Vous êtes sur la page 1sur 30

See discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/245410312

Seismic Stability of earth dams and


embankments

Article in Géotechnique · January 1975


DOI: 10.1680/geot.1975.25.4.743

CITATIONS READS

144 1,445

1 author:

Sarada K. Sarma
Imperial College London
60 PUBLICATIONS 1,715 CITATIONS

SEE PROFILE

All content following this page was uploaded by Sarada K. Sarma on 01 March 2014.

The user has requested enhancement of the downloaded file.


SAllMA, S. K. (1975). Giotechnique 25, No. 4, 743-761.

Seismic stability of earth dams and embankments

S. K. SARMA, BTech, PhD, Die*

In the design of earth daros and embankments under Dans le calcul des barrages en terre et des remblais
earthquake loading conditions, displacements offer a sous I'action des tremblements de terre, les déplace-
ments offrent un meilleur critere qu'un facteur de
better criterion than a factor of safety on shear sécurité sur la résistance a la rupture au cisaillement.
strength. The displacement depends on shear Les déplacements dépendent de la résistance a la
strength, inertia forces and pore-pressures generated rupture au cisaillement, des forces d'inertie et des
during the earthquake. The model of a rigid block pressions intersititelles produites pendant le tremble-
ment de terre. Le modele d'un bloc rigide sur une
on a plane surface is used to analyse the effect of surface plane est utilisé pour analyser I'effet des
inertia forces and pore-pressures on the stability of forces d'inertie et des pressions interstitielles sur la
an earth dam or embankment during strong earth- stabilité d'un barrage en terre ou d'un remblai pen-
dant de forts tremblements de terre. L'étude est
quakes. The study is based on limit equilibrium basée sur des príncipes d'équilibre limite et la
principies and the material obeys the Mohr- rnatiere obéit au critére de rupture Mohr-Coulomb
Coulomb failure criterion with effective stresses. lt en contraintes effectives. 11 est démontré qu'en
is shown that by using the pore-pressure parameters utilisant les parametres A et B des pressions inter-
stitielles, l'exces de pression interstitielle produit
A and B, the excess pore-pressures generated during pendant un tremblement de terre peut étre expliqué
an earthquake can be accounted for in the effective par l'analyse des contraintes effectives. 11 est aussi
stress analysis. It is also shown tbat displacements démontré que les déplacements produits pendan! un
during an earthquake can be easily calculated by tremblement de terre peuvent facilement étre calculés
en utilisant de simples pulsations. Plusieurs con-
using simple pulses. Severa) interesting conclusions clusions intéressantes sont tirées et on donne un
are drawn and an example of application of the exernple de l'application de la méthode aux barrages
method to earth dams and embankments is given. en terre, et aux remblais.

lt has been recognized for sorne time by earthquake engineers that the usual concept of a factor
of safety on shear strength <loes not properly assess the performance of an earth dam or
embankment during strong earthquakes. Toe performance should rather be measured in
terms ofthe relative displacements that the failed mass may undergo during strong earthquakes,
Newmark (1965), Seed (1966). The factor of safety is defined as the factor by which the
available strength along the failure surface should be reduced so as to bring it to a state of
limiting equilibrium with the stresses. In terms of the limit equilibrium principie, a factor of
safety less than one represents failure. This signifies that when the factor of safety is less than
one, the mean strength along the failure surface is less than that required to maintain equili-
brium. Therefore, a section of the dam or embankment will slide along the failure surface
and will come to rest again at a time and a place when the new mean stresses do not exceed the
available strength. Depending on the amount of this relative displacement of the failed mass,
which may vary from a fraction of an inch to a few yards, the structure will be said to be safe
or to have failed. It is therefore obvious that a factor of safety less than one cannot be
permitted under static conditions as the stresses producing this stage will exist until large
displacements change the geometry of the structure.
However, under earthquake conditions, it may be possible to allow the factor of safety to
• Civil Engíneering Department, Imperial College. London,
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
744 S. K. SARMA

NOTATION
A and B Skempton's pore-pressure para- 8 Inclination of inertia force to
meters horizontal
a Base area of block Inclination of inertia force which
F Factor of safety on a possible produces minimum F4
failure plane or maximum Xm
,
Factor of safety before earthquake C1 Effective normal stress on possible
Maximum ground acceleration failure plane
Seismic coefficient as fraction of Total normal stress on sliding
gravity plane during earthquake
Critica) seismic coefficient which Effective normal stress on sliding
produces factor of safety equal to plane during earthquake
one on the failure surface Total normal stress on sliding
Maximum value of seismic co- plane before earthquake
,
efficient C1 o Effective normal stress on sliding
Average seismic coefficient for plane before earthquake
sliding surface in dam as fraction Total major principal stress before
of maximum ground acceleration earthquake
Modified acceleration factor for a' 1, o Effective major principal stress
condition u0 #= O before earthquake
q First root of equation Total minor principal stress before
q=a+(Km/K0)(cos a-cos q)
, earthquake
T Predominant period of earthquake C1 3, O Effective minar principal stress
acceleration record before earthquake
Pore-pressure during earthquake T Shear stress on possible failure
Pore-pressure at failure plane
Initial pore-pressure befare earth- Shear stress on sliding plane before
quake earthquake
Excess pore-pressure generated Shear stress on sliding plane during
during earthquake earthquake
w Weight of sliding block ef,' Effective shear strength parameter
X Displacement of sliding block of the soil
measured down the slope Effective shear strength envelope
Maximum displacement of block reduced by F
sin -1(K 0/ Km) Effective shear strength envelope
f3 Inclination of sliding plane to reduced by F4
horizontal Y,o Effective shear strength envelope
reduced by F0

drop below one as this state will exist only for a short time. There are two possible ways by
which the factor of safety may be reduced to one or below one during an earthquake, even
though the static factor of safety is greater than one. First, earth dams and embankments,
when subjected to strong ground motion will be set into vibration which will produce inertia
forces on the structure. This together with the pre-earthquake static forces may bring the
factor of safety to a value below one. But, if the strength of the soil remains unchanged and
if the displacement of the failed mass is not Jarge enough to change the configuration of the
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABll.ITY OF EARTH DAMS AND EMBANKMENTS
745

Flg. l. Model oí tbe rlgid block on a sloping suñace

structure, the factor of safety after the earthquake will be the same as that before the earth-
quake. Second, it is known that pore-pressures within the soil mass of an earth dam or em-
bankment may increase during earthquake loading (Seed and Lee, 1969). Cyclic loading
tests on soil samples in the laboratory have shown that the pore-pressure within the samples
increases with every cycle of loading and unloading. This continues until the sample reaches
a state of failure after which the pore-pressure remains constant (Shaal, 1972; Seed and Lee,
1966; Lee and Seed, 1967; Finn et al., 1971; Peacock and Seed, 1968). This increase of pore-
pressure will reduce the total strength of the material which may lead to a state of failure.
In the extreme case, this increase of pore-pressure may produce liquefaction within the soil
mass (Seed and Wilson, 1967; Seed et al., 1969; Seed, 1968; Shibata et al., 1971; Seed et al.,
1973). Since dissipation of the excess pore-pressure will take sorne time, the factor of safety
will only gradually increase to its pre-earthquake value, and, therefore, even a small aftershock
may create a state of failure.
Thus, the increase of inertia forces as well as the decrease of total strength dueto pore-water
pressure may reduce the factor of safety of the structure to unity at sorne time during an
earthquake. Any further increase of the inertia force will produce displacement of the soil
mass along the failure surface relative to the main body of the dam or embankment. The
behaviour of the pore-pressure along the slip surface during such displacement is unknown but
may be assumed to be constant and equal to that at the time when the factor of safety was one.
Newmark (1965) proposed the model of a sliding block to compute the displacement of the
failed mass. Goodman and Seed (1966) showed that this model gives reasonable results for
dry cohesionless soil. Finn and Miller (1973) used a finite element model to compute the
displacement. As pointed out by Seed (1966), the displacement depends upon determining
the critica! acceleration as accurately as possibíe, The Author propases to use the Newmark
model to analyse the effects of the inertia forces and the pore-pressures on the factor of safety,
the critica) acceleration and the subsequent displacement during an earthquake. The analysis
is based on limit equilibrium principie and the material obeys the Mohr-Coulomb failure
criterion with effective stresses. In the following analysis, for the sake oí simplicity, we
shall use only cohesionless material. This simplifying assumption does not restrict the model
in any way. In fact, the model of the cohesive material was used by Sarma (1968) to compute
the critica! acceleration for a dam section.

ANALYSIS: FACTOR OF SAFETY


First, the stability oí a rigid block resting on a plane surface will be examined. The surface
is inclined at an angle f3 to the horizontal. It is assumed that the block and the plane are
separated by a thin layer of cohesionless soil having e/,' as the effective strength parameter.
The soil may be dry or wet. The block has a weight Wand base area a (Fig. 1).
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
746 S. K.SARMA

The total normal stress on the plane before the earthquake is


o0 = WcosfJ/a (1)

and the shear stress is


To = W sin fJ/a (2)

Let it be assumed that an initial pore-pressure exists before the earthquake at the base of the
sliding block and is equal to Uo· Since the material obeys the Mohr-Coulomb failure criterion
with effective stresses, the failure surface has a static factor of safety F0 given by
F0 = (cos fJ-UoO/W) tan 4,'/sin fJ (3)
Let an earthquake acceleration Kg act on the base at an angle (} to the horizontal. This will
induce an inertia force KW on the block acting at an angle 8 as shown in Fig. 1. During the
earthquake, the total normal stress on the plane is

od = W[cosfJ-Ksin (fJ-8)] (4)


a
and the shear stress is

Td = W[sinfJ+Kcos(fJ-8)]
a
. (5)

Due to the earthquake, an excess pore-pressure Llu may develop which gives
� = u0+Llu (6)
During the earthquake, then, the failure surface has a factor of safety Fd given by
Fd = [cos fJ-K sin (fJ- 8)-uo0/W-Lluo0/WJ tan 4,'/[sin fJ+ K cos (fJ- 8)] (7)
From eqns (6) and (7), it can be seen that if Llu can be evaluated, then F¿ can be calculated.
At this stage the hypothesis is introduced that if a' and -r are the effective normal and shear
stresses on a possible failure plane and F is the factor of safety, then the state of stress at any
point along the failure surface wil1 be the same as if the friction angle of the material were
,f,=tan-1 (tan </,'/F). With this hypothesis, it is possible to draw the Mohr's circle of stresses.
The circle will pass through the point (a', -r) and will be tangent to the strength envelope inclined
at an angle ,f, at the point. With this hypothesis, it will be seen that for a given failure surface,
the direction of the principal stresses will change during the application of the stresses. It is
of interest here to note that Seed (1966) proposed another hypothesis for the purpose of draw-
ing the Mohr circle, where he assumed that principal stress directions do not change and are
those that are obtained at the failure state. However, it is a fact that during shearing, principal
stress directions rotate and therefore Seed's hypothesis does not seem reasonable.
With the help of the hypothesis, the Mohr circle of stresses for the static condition is now
drawn (Fig. 2). From the geometry of the circle,
o'i. 0 = o'0+T0(tan i/io+sec ,f,0) (8)
u'a.o = u'0+T0(mn,f,0-secif,0) (9)
where
,f,0 = 1an-1(mn 4,'/F0) (10)

Adding u0 to both sides of eqns (8) and (9),


01.0 = oo+To(mn,f,o+sec,f,o)
o3• 0 = o0+-r0(bln ,f,o-sec ,f,o)
ered by ICEVirtualLibrary.com to: IP:
86.0.218.4
On: Fri, 08 Oct 2010 13:00:22 (11)
(12)
SEISMIC STABIUTY OF EARTH DAMS ANO EMBANKMENTS
747

.,.
.,. .,.
---="-.., - -- . . .
/

<,
<,

'"
' \
\
1
1

Fig. 2. Mobr's circle of stresses for the static and dyoamic conditions

Let it be assumed that Fd is also known. In this case it will be possible to draw the Mohr
circle in terms of the effective stresses and from geometry the following will be obtained
o\, d = a'd+Titan ,f,d+sec if,.J (13)

o'«: d = u'd+Td(tan r/,d-sec r/,J (14)


where
'Pd = tan-1(tan 4,'/F.J (15)

Adding ud to both sides of eqns (13) and (14)


ª1, d = ad+Td(tan 'Pd+sec y,J (16)

a3, d = crd +Titan 'Pd - sec if,J (17)

Introducing
Lla1 = ª1, d-a1, o (18)
(19)
pore-pressure Llu can then be related to Lla1 and Lla3 by Skempton's equation (1954). This
assumes that the rotation of the principal axes will not have any significant effect on
the pore-pressure parameters A and B. These parameters are evaluated at failure and are
assumed constant during shearing. A method of evaluating these parameters is given later.
Thus
Llu = B[Lla3+A(Lla1 -Lla3)] (20)
Substituting the values of Lla1 and Lla3 in eqn (20) and introducing eqns (1), (2), (4), (5), (11),
(12), (16) and (17)
áuajW = B{-Ksin (JJ-8)-sin /?[tan y,0-(l-2A) sec ,/,0)
+ [sin P+K cos (JJ-8)) [tan ipd-(l -2A) sec ifod]} (21)
Substituting Llu from eqn (21) into eqn (7) and rearranging terms,

_!l_+Btan</,' B(l-2A)y(l+tan24,'/Fl) =
tan é' Fd
cos P- K(I - B) sin (JJ- 8) + B sin /?[tan ,f,0 -(1 - 2A) sec i/,0)- u0a/ W ( )
22
sin P+K cos (P-8)
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
748 S. K. SAkMA

where from the definition of if¡4


sec !fd = y'(l +tan" cf,'/F/)

Equation (22) is a quadratic in F¿ and can be solved to obtain F4• Since only the positive
values of sec !fd are meaningful, only two of the four roots are admissible. One of these then
can be rejected on physical grounds. Also, depending on combinations of parameters, it is
possible to arrive at imaginary roots, which combinations cannot possibly exist in nature.

CRITICAL DIRECTION OF INERTIA FORCE


It can be seen from eqn (22) that the factor of safety F¿ is a function of the angle 8
among other parameters. Therefore, there exists a critica) angle (Je which will produce a mini-
mum factor of safety. Differentiating F4 with respect to 8 and equating it to zero, we obtain
-sin Be+ B sin /3{cos (/3- 8e) + sin (/3- 8e)[tan 1p0-(l -2A) sec if¡0]}
- u-a] W sin (/3- 8e) = K( 1 - B) (23)
This shows that the critica) direction for minimum F¿ is independent of ,f,' but dependent on
other parameters.

Special cases
Fully saturated sand: B = l. Equation (23) will give Be= /3 and is independent of the value of
A. This shows that for fully saturated cohesionless materials, the mimimum factor of safety
is obtained when the inertia force acts parallel to and down the slope. This implies that fully
saturated sand behaves as a purely cohesive material under earthquake conditions.
Dry sand: B=O and u0=0. Equation (23) will give Be= -sin-1(K) and the value of A is
immaterial.
For other values of B, the Be will lie between (3 and - sin -1(K). Fig. 3 shows the variation
of Be with B for a particular pair of K and (3 values with u0 equal to zero.

CRITICAL ACCELERATION
Critica! acceleration is defined as that acceleration which when applied on the block will
produce a state of limiting equilibrium. This implies that Fd= 1 when K=Ke, Eqn (22)
therefore gives
K _ cos /3 + B sin /3[tan !fo -(1 - 2A) sec ¡/J0]- P sin f3-u0a/ W (24)
e - (1- B) sin (/3- 8) + P cos (/3- 8)

where
P = cot ,f,' +B tan ,f,'-B(l-2A) sec ,f,' (25)

Equation (24) also shows the dependence of Ke on the direction of the inertia force. The
critica! direction Be which gives minimum K¿ will be obtained by differentiating K¿ with respect
to 8 and equating it to zero, which leads to the condition
tan (/3- Be) = (1- B)/P (26)

For B= 1, this again gives 8e=/3 and substitution in eqn (24) gives
minimum K¿ = {cos /3+B sin .B[tan ¡/J0-(1-2A) sec ¡/J0]-P sin f3-u0a/W}/P (27)
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS ANO EMBANKMENTS 749
A - 1·0 A= 0·5 A.� 0·25

1 1 1 1 1 1 1 /-1 /
1,2 lil 1/4 1/5 1/4 1/5 1/) 1/4 1/5 � Slope

,- 451 \ \ \ 1
1
1 1 / I

o< K = 0·20 _ 1 1 1 1
I
/
'- ,� Slope = 1 /3 40
1 1 1
/ .
a.
U0 = 0
/ / j;
�� / '
·--··
\\\ �
\" \\ -,
u-'
\'\- C\\
\ \\\\
- t----·
�\ \\ [\
A - l·O�
-i-
·5- � � [\\ - _j ,
1
0 : -, ,\. \
�\ --
�-------L--· -
0 25
· ..
_, o
1\. \' \\
o' \
' \

_J.2s
\ \\ .:-...
\. \
!
\ \�
��
l---+---·-+rTJ----
' 1 �1
rT77T j


O·J 0
-2 1·0 O·I 02 ·4 0·5 06
o 0·5
Cric lcil accclcratlon, K e : g
B

Fig. 3. Typical variation of 80 with pore- Fig. 4. Variation of critica) acceleration with .¡;
pressure parameter B

For B = O, eqn (26) gives 80 = - (</>' -/3)' and eqn (24) then gives
minimum K¿ = sin(</>' -{3) (28)
which is the same as that shown by Seed and Goodman (1964). Fig. 4 shows the relation-
ship of K0 to </>' for various combinations of A and f3 with B = 1 and u¿ = O. In these curves, (J is
assumed to be equal to zero for simplicity and also because the angle has little influence on the
final result. One interesting feature of these curves is that for large values of the parameter A,
an increase in</>' will at first cause an increase in K0; but, after a certain value, K0 starts reducing.
Though analytically correct, this is not reasonable; on closer examination, it is seen that it
happens because of incompatible combination of parameters. For example, one does not
expect to obtain a large value of A for materials having a large </,'. Therefore, upper portions
of these curves are rejected. Fig. 5 shows the stress path to failure as K increases from zero
to K0• In obtaining these paths it is assumed that A is constant and equal to that at failure,
which is not strictly the case.
The upper parts of Figs 6(a) and (b) show the factor of safety F« which reduces from F0 to
unity when K increases from zero to the critica! value K0• Full lines show the factor of safety
when the direction of application of the inertia force is critica!. The dotted Iines show the
factor of safety when the direction of application is horizontal. As can be seen, there is not
much difference between the two values and therefore, horizontal acceleration may be adopted
for stability analysis calculations.
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
750 S. K. SARMA

B-0-S
=
Slope 1/3
8- (J,
ef,-35°
Uo -0

..
.-
.- .-
----
Numben wlthln brackets show K¿

O' C>'o

Fig. S. Typical stress path to failure from static coodition

The effect of the initial pore-pressure u0 is not shown in Figs 6(a) and (b). It is obvious
that the effect of the initial pore-pressure is to reduce or increase both the static factor of
safety F0 and the dynamic factor of safety Fd, depending on whether the initial pore-pressure is
positive or negative. For the condition u0#0, it can be shown that the factor ofsafety F¿ can
be obtained by using the same graphs but by using a new acceleration K¿ which is given by
K¿ = (K - Q tan /3)/(1 + Q) (29)
where
Q _ cos /3(tan ,/¡0-tan /3)(B sin /3-cos f3 cot ,/¡0)-0·5B(l-2A) sin 2f3(sec ,/¡0-sec f3) ( 30)
- l-B(l-2A} sin f3
and
n f3
tan ,/¡ O = --,,.s..i� --,-.,. ,. (31)
cos f3-u0a/W

Equations (29)-(31) are derived for horizontal earthquake accelerations As an example,


referring to Fig. 6(b), for u0 = O and </,' = 45º, K = O· I (horizontal} gives a factor of safety of 1 ·94.
Other things being equal, the condition u0 # O, say of u0a/ W = O· 3, will result in increasing K to
K¿ = 0·25, to which the corresponding factor of safety is 1 · 33. Similarly, Fig. 4 can be u sed to
determine the critical acceleration for any given u0•
Solution of eqn (22) in a graphical form is given in the Appendix, which may be used
to determine the factor of safety or critica) acceleration for combinations of parameters with
(J equal to zero.

DJSPLACEMENT
If the earthquake acceleration is bigger than Kc1s, the factor of safety will be less than one,
which constitutes a state of failure. In this case the block will slide along the surface and will
come to rest sorne time after the earthquake acceleration has become less than Kc1s. The net
displacement depends on the magnitude and duration of the earthquake acceleration.
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF l!ARTH DAMS ANO EMBANICMENTS 7Sl

8

1
..,o
"' "' .....
S:'.

� '<
NffGr-4

·¡ 'e 1¡
¡"'
"&- .e
���� .. ;!::
t. [Ir:

=e
,¡¡

,¡¡

¡."
_m�
C4

:a
'CI
=

- -+���
C4
.t>
�C4
'S"'
..
� 8. :1º
o
¡;; ....Q
�"
----
·e"
C4

..=....
'CI
Q

..... ·.e=Q
·!C4
;...
.:.. M <;>
-r \/;
"' ·1 .. �M "'
e: �
.... "<
E i

Delivered by ICEVirtualLibrary.com to:


IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
752 S. K. SARMA

When K > K0, the driving force down the slope is


D = W(sin .B+K cos (.8-0)] (32)

and the resisting force is


R = W[cos .B-K sin (.8-0)-ura/W] tan e' (33)

We assume that the pore-pressure does not change during movement and that it is equal to that
at the state of limiting equilibrium. This means that ur equals ud when K equals K0• There-
fore, substituting eqn (24) in eqn (21) and the resulting expression for Llu in eqn (6),

ur = W {[cos .B-Kc sin (.8-0)]-[sin .B+K0 cos (.8-0)] cot </,'} • (34)
a
The acceleration of the block relative to the plane surface is then given by
-Wx.. = D - R
. (35)
g

where x is the relative displacement measured down the slope and x=d2x/dt2•
Eliminating D and R by eqns (32) and (33) and u, by eqn (34),

x= cos (,8-0-</,') (K- "') . (36)


g COS </,' no

The solution of the above eqn (36) depends on the variation of K with time.
For a rectangular pulse of duration T/2 and amplitude Kmg (Fig. 7(a)), solution of eqn (36)
gives
4xm [ COS </,' ] . (37)
KmgT2 cos (/3- 0-4") = 1(Km/Kc-1)

where Xm is the total displacement.


For a triangular pulse of duration T/2 and amplitude Kmg (Fig. 7(b)), solution of eqn (36)
gives

K�;T2 [cos �� t�</,')] = J4 {4(1-Kc/Km)(l -.\Kc/Km)-[l -.\(K /Km) ]}/(K /Km) 0


2
0

forO � K0/Km � [l-y1(1-.\)]/.\ (38)


= i {(l-K /Km)ª[2-2y1(1-.\)-.\]}
0

for [l-y1(1-.\)]/.\ � K0/Km � 1 (39)

For a half sine pulse of duration T/2 and amplitude Kmg (Fig. 7(c)), solution of eqn (36)
gives

for 1 � K0/Km � 0·725 (40)


= [K.,/Km+a-1r+cos2 (a/2) cot (a/2)]/,r2
for O � K0/Km � 0·725 (41)
where
q = a+Kc/Km(COS a-cos q) (42)
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS AND EMBANKMENTS
753

Kmf-------, Km ---

T/2 T/2
(a) (e)

Km ---

10 IS 20 25
Time: s
(d)

Fig. 7. (a) Rectangular pulse; (b) triangular pulse; (e) balf sine pulse; (d) typical earthquake record
and
(43)
The solution for earthquake records (Fig. 7(d)) is obtained by assuming piece-wise linear
acceleration. In this case movement is allowed in the downhill direction only.
Figure 8 shows the relationship [(4xm)/(KmgT2)]/{(cos fl/[cos (�- 8-,/>')]} against Kc/Km
for the four cases just described. For the earthquake records, T is the predominant period
obtained from the acceleration spectrum of the record; Km is the maximum acceleration of the
record. Severa! earthquake records are chosen and these also include records obtained from
nuclear blasts. For the triangular case A is chosen at 0·5. As can be seen from the graph, the
points for earthquake records corresponding to higher values of Kc/Km can be made to coincide
with the triangular pulse curve by slight change ofthe period T. Even otherwise, the difference
between the points for earthquake records and the triangular pulse curve in terms of the
value of Xm is almost negligible. For small val u es of Kc/Km, the effect of the number of pulses
in earthquake records seems to become important. The points still appear to be bounded by
the rectangular pulse curve. lt can therefore be said that for K] Km greater than 0·5, the
triangular pulse and for Kc/Km less than 0·5 the rectangular pulse will effectively give the dis-
placement of a sliding block subjected to an earthquake.
Equation (36) shows that the net acceleration and therefore the maximum displacement
depends on the angle 8 as well as on the critica! acceleration which is a function of 8. There-
fore, there exists a critica! angle which will produce maximum displacement. The lower
part of Figs 6(a) and (b) shows the dimensionless displacement (4xm/KmgT2 ) plotted against

Km when Km is greater than K; for the rectangular pulse and for u« equal to zero. The
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
754 S. K. SARMA

o
T ··-

y CD Rectangular pulse

a> Triangular pulse ----


c .. cos
cos¡e + 13- qi) -
-
-,
\ -. - (]) Half sine pulse --�--- ---- O..., El Centro
N
S (T- 0·5 sec)

1940 -

- • " Koyn, 1967 -


\\ �� x
L (T-0·15 sec)
== Pacolma 1971

1 S 16E (T 0·4 sec) -


·-� ., Parkfield 2 1966

' -,
-.;:· """ <,
- . --
N65 E (T- 0·6 see)"
<, <,
v = Several nuclear -
-, <; r-, blasu -·

. .. "'- � --......,...

<,
s-,
<,
�-



1
.
. - -
<,
�-- �--
'
-- -·
..

·'-·
-. ·-
!"-.CD
-·,
-

'..:,.- --
' ·- - ·-·-
' '
;
·�
-·- - <, --
-- -- """')
'-- -··--·- 1-----

-, \

--
-,,,
O·O 1
1

-·-·-·---
1
\
'
·- -· \
\
-- \
f--·-
"\ \
0·00 l._ - - \ \
O·I 0·2 O·J 0·4 0·5 0·6 0·7 0·8 0·9
K, /Km

Fig. 8. Variation of dimensionless displacement with K0/Km

corresponding factors of safety are also shown and these are less than one. The dotted lines
show the case when earthquake force acts horizontally and the full line represents the critica!
case. From the figures, it can be seen that the direction of the inertia force has Iittle effect on
the displacement. Also for factors of safety between 0·9 and 1 ·0, the displacements are small.

DISCUSSION
lt is mentioned in the introduction that the computation of the displacement of the sliding
block depends upon determining the critica! acceleration as accurately as possible. From the
foregoing analysis, it is apparent that the excess pore-pressure generated during earthquakes
plays a dominant part in determining the stability of earth structures, whether in terms of the
dynamic factor of safety or critica! acceleration. In this analysis, the excess pore-pressure is
accounted for by a relationship of the type
Llu = B[Lla3+A(Lla1 -Lla3}]

which was proposed by Skempton (1954). The purpose of this expression is to relate the change
of pore-pressure to the change of stresses. This relationship was derived under static condi-
tions and no such relationship has been proposed for cyclic tests. Since this expression is of a
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS AND EMBANKMENTS
755

Number of cycles
I to cause faiJure
1

Fig. 9. Variation of A and K0 with number of cycles

0·9

0·8

0·7 IJ - o
«o= O

l<"' 0·6
e

s
o 0·5

] 0·4

0·3

Q-2

Fig. 10. Relationship of seismic acceleration and dynamic factor of safety

general nature, it is assumed that by proper determination of the parameter A, this expression
can be used for cyclic loading conditions. A possible alternative to the use of this equation
would be to derive a relationship between stress change and pore-pressure change from cyclic
loading tests and modify the theory accordingly.
From the static tests, it is seen that the parameter A is dependent on stress path. It is also
expected that this parameter will vary from cycle to cycle in cyclic tests. It is noticeable that
the rate of increase in pore-pressure is decreased with the number of repetitions (Finn et al.,
1971; Lee and Seed, 1967). This implies that the parameter A increases at a reducing rate
with every cycle when the pore-pressure is measured from the beginning of the test.
A possible experimental procedure to determine the parameter A for use in the stability
analysis of an earth dam or embankment is outlined in the following. First, the expected
failure surface is judged and the pre-earthquake static stresses determined. The laboratory
sample should then be consolidated anisotropically under these stresses. The sample is then
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
756 S. K. SARMA

! ! 1
0·9>---+--+--+--+-+-+-+-+-+--
B= l·O 0·91 . B= l·O '
O· -+---+--+--+--; 0·8� --,----1----+--+--+--
.,:: O 71--f--l,__·f--lf--1--t--t--t----t- ,:- 0·7 . A = 0·5 1

: 1
2 O· ·-+---+- -+--� Slopc

�g o . 'v-+--C----,- ·- - --- -

·� 0·4'--+-+---+---r- +--+--t--t- Slope


�e o) 1 i
o Vi-t-t::P-:�:==:;!
< 0·2f--l-t,.....¡"-t-t-t-±=l:::-:±- Y--!-+,,..17-''F-::::�t==Fl tl
0· 1HfL-J-1717t"'::::f--t--t-+

tan r/,'1Fd

09'1--1 1
1-- lt-
-- - tt- _• J 0·9 >--J----+---+---+---+-----t-----t--+- Slope
) B = l·O-+-+-t--t--[f--1--
0 -+--t----+-+-+-+--c o
� 0 7 f--lf--l = -;·7_-+----+---+--+-+-·-t-
� 0·61--f--1--1--1---t----t�---+--t--
] O·S Slope 115

·!0·4>-----+---+---+----+..---±::-.:=!=-"'!"'--= 1-1-1-1-A-t'--l-.r!�=-l--=-1 t)
IH

� 0·1f-i-l-f---,,f.L:-t-t-t-t--
_.,,V : -- - -111154 1, 2


�-----¡¡¡
(}21
' /
11/1 /,V_., ..,....-,--·-----
1/ 2
O O· I 0·2 O·l 0·4 0·5 0·6 0·7 0·8 0·9

Fig. 11. Relationsbip of seismic acceleration and dynamic factor of safety


subjected to pulsating deviatoric stresses of different magnitudes under undrained conditions
whilst pore-pressure is measured and the curve shown in Fig. 9 determined. This curve
is similar to that produced by Seed and Lee (1966) and various other investigators.
The parameter A is also calculated in each case, the pore-pressure being measured from the
beginning of the test.
Once these parameters are determined, the earth dam section may be analysed applying the
limit equilibrium principie, as shown by Sarma (1968). In this analysis, the technique of
including the excess pore-pressure generated during the earthquake is the same as that shown
in the sliding block problem. But this analysis must also include the initial pore-pressure and
the cohesion in the material. The critical acceleration K0 is determined for each value of A
and can be plotted as shown in the lower part of Fig. 9. This then gives the required number of
cycles to failure at various levels of K0• From the design earthquake record, the number of
pulses greater than or equal to each value of K0 is determined. The smallest value of K0
which will satisfy the requirement of the number of cycles will be the critica! acceleration for
that surface. From severa! tria! surfaces, the surface with the mínimum K0 will give the
critica! surface.
When the maximum design acceleration is greater than K0 it would be necessary to compute
the displacement of the failed mass. In order to utilize the sliding block model, two other
parameters are needed; these are the equivalent slope angle f3 for the failure surface and the
average.¡,'.
A very simple way of determining the equivalent slope angle is to sum the shear forces along
the failure surface vectorially. The direction of the resultant shear force vector will give {3.
Knowing the values of K0, f3 and the pore-pressure parameters, a value of the average ef,' can be
obtained from Figs 10-13. The displacement of the mass can then be derived from Fig. 8.
(In case of cohesive material, a weighted average value of ef,' should be determined for the
slip surface and Figs 10-13 should not be used.)
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS AND EMBANKMENTS 757
Slope:O 1/5 l/4 Slope Ol/5 1/4 113 112
v'
0·9
l/3

1' V
1

08 o / B-1-0

.,. J A--05
..:-01 1,2 .. o. 1
g 06 ··-·r-· �
� O·S
o
5 -
�o-.. I J
] O· ·3
4
V n1
,< O· 2
/ '/
V I
O· ·1/
O 01
V; /
0·2 0·3 04 05 O·• u ., O·• , .,

115 l/4 113 112

� O·
.:! 0·51--t--,r--t-r-t--f-t-'rf--rcrt----t-
c
.g 0-4,--,---,--,,--,-,�..,.,.--,.,-,--,--
i� O· )t--t--t,�-'l-f-h'-+-f1-+--t--
� O·lr--;--r1---,-,-,-.,.._-,.,,-,--+--+--

Fig. 12. Relationsbip of seismic acceleration and dynamic factor of safety

1 1

" o ·8>-- � A-1·0


..
B-0·5
.,.0·8
·1 �07
·6 §o-6ir-t--+-,r--+--t--r'-,r--t--t---,, 115

4
¡._.- 1--
- � 0·4 113

·S Slope:� 0-S 1/4


ir-t--+-,r--+-,r-+-,1--:*"'°-j,,'�
· o .g
vv
1/5 �0-3
·3
,,v 1--: j:::::: ¡::.... +-- 1/4 < O·
·2

11/
1/
1/.'. V,. V -¡...--
-
j� V ¡....--,_
l/3
- 1,2 01

O· I 0·2 O·l 0·4 0·5 0·6 0·7 0·8 0·9 O·I


tan </>'tFd Slope:

0·9
1/5
o-0·8 ---t-t--t-,"T--t-�rr- 114
Slope: .. 0.7
o l;
] 0·6
� 0·5
1/5

114
5 0·4
-¡;
1/3

112
-ct: 0·3
0·2
O·I
Fig. 13. Relationship of seismic acceleration and dynamic factor of safety
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
758 S. K. SARMA


c.,. 1 Case2

Surf1ce Fo K, (%8) f2.. K, (%1)


2·96 0·57 1·76 0·19
3-01 0·56 1·81 0·19
3·31 0·57 2·01 >0·20
3-92 0·60

I�

Cue2
.l B 015

c'(lbift2) ro
-y (lbift l) 350 350
i�s ,�s
'• O·�

tan ,p 0761 0·76l


Fig. 14. Homogeneous earth dam with four possible slip surfaces

EXAMPLE
Figure 14 shows a homogeneous earth dam. The properties of the material and the results
of the stability analysis of the dam performed by the Sarma (1968) method is shown in the inset.
Let us assume that surface No. 2 is the critical surface, which can withstand an acceleration
equal to 0·2g with the factor of safety being greater than or equal to one.
Figure 15 shows the vector diagrams of forces for this slip surface. This gives the angle for
the equivalent slip surface as 14·5º. Average</,' for this surface is 37·4º.
In order to obtain Km, let it be assumed that the dam is subjected to an El Centro type
earthquake (Fig. 7(d)) which has a maximum ground acceleration gmax =0·5g and predominant
period T=0·5 s. Assuming that the fundamental period of the dam is 0·25 s, for a 100 ft high
dam, this will correspond to a shear wave velocity of 1040 ft/s in the dam. From Fig. 16
extracted from Ambraseys and Sarma (1967), the seisrnic coefficient for the surface is K = l ·8.1
This therefore gives Kmg=Kgmax=0·9g. Therefore, for this surface (Kc/Km)=0·2/0·9=0·22.
From Fig. 8, so, using the rectangular pulse, the displacement that the failed mass will undergo
is Xm =43 in. which gives a vertical displacement of 10·8 in. anda horizontal displacement of
41 ·6 in.
This shows that the dam which can withstand an acceleration of 0·2g with a factor of safety
equal to one, will undergo sorne deformation when subjected to a ground acceleration as large
as 0·5g with a predominant period of0·5 s. Provided there are no buried structures inside the
dam which may be incapable of deforming by 42 in. in the horizontal direction or 11 in. in the
vertical direction, the dam may be considered safe during the earthquake.

CONCLUSIONS
From the analysis of the stability of the block the following conclusions may be drawn.
Both the factor of safety and the displacement along a failure surface depend on geometry,
strength of material, pore-pressure parameters and the size of the inertia force.
' This in fact is an overestimate, since the increase in the angle b of the slip surface will reduce the seismic coefficient.
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS AND EMBANKMENTS 759
The duration of the pulse is a major factor in determining the total displacement. Dis-
placement is proportional to the square of the duration.
Both the factor of safety and the displacement are insensitive to the inclination of the inertia
force. Therefore, a horizontal earthquake acceleration can be used in the stability analysis
without creating much error.
In general, for a factor of safety less than l ·O but greater than 0·9, instability of the slope
induced by a strong earthquake implies very small displacements.

Slip surface 2 (case 2)


S = Shear force vector

Ñ - Normal force vector

Scale of pcunds force

0 ( 2 ) 4 5 X IOS

Fig. 15. Vector diagram of forces for the critical slip surface

Fig. 16. Variation of K with fundamental period and depth (after Ambraseys and Sarma, 1967)
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
760 S. K. SARMA

For calculating displacements, a triangular pulse adequately represents a strong earthquake


when the amplitude of the pulse is equal to the maximum of the record and the duration is half
the predominant period. This applies when the critical acceleration is greater than half the
maximum, which will generally be the case in the design of dams. When the critica! accelera-
tion is Iess than that, a square pulse should be adopted.
The displacement computed from the rigid block analysis offers a method of calculating
the approximate displacement of a section of an earth dam or embankment, when the design
acceleration for the surface is bigger than the critica! acceleration. Since the displacements
depend on the material properties and the possible earthquake, which are not really known, a
more refined method of analysis will not produce any better result for design purposes.

ACKNOWLEDGEMENTS
The work presented in this Paper formed part ofthe thesis ofthe Author for the PhD degree
presented in 1968. As such, acknowledgements are due to ali my friends who actively helped
me with many discussions at that time. Particular acknowledgements are due to Professor
N. Ambraseys who supervised the research at that time. The Author is grateful to Dr D.
Papastamatiou, Dr J. Tchalenko and Dr P. Vaughan for reading the manuscript.

APPENDIX
Equation (22) may be written (with ll=O)
�+ B tan</,' -B(l-2A)v'(l +tan2 rf,'/F.2) =
tan é' F4
cos {3- K0(l - B) sin {3 + B sin {3[tan {3- (1 - 2A) sec {3]
(44)
sin f3+K0 cos {3
where

K0 = (K- Q tan {3)/(1 + Q) . . (45)


and for {3 #- O

Q _ cos {3(tan .µ0-tan ,B)(B sin ,8-cos ,8 cot ip0)-0·5B(l - 2A) sin 2,B(sec .µ0 - sec {3) . (46)
- 1-B(l-2A) sin {3
and
tan .Po = sin ,8/(cos ,8-uoa/ W) (47)
For ,8=0,
Q = -u0a/W (48)

Figures 10-13 show plots of tan rf,'/F4 against K0• These figures may be used either to determine the
factor of safety Fo for a given K orto determine the critica] acceleration for a given </,' when the pore-pressure
parameters and the slope is given. When the initial pore-pressure u¿ is zero, K0 equals K and therefore these
curves directly give Fo or K0• For the condition u0 # O, the procedure is as follows. In order to obtain the
factor of safety F4 for a given K, first convert K to K0 with eqns (45) (46) (47) or (48) and determine tan
rf,'/F4 for this K0• In order to obtain K¿ for a given rf,', first determine K0 for tan rf,'/Fo=tan rf,' and convert
K0 to K through eqns (45)-(48), which gives K.,.

REFERENCES
Ambraseys, N. N. & Sarma, S. K. (1967). Response of earth dams to strong earthquakes. Géotechnique 17,
No. 3, 181-213.
Goodrnan, R. E. & Seed, H. B. (1966). Earthquake induced displacements in sand embankments. Jnl Soil
Mech. Fdn Div. Am. Soc. Civ. Engrs 92, SM2, 125-146.
Finn, W. D. L. & Miller, R. I. S. (1973). Dynamic analysis of plane non-linear earth structures. 5th
World Conf. Earthquake Engng, Rome, Session lD, paper 42, 360-367.
Finn, W. D. L., Pickering, D. J. & Bransby, P. L. (1971). Sand liquefaction in triaxial and simple shear
tests. Jnl Soil Mech. Fdn Div. Am. Soc. Civ. Engrs 97, SM4, 639-659.
Delivered by ICEVirtualLibrary.com to:
IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
SEISMIC STABILITY OF EARTH DAMS AND EMBANKMENTS 761
Lee, K. L. & Seed, H. B. (1967). Cyclic stress conditions causing liquefaction of sand. Jnl Soil Mech. Fdn
Div. Am. Soc. Ciu. Engrs 93, SMI, 47-70.
Newmark, N. M. (1965). Effects of earthquakes on dams and embankments. Géotechnique 15, No. 2.
139-160.
Peacock, W. M. & Seed, H. B. (1968). Sand liquefaction under cyclic loading simple shear conditions. Jnl
Soil Mech. Fdn Div. Am. Soc. Civ. Engrs 94, SM3, 689-708.
Sarma, S. K. (1968). Response characteristics and stability of earth dams during strong earthquakes, PhD
thesis, University of London.
Seed, H. B. (1966). A method for earthquake resistan! design of earth dams. Jnl Soil Mech. Fdn Div. Am.
Soc. Civ. Engrs 92, SMl, 13-40.
Seed, H. B. (1968). Landslides during earthquakes dueto liquefaction. Jnl Soll Mech. Fdn Div. Am. Soc.
Civ. Engrs 94, SMS, 1053.
Seed, H. B. & Goodman, R. E. (1964). Earthquake stability of slopes of cohesionless soils. Jn/ Soi/ Mech.
Fdn Div. Am. Soc. Civ. Engrs 90, SM6, 43-73.
Seed, H. B. & Lee, K. L. (1966). Liquefaction of saturated sands during cyclic Joading. Jnl Soil Mech,
Fdn Div. Am. Soc. Civ. Engrs 92, SM6, 105-134.
Seed, H. B. & Wilson, S. D. (1967). The Tumagain Heights landslides, Anchorage, Alaska. Jnl Soil
Mech. Fdn Div. Am. Soc. Civ. Engrs 93, SM4, 325.
Seed, H. B. & Lee, K. L. (1969). Pore water pressures in earth slopes under seismic loading conditions.
Proc. 4th World Conf, Earthquake Engng; Santiago, Chile, Session AS.
Seed, H. B., Lee, K. L. & Idriss, l. M. (1969). Analysis of Sheffield dam failure. Jnl Soit Mech. Fdn Div.
Am. Civ. Engrs 95, SM6, 1453.
Seed, H. B., Lee, K. L., Idriss, l. M. & Makdisi, F. (1973). Analysis of the slides in the San Fernando dams
during the earthquakes of February 9, 1971. Report No. EERC-73-2, Earthquake Engineering Research
Centre, University of California, Berkeley.
Shaal, B. Z. (1972). The behauiour of cohesionless soils in simple shear under cyclic /oading. PhD thesis,
University of London.
Shibata, T., Kawakami, F., Yamagisawa, E. & Kakayama, T. (1971). General report on the Tokachi-Oki
earthquake of 1968. Part IV 1-3: Damages to soil structures. Tokyo: Keigaku Publishing Co.
Skempton, A. W. {1954). Pore pressure parameters A and B. Géotechnique 4, No. 4, 143.

Delivered by ICEVirtualLibrary.com to:


IP: 86.0.218.4
On: Fri, 08 Oct 2010 13:00:22
View publication stats

Vous aimerez peut-être aussi