

proportional with depth, as in the Rankine solution.


Accordingly, the point of application of seismic thrust is
located at a height of H=3 above the base of the wall. It is
well known from experimental observations and rigorous
numerical solutions, that this is not generally true. The
source of the difference lies in the distribution of inertial
forces in the soil mass (which is often sinusoidal like
following the time-varying natural mode shapes of the
deposit), as well as the various kinematic boundary
conditions (wall exibility, foundation compliance, presence of supports). Studying the above factors lies beyond
the scope of this article, and like will be the subject of a
future publication. Some recent developments are provided
in the Master thesis of the second author [39] as well as in
Refs. [11,1618,37,38].

4. Discussion: simplicity and symmetry


It is instructive to show that the proposed solution can
be derived essentially by inspection, without tedious
algebraic manipulations as in the classical equations.
Indeed, basis of Eq. (12) is the familiar Rankine ratio
1  sin f=1  sin f. The terms cosD2  d and cosD1 
b in the numerator and denominator of the expression
reect the fact that stresses sb and sw are not principal.
Both terms involve the same double signs as their multipliers ( sin f and  sin f, respectively). Angle b and
associated angle D1 have to be in the denominator, as an
increase in their value must lead to an increase in active
thrust. The exponential term is easy to remember and
involves the same double sign as the other terms in the

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

numerator. With reference


to the factors outside the
p
brackets, 1= cos d 1 tan2 d stands for the vectorial
sum of shear and normal tractions at the wallsoil
interface. Factor cos b arises from the equilibrium of the
innite slope in Eq. (4a). Finally, coso  b=cos2 o is a
geometric factor arising from the integration of stresses
along the back of the wall, and is associated with the
inclination of the wall and backll.
In light of the above, the solution for gravitational
pressures can be expressed by the single equation
Kg

coso  b cos b
2
cos d cos o

1  sin f cosD2  d

exp2y tan f,
1 sin f cosD1 b

26

which is valid for both active conditions (using positive


values for f and d and passive conditions (using negative
values for f and d. It is straightforward to show that this
property is not valid for the MononobeOkabe solutions in
Eq. (4). The lack of symmetry in the limit equilibrium
solutions can be attributed to the maximization and
minimization operations involved in deriving the limit
thrusts. An application example elucidating the simplicity
of the solution is provided below.

967

average between the predictions of the MO solution


and the proposed stress solution (both available in
closed forms) yields results which are comparable to
those obtained from rigorous numerical solutions.
(5) The pseudo-dynamic seismic problem can be deduced
from the corresponding static problem through a
revolution of the reference axes by the seismic angle
ce (Fig. 5). This similarity suggests that the Coulomb
and MO solutions are essentially equivalent.
(6) Contrary to the overall gravitational-seismic thrust PE ,
the purely seismic component DPE PE  P cannot be
put in the context of a lower or an upper bound. This
holds even when PE and P are rigorous upper or lower
bounds.
(7) In the realm of the proposed model, the distribution of
earth pressures on the back of the wall is linear with
depth for both gravitational and seismic conditions.
This is not coincidental given the similarity between the
gravitational and pseudo-dynamic problem.
It should be emphasized that the verication of the
proposed solution was restricted to analyticalnot experimental results. Detailed comparisons against experimental
results, including distribution of earth pressures along the
wall, will be the subject of a future publication.

5. Conclusions
A stress plasticity solution was presented for determining
gravitational and earthquake-induced earth pressures on
gravity walls retaining cohesionless soil. The proposed
solution incorporates idealized, yet realistic wall geometries
and material properties. The following are the main
conclusions of the study:
(1) The proposed solution is simpler than the classical
Coulomb and MononobeOkabe equations. The main
features of the mathematical expressions, including
signs, can be deduced by physical reasoning, which is
hardly the case with the classical equations. Also, the
proposed solution is symmetric with respect to active
and passive conditions, as it can be expressed by a
single equation with opposite signs for soil friction
angle and wall roughness.
(2) Extensive comparisons with established numerical
solutions indicate that the proposed solution is safe,
as it overestimates active pressures and under-predicts
the passive. This makes the method appealing for use in
practical applications.
(3) For active pressures, the accuracy of the solution is
excellent (maximum observed deviation from numerical
data is about 10%). The largest deviations occur for
high seismic accelerations, high friction angles, steep
backlls, and negative wall inclinations.
(4) For passive resistances, the predictions are also
satisfactory. However, the error is largerespecially
at high friction angles. Nevertheless, the improvement
over the MO predictions is dramatic. Taking the

Acknowledgments
The authors are indebted to Professor Dimitrios
Atmatzidis for his constructive criticism of the work.
Thanks are also due to two anonymous reviewers whose
comments signicantly improved the original manuscript.
Appendix A. Application example
Active and passive earth pressures will be computed for a
gravity wall of height H 5 m, inclination o 5 and
roughness d 20 , retaining an inclined cohesionless
material with f 30 , g 18 kN=m3 and b 15 , subjected to earthquake accelerations ah 0:2 and av 0. The
static counterpart of the problem has been discussed by
Powrie [35].
The inclination of the resultant body force in the backll
is obtained from Eq. (1):
ce arctan0:2 11:3 .

(A.1)

The two Caquot angles are determined from Eqs. (6), (9)
and (15) as
D1 sin1 sin15 11:3= sin 30 62:4 ,

(A.2)

D2 sin1 sin20= sin 30 43:2 .

(A.3)

The angle separating the major principal planes in regions


A and B is computed from Eq. (21):
2yE 43:2  62:4 20 15  2  5  11:3 45:5 .
(A.4)

ARTICLE IN PRESS
968

G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969

Based on the above values, the earth pressure coefcient is


obtained from Eq. (21):
cos5  15 cos15 11:3
K AEg
2
cos 11:3 cos 20 cos 5

1  sin 30 cos43:2  20

1 sin 30 cos62:4 15 11:3


p
tan 30 0:82
A:5
 exp 45:5
180
from which the overall active thrust on the wall is easily
determined (Eq. (11)):
PAE 12 0:82  18  52 185 kN=m.

(A.6)

Both MO and ChenLiu solutions yield K AEg 0:77,


which elucidates the more conservative nature of the
proposed approach.
For the gravitational problem, the corresponding
parameters are D1 sin1 sin 15= sin 30 31:2 , D2
sin1 sin20= sin 30 43:2 ,
2y 43:2  31:2 20 15  2  5 3 , K Ag 0:42.
Thus,
PA 12  0:42  18  52 94:5 kN=m.

(A.7)

The horizontal component of gravitational soil thrust is


determined from Eq. (24)
PAH 94:5  cos5 20 85:6 kN=m.

(A.8)

Note that according to Powrie [35], the horizontal


component is (Eq. 9.42, p. 333)
PAH 12  0:395  18  52 1 tan 5  tan 20 91:7 kN=m,
(A.9)
which is clearly in error as: (1) Ka, as determined from
Powries equations, should be 0.385not 0.395; (2) the
sign in front of product tan b  tan d should be minus
one. (3) Powries equation does not encompass factor
coso  b= cos o cos b arising from the integration of
stresses on the back of the wall.
For the passive case, the corresponding parameters are:
ce Arctan0:2 11:3 ,
D1 sin1 sin15  11:3= sin 30 7:4 ,
2yE 43:2 7:4 20 15  2  5 11:3 86:9 .
The passive earth pressure coefcient and resistance are
obtained from Eqs. (21) and (11):
K PEg

cos5  15 cos15  11:3


2
cos 11:3 cos 20 cos 5

1 sin 30 cos43:2 20

1  sin 30 cos7:41  15  11:3


p
tan 30 6:31,
 exp 2yE
180

PPE 12  6:31  18  52 1420 kN=m.

A:10
(A.11)

The MO and ChenLiu solutions predict K PEg 10:25


and 8.01, respectively. Note that the average of the two
closed-form solutions, 10:25 6:31=2 8:28, is very
close to the more rigorous result by Chen and Liu.

References
[1] Coulomb CA. Essai sur une application des regles de maximis et
minimis a quelqes problemes de stratique relatifs a l architecture.
Memoires de mathematique et de physique. Presentes a l academie
royale des sciences 1776; Paris, 7: p. 34382.
[2] Heyman J. Coulombs memoir on statics; an essay in the
history of civil engineering. Cambridge: Cambridge University Press;
1972.
[3] Lambe TW, Whitman RV. Soil mechanics. NY: Wiley; 1969.
[4] Clough GW, Duncan JM. Earth pressures. In: Fang HY, editor.
Foundation engineering handbook. New York: Chapman & Hall;
1990. p. 22335.
[5] Okabe S. General theory on earth pressure and seismic stability
of retaining walls and dams. J Jpn Soc Civil Eng 1924;10(6):
1277323.
[6] Mononobe N, Matsuo O. On the determination of earth pressure
during earthquakes. In: Proceeding of the world engineering
congress, vol. 9. Tokyo; 1929. p. 17987.
[7] Matsuo M, Ohara S. Lateral earth pressures and stability of quay
walls during earthquakes. Proceedings, second world conference on
earthquake engineering, Tokyo, Japan; 1960.
[8] Seed HB, Whitman RV. Design of earth retaining structures for
dynamic loads. In: Proceedings of specialty conference on lateral
stresses in the ground and design of earth retaining structures. Ithaca,
New York: ASCE; 1970. p. 10347.
[9] Ebeling RM, Morrison EE, Whitman RV, Liam Finn WD. A manual
for seismic design of waterfront retaining structures. US Army Corps
of Engineers, Technical Report ITL-92-11;1992.
[10] Chen WF. Limit analysis and soil plasticity. Developments in
geotechnical engineering. Amsterdam: Elsevier; 1975.
[11] Kramer SL. Geotechnical earthquake engineering. Englewood Cliffs,
NJ: Prentice-Hall; 1996.
[12] Wood JH. Earthquake induced soil pressures on structures. Doctoral
Dissertation, EERL 7350, Pasadena, CA: California Institute of
Technology; 1973.
[13] Steedman RS, Zeng X. The inuence of phase on the calculation of
pseudo-static earth pressure on a retaining wall. Geotechnique
1990;40:10312.
[14] Sheriff MA, Ishibashi I, Lee CD. Earth pressures against rigid
retaining walls. J Geotech Eng 1982, ASCE; 108: GT5. p. 67996.
[15] Finn WD, Yogendrakumar M, Otsu H, Steedman RS. Seismic
response of a cantilever retaining wall: centrifuge model test and
dynamic analysis. In: Proceedings of fourth international conference
on soil dynamics and earthquake engineering. Southampton:
Computational Mechanics Publications; 1989. p. 331431.
[16] Veletsos AS, Younan AH. Dynamic soil pressures on rigid retaining
walls. Earthquake Eng Struct Dyn 1994;23:275301.
[17] Theodorakopoulos DD, Chassiakos AP, Beskos DE. Dynamic
pressures on rigid cantilever walls retaining poroelastic soil media.
Part I: rst method of solution. Soil Dyn Earthquake Eng
2001;21(4):31538.
[18] Theodorakopoulos DD, Chassiakos AP, Beskos DE. Dynamic
pressures on rigid cantilever walls retaining poroelastic soil media.
Part II: second method of solution. Soil Dyn Earthquake Eng
2001;21(4):33964.
[19] Pecker A. Seismic design of shallow foundations. In: Duma A, editor.
State-of-the-Art: 10th european conference on earthquake engineering. Balkema; 1995. p. 100110.
[20] Richards R, Elms DG. Seismic behaviour of gravity retaining walls.
J Geotechn Eng Div 1979;105(GT4):44964.
[21] Whitman RV, Liao S. Seismic design of gravity retaining walls. US
Army Corps of Engineers, Miscellaneous paper GL-85-1; 1985.
[22] Collins IL. A note on the interpretation of Coulomb analysis of the
thrust on a rough retaining wall in terms of the limit theorems of
plasticity theory. Geotechnique 1973;24(1):1068.
[23] Caquot A, Kerisel L. Traite de mecanique des sols. Paris: GauthierVillars; 1948.

ARTICLE IN PRESS
G. Mylonakis et al. / Soil Dynamics and Earthquake Engineering 27 (2007) 957969
[24] Finn WD. Applications of limit plasticity in soil mechanics. J Soil
Mech Found Div 1967;93(SM5):10120.
[25] Davis RO, Selvadurai APS. Plasticity and geomechanics. Cambridge:
Cambridge University Press; 2002.
[26] Atkinson J. Foundations and slopes. London: McGraw-Hill; 1981.
[27] Parry RHG. Mohr circles, stress paths and geotechnics. E&FN Spon;
1995.
[28] Sokolovskii VV. Statics of granular media. New York: Pergamon
Press; 1965.
[29] Papantonopoulos C, Ladanyi B. Analyse de la Stabilitee des Talus
Rocheux par une Methode Generalisee de lEquilibre Limite. In:
Proceedings, ninth Canadian rock mechanics symposium. Montreal;
December 1973. p. 16796 [in French].
[30] Lancelotta R. Analytical solution of passive earth pressure.
Geotechnique 2002;52(8):6179.
[31] Chen WF, Liu XL. Limit analysis in soil mechanics. Amsterdam:
Elsevier; 1990.
[32] Briske R. Die Erdbebensicherheit von Bauwerken. Die Bautechnik
1927;5:42530 4537, 54755.
[33] Habibagahi K, Ghahramani A. Zero extension theory of earth
pressure. J Geotech Eng Div 1977;105(GT7):88196.
[34] Terzaghi K. Theoretical soil mechanics. New York: Wiley; 1943.

969

[35] Powrie W. Soil mechanics: concepts and applications. London:


E&FN Spon; 1997.
[36] Lee IK, Herington JR. A theoretical study of the pressures acting on
a rigid wall by a sloping earth or rock ll. Geotechnical 1972;
22(1):126.
[37] Ostadan F. Seismic soil pressure for building walls: an updated
approach. Soil Dyn Earthquake Eng 2005;25:78593.
[38] Paik K, Salgado R. Estimation of active earth pressure against rigid
retaining walls considering arching effects. Geotechnique 2003;
53(7):64353.
[39] Kloukinas P. Gravitational and earthquake-induced earth pressures
on gravity walls by stress plasticity theory. Ms. Thesis, University of
Patras; 2006 [in Greek with extended English summary].
[40] Salencon J. Applications of the theory of plasticity in soil mechanics.
New York: Wiley; 1974.
[41] Davies TG, Richards R, Chen KH. Passive pressure during seismic
loading. J Geotechn Eng 1986;112(4):47983.
[42] Salencon J. Introduction to the yield design theory. Eur J Mech A
Solids 1990;9(5):477500.
[43] Psarropoulos PN, Klonaris G, Gazetas G. Seismic earth pressures on
rigid and exible retaining walls. Soil Dyn Earthquake Eng 2005;
24:795809.

Potrebbero piacerti anche