_.__ _,_ _ O_ Asso¢i.tiorl for |nformation and Imllge M.rlagemeltt _,,_ _._ _.

j,

Centimeter
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 mm

i,,,,i,,,,i,,,,i,,,,i,,,,I,,,,i,,,,i,,,,i,,,,I,,,,I,,,,I,,,, I
1 2 3 4 5

Inches iiii1,.o
,,,,,3_____.2
illll:
'" _,-_
IllU_
,_ BNL-60611

SEISMIC SAFETY OF EARTH DAMS. A PROBABILISTIC APPROACH

N. Simos, C.J. Costantino and M. Reich

Department of Advanced Technology
Brookhaven National Laboratory, Upton, NY 11973

ABSTRACT

The evaluation of the potential .for slope sliding and/or liquefaction failure of earthen
darns subjected to earthquake loadings i'J most often based on deterministic procedure8 of
both the ezcitation input and of the physical model. Such treatment provides answers in the
form of either factor of safety values or a yes or no as to whether liquefaction will occur
or not. Uncertainties in the physical properties of the soil in the embankment and the
foundation layers underl_ling the darn are typically treated with parametric studies. While
eztensive soil testing can compensate for the luck of such information, questions on what
deterministic earthquake to use aJ representative of the site remain.
Consideration o,f probabilities pertainin9 to the uncertainties of the earthquake and of
the site characterization is ezpected to augment the prediction of failure potential by as.
sociating slope and liquefaction failure to generic properties of the earthquake and of the
site characterization. In thin study, the procedures for conditional slope failure/liquefaction
probabilities are formulated based on a series of simulated deterministic analyses of a dam
cross section. These synthetic earthquakes emanate from a I.D stationary stochastic pro-
cess of zero mean and an analytical form of power spectral density function.
The response of the dam section is formed upon a dynamic finite element approach
which provides the temporal variations of the _tresJes, attains and pore water pressure
throughout the model. The constitutive response of the granular soil skeleton and its cou-
pling with the fluid phase is formulated based on the Bier dynamic equations of motion
with nonlinear terms compensated for into soil hvsteretic damping. Lastly, a stochastic
approach to liquefaction based on the translating of the input motion statistics to the cross
section is presented.

1.0 Introduction involving the Unit Response of the do-

main that can eventually provide the sys-
During an earthquake event of consid- tern response resulting from a seismic event
arable duration and intensity an earthen which in turn represents a stochastic pro-
dam can experience partial or total failure cess. Such procedure, however, implies
that stems from either loss of soil strength that the domain exhibits linear behavior.
due to liquefaction or reduction of the in- While in typical soils this is only true in
herent resistance to sliding along a poten- small strains, the benefits of the Unit Re-
tial failure surface. In trying to assess sponse approach compensate for the lack
the potential to failure in either mode one of nonlinear considerations.
has to essentially incorporate two types of For a study case a hypothetical earth
probabilities. One that is associated with dam built over a saturated soil layer was
the occurance of an earthquake in the prox- considered, Figure 1. The constitutive re-
imity of the structure and the other with sponse of the granular soil skeleton and its
the state of the soll and its tendency to coupling with the fluid phase is formulated
liquefy or fail in shear, around the Blot dynamic equations of too-
The coupl:ng of the two essential corn- tion. The finite element analysis utilized
ponents (excitation and soil state) with in- in the evaluation of the dynamic response
herent statistical properties can only be is a linear in character but it treats the soil
achieved through response analyses that as a two-phase medium. While the draw-
allow for the combined statistics to partie- back of linearity is somewhat compensat-
ipate. A promising approach is the one ed with the equivalent hysteretic damping,

Th±s work was done under the auspices of the U.S. Dept. of Energy.
 t

it is the two-dimensional pore water/soil The stress field in the system is the
sceleton interaction that provides a realis- result of the soll overburden end of the hy-
tic description of the behavior of the soil in drostatic action of the water in the reser-
a dynamic mode. The solution takes place voir. The discretized cross section of the
in the frequency domain and the resulting embankment and its foundation are consid-
harmonic response, inverted with the use ered to be in a plane strain state of stress
of Fast Fourier Transform techniques, pro- witile values of the elastic properties for the
rides the intergranular stress as well as the different soils were assumed.
pore water pressure fluctuation during the The effective overburden stress o'0 is
seismic event, deduced from the relation,
The failure potential, viewed in the for-
m of a factor of safety against slope failure, or0 = 7°vv - ap/ (1)
is evaluated by incorporating the in-situ
and seismically induced dynamic stresses where ru_ is the overburden stress, a is
over various potential surfaces were sliding the compressibility of the soil fraction and
can occur during an earthquake event. The Pf is the pore pressure deduced from the
driving forces in the liquefaction process is solution of the equation governing seepage
the dynamic (cyclic) shear stress that is dh(z,y) _h(z,y)
generated in the soil layers and the buildup kz + kv - 0 (2)
of pore pressures. While a linear analysis Oz2 tgY_
cannot predictthe buildup and the dlssi- and the relation
pationofthe pore pressures, itcan provide
the level of sheafing that the soil experi- Pf = h - Y (3)
ences during the seismic event. This in
turn can become the basis for assessing the where, h(z,y) = total head and kz,ky =
susceptibility to liquefaction soil permeabilities.

2.0 Seismic Response of a Dam Dynamic analysis of the 2-phase

medium

Initial Effective Stress State In assessing the dynamic response of

The stress field prior to a seismic even-
t that exists in the embankment and the
foundation layer are an important elemen-
t in the stability integrity evaluation. The
static stress conditions are key components
in the definition of the factor of safety a-
gainst slope failure. Further, the overbur-
den initial stress is vital to a liquefaction
potential analysis because of its relation
with the effective stress that controls the
process of liquefaction,
For the case of a saturated medium,
the initial total stress vector
['r°] = [r°z rl_°y -rz°y]T
is computed by POROSLAM. These
stresses are the result of the body forces
while taking into account the presence of
water in the pores (increased effective den-
sity) and the hydrostatic loads resulting
from the reservoir behind the dam.
the embankment-foundation the saturated
state of the soil must be accounted for.
The pore pressure of the water trapped in
the soil skeleton will fluctuate during the
earthquake and impact on the intergranu-
lax soil stresses. Since the strength of the
soil is tied to the intergranular stresses, it
is vital that the dynamic pore pressure be
captured. The coupled behavior of pore
water and soil skeleton requires that the
medium must be treated as a two-phase
one with governing equations that reflect
the coupling.
Further, the ability of the soil to resist
liquefaction is one hand dependent on its
initial stress state (effective stress) and on
the other on the intensity of the dynamic
shear stress. The shear stress variation at
different locations in the embankment and
the foundation as well as the number of
stress cycles during the earthquake event
determine whether the soil is susceptible to
such failure.
i

Therefore, to effecti,,_ly analyze displacement vector while considering hys-

the system, the employed theoreti- teretic damping, takes the form
cal/eomputational model must enable;

a. The
phasedescription
medium. of the domain as a two- {r} = Ec In1] [Do] + [D31 _ {uz, uy} T+
b. The implementation of actual or repre-
sentative earthquake input, a2M [D2] {ut uy} T + aM [D2] {wz tog} T
c. The evaluation of the time variation ' '
of stresses resulting from the seismic (5)
input, where,
In order to perform the dynamic analy- [ 1 1v-u 0 ]
the POROSLAM code is employed. The
sis,
codewhich
is a satisfies the above finite
two-dimensional requirements,
element
representation of Blot's dynamic equation-
[Do] = 1-u
[ ..x_u
0 10
1-2v
0
2(1-----_]
s for both soil and fluid phases. Blot's
equations are a linear description of the re- [ )_c _ 0
water in the form, [D3] = _l-u )_e 0
0 0 _,(1--2_)
sponse of the soil skeleton and of the pore [ 1-v
0_'zz Or_y 2(1-v)
--
Oz + _---_ = g6z + g ftbz

[ ] [0o]
_ 0 _ N

Oz -_ 0y -- g'u'V + gfwV 0-'_ 0-'_

and

Opi 1 , . PI = -aM (e,,, + evu)-M \ Oz + Oy ]

Oz = gf6z + ]gfwz + _wz
_c is the hysteretic damping ratio associ-
Opf 1 . r/. ated with hydrostatic compression while
Oy = gf_v + ] #fwv + h wv (4) )_, represents the damping ratio(l-,,)e
associated
with shear strains and E¢ = (l+u)(i-2u)"
where,
[uz,u./j] = components of displacement The implementation of transmiting
of tnesou boundaries on the two sides of the model al-
[w,, wy] = components of displacement low
Theseforboundaries
the propagation
ensureofthe
waves outward.
continuation
of the pore water of both intergranular stresses and pore
pressures in the saturated soil. The prop-
{r} = (rzz,r_v,rzv)r = agation is based on the one-dimensional
wave equations of saturated soils where,

while, f = porosity, _ = total mass density, rzz = Ec \ Oz + he Ot Oz

gI = fluid mass density, a = compressibili-

ty =°fsolid,viscosityM
fluid = compressibilityof
and h he=soil permeabil-
t fluid, +a2M-'_Ou, + aMOWZ
0z
ity.
Thetotal
es the resultant
stress equation
vector in that
termsexpress-
of the P/= _-
_aMOU® M_Owz (6)
Oz Oz
 D

while shear transmission is governed by the A = [Cg,Wg, So] T (9)

principle:
r=v = _,V,6, (7) the synthetic time history 9(t) generated
where ti, = velocity in the direction of from tlle form
shear and Vs = shear wave velocity of the
soil. N
g(t) = 2_ (t) E V/-S= (_i)A_c°s(cvi_ + _bi)
Harmonic Solution i=1
The dynamic input can be either de- (10)
terministic or stochastic. In the deter- is a realization of the process described by
afinistic mode the given ground excitations the power spectrum. In order to accommo-
or dynamic loads are expressed in the fol- date the non-stationary part of the ground
excitation, the simulated acceleration or
lowing form force is multiplied by the nonstationary
N function _(t). In the above expression
g(t)= Z x Wu
/==1 w i = i A _v A _,,=
N
where _k are the frequencies of the har-
monic analysis.In thestochasticmode the where _u isa cutofffrequencyand _bi is
base excitationor the forcingfunctionis a vectorofrandom phase anglesuniformly
expressedin terms ofthe complex input distributedbetween 0 and 27r. Differen-
t choicesof the vector of random phase
X (¢#k) = eiWht angles will lead to a different simulated dy-
namic inputs. The synthetic process g(t) is
Evaluation of the response at the frequen- periodic with a period To a- _-_.
2_
cies _,'l=leads to the complex frequency re-
sponse H(=,). The Fourier coeffcients of
a responsequantityY(w) (displacementnr 3.0 Evaluation of Failure Potentials
stress)from a syntheticinput with coeffi-
cientsX(a,) can be obtained through the Slope Failure
relation The stability of the dam isviewed in
terms of a safetyfactoralong any poten-
Y (¢,,)= H (t#)X (w) (8) tim failure surface as shown in Figure 1.
The margin of safety against slope sliding
In probabilisticanalysisthat isbased can typicallybe seen as the ratioof the
on a Monte Carlo scheme (multipleevalu- shear strengthat a given effective stress
ationof the responseat the same intensity to the correspondingshearstrengthon the
levelbut with random distributionofpeak- envelopeline.While debate surroundsthe
s) the simulated earthquakesor dynamic definition of an appropriatesafetyfactor,
loadsbelong to a certainfamily.Such faro- in thisanalysissafetyfactorisdefinedin
ily can be chmracterizedby a response terms of the stateof stressat any instant
or a power spectrum. Earthquake (or duringthe seismiceventthrough thestress
load) recordscan be synthesizedon the invariants oftheintergranular stresses.
basis of these properties. For the case Sincefailureisexpectedto occur over
of an earthquake familydescribedby the an entireplane,the safetyfactoralongany
Kanai.Tajlml power spectrum such potentialsurfaceisdefinedas a con-
tributionfrom allthepointstransversedby

1 +4¢ ;;,
_ the surface as follows,

(.),],
()'
a

where Ai = area of the finite element trans- Liquefaction potential

versed by failure surface and (SF)i = safe-
ty factor for element i. The safety factor
for an individual element is formed on the
basis of the intergranular
!
stress invariants
J1 and J2 and the Mohr-Coulomb
envelope where,
Two methods both btdlt into the pro-
gram can be utilized in assessing liquefac-
tion susceptibility. The first has been intro-
failure duced by Seed & Idriss and is called Sim-
plifled Liquefaction Procedure. it rep-
resents the classical method used in evalu-

_a J1 + _f_ = k sting approach

ond liquefaction susceptibility.
is a Probabilistic The see-
method
that reflects the statistics of the excita-
J1 = o'z + _ry + crz tion and it estimates probability of failure.

, [(trz - _r_)2 + (o'y - crz)2 + (cr_ - trz) 2] observations

In applying either
must bemethod
considered.
the following
Lique-
J2 = 6 faction potential has been assessed to be
greatly affected by certain parameters par-
+ .r-2
zv (12) ticipating during the dynamic event. These
3C tartqb include the soil type, the initial effective
k = a = stress and the nature of the excitation (i.e.
V/9 + 12tart2qb V/9 + 12tart2_b earthquake type). Consequently, the soil
such that type with inherent damping and the inten-
sity and type of the input determine the
SFi - k + ctJ1 (13) intensity and the cyclic profile of the dy-
namic shear stresses.
a. soll type
Three parameters are expected to greatly It has been observed that sand deposits
influence the safety factor defined above, are more susceptible to liquefaction than
namely the intensity of the ground steeler- deposits of silt, clay, gravel or coarse sands.
ation as well as its spectral characteristics, This is attributed to the uniform (gradin_
the frictional angle _ and the hysteretic of the sand as compared to the other soil
damping _,fthe soil. types. The susceptibility of the soil to
Because of the linearity of the consti- liquefaction also depends on its void ratio
tutive equations, the dynamic stresses that or relative densit_l. The looser the sand the
will result from a scaled-up earthquake will higher the potential for liquefaction for a
be subject to similar increase, except for given earthquake.
the effects of the hysteresis. The amplifica-
tion of the fluctuation of the intergranular b. initial effective stress
stresses and pore pressures will, at vail- It has been observed in laboratory test-
ous times of the seismic duration, bring the s that there is a direct link between the
stress state of the points closer to the fai!- stress required to initiate Liquefaction un-
ure surface causing significant reduction of der cyclic load and the effective stress at
the safety factor. This issue is addressed the location of interest. Specifically, the
with the evaluation of the safety factor over required driving stress (or soil strength)
a chosen failure surface subject to incre- increases with increased effective stress.
mental changes of the peak acceleration
of the same earthquake. Figure 2 depict- c. dynamle input properties
s (a) a typical acceleration record acting The vulnerability of the soil to liquefaction
as base excitation to the dam section, (b) is determined by the level of stress or strain
the temporal variation of the dynamic s- that develops during a dynamic event. For
hear stress at a particular location induced a shear stress failure the developed stress
by the input excitation and (c) the tern- in the soil must exceed the soil strength
poral variation of the safety factor along a while for a strain driven failure the thresh-
selected failure surface, old strain must be exceeded. The shear
 6

stressesor strainsthat developin the soil

are directlyrelatedto the intensityof the
earthquake. In laboratoryteststhe soil,
subjectedto a cyclicload of a giveninten-
sity, fails after a number of such loading
cycles. In the field and under the action of
an earthquake,theeventmust have enough
durationsuch that a significant number of
loading cyclescan develop. These cycles
are not uniform, but they can effectively
be representedby an equivalentuniform
cyclerecord,
LQF = Strength at Nequlv (15)
0.65 rmaz
for the shear mode. For complex stress s-
rate,however, where the impigingseismic
waves are not only shear waves itisaug-
earedthatsimilarstrengthcurvesbe devel-
oped (from laboratorytests)reflecting the
strengthof the soilas a functionof more
than just the shear stress. This can be !

Simplified Liquefaction Procedure achievedthrough thestressinvariantVr_.

The liquefaction
safetyfactorcan then be
This Seed & Idrissprocessisoutlined equivalently
definedas
in the followingsteps:
8o Definitionof the earthquake that is j_//'_
likelyto occur at/orin the vicinityof LQF = v "soiltestlng
the siteand applicationas base exci- j./_f
ration. The excitationin the seismic V '_2seismically induced
liquefactionprocessisconsideredto be
verticallypropagatingshear waves (s- For eitherapproach the index LQF will
hear stressesgovernthe process), determinethe liquefaction
susceptibility
of
b. Calculationof the time historyrecord thesoilat the _ivenlocation(LQF > 1.0:
of the intergrannularshearstressesat no liquefaction).
selectedlocations.
c. Transformation of the time historyof 4.0 Stochastic Assessment of Failure
the shear stressinto an equivalent
record of uniform intensity.This is POROSLAM Probabillstlc Analysis.
based on strengthdata fortheparticu- Monte Carlo Simulations
larsoilsimilarto the ones in Figure 3
and achievedwith the relation The processbuiltin the program can
apply to assessmentsofeitherslopefailure

Neq : NiN (14) tional limit state

or liquefaction. Foriseither
defined as
optiona condl-
i=1 _r_q av
FSF -- ( SF) - 1.0 - 0
where,
Neq = Equivalent Uniform Cycles or
Ni - Induced Shear stress cycles at level FLQF = (LQF) - 1.0 : 0 (16)

ao where,
N_ q : Equivalentuniform cyclescausing
SF --SafetyFactoralonga failure
surface
liquefaction at rl
Nat,- Stresscyclesinducingliquefaction at time t
at 7"at,= 0.657"mar LQF = Liquefaction Potential at time t

d. Evaluation of the soil strength (_o)eq Further, the conditional limit state proba-
from the equivalent uniform cycles, bilities FSF;LQF are the probabilities that
e. Computation of the factor of safety a- at a given intensity of the input ai (which
gainst liquefaction according to the re- is included in the power spectrum of the
lation process and specifically in S0) the factor of
L i

tt

safety against slope or liquefaction failure POROSLAM Probabilistic Analysis.

is less than unity or equivalently Stochastic Approach

FSF;LQF < 0.0 (17) Consider the case when the input forc-
ing function (earthquake or load) belongs
with slope or liquefaction failure occurring to a weakly stationary process a(t) of dura-
when FSF;LQF < 0. tion T, zero mean and autocorrelation and
The statistical simulation process proceeds power spectra/density functions given by
as follows: the relations,

• Selects the range of the various inten- E[a(t)] = 0

sity levels of the input which in turn e-
valuate the power spectrum of the pro-
cess and consequently the time history E[a(t + r)a(t)] = R_a (r)
and the Fourier coefficients of the input
forcing function g(t). The response of 1 [,o -iW_dr
the system is evaluated via Equation q'a = _ d- OO Ran (r)e (19)
(¢).
• For the same intensity value but for M /_o
vectors of the phase angle 4'i (random- Ran (r) = j_ OO q'a (w)ei'_Tdw
ly selected) the process of the first step
is repeated. For each synthetic history The synthesized time history
at the intensity level the temporal vari-
ation of the Safety Factor or the Lique- N
faction Potential is evaluated. Figure
2 is a sample of a generated input base g (t) = 2 _ V/'I'a (wi) Awcos (wit + ¢i)
motion, shear stress and safety factor i=1
variation. The criteria for failure dur-

ing
safetyanyfactors
such simulated
drop belowevent
1.0 are
or /_ / _ and
can be has
assumed
both the
to represent
mean and a(t)as
the autocor-
N
the potential to liquefaction is greater relation of the stochastic process a(t).
than 1.0. The potential to liquefac-
tion is evaluated based on the resulting With Complex Unit Response Function of
dynamic shear stresses at a particular the multi-degree of freedom system H(w)
location (Figure 2). available for either unit base excitation or
unit harmonic load, the following powerful
• The probability of failure at any given relation holds,
intensity cti is computed in the form

P [FsF;LQF < OI]PGA = a] = Mf,_,

11,I
(is)
where, M/',it are the number of
simulated earthquakes at PGA = a
that induced failure and M is the total
number of simulations,
Ir should be noted that the larger the
number of simulations M at each level
the better the probabillstic assessment,
t Repeating the process for the select-
ed intensity range (cti ..... ctk) a fragility
curve is developed reflecting the sus-
ceptibility
of failure. of the system to either mode
#out (w) = H (w) Oi.t, (w) H "T (w) (20)
where Oinpt and Oout are the Power
Spectral Densities of the input and out-
put respectively. Provided that the excita-
tion is stationary, the response is also sta-
tionary and all the mathematical relations
pertaining to a stochastic process, which
satisfies that criterion, hold. These statis-
tical properties of the response proce:s are
listed below
tr_ = q,y (w) dw
 8

k_i.) =
/0 wi@y(w)dw, i= 1,2,..
(21)
6. References
1. POROSLAM. Two.Dimensional Dy.
namic Solution of Elajtic Saturated
By utilizing the statistical properties Porous Media, N. Simos, C.J. Costanti-
of Equation (21) the probability density no, C. Miller, Earthquake Research
function of the response quantity (say s- Center, City Univ. of New York.
hear stress at a location) can be evaluated. 2. Seismic Risk Assessment of Small
For temporal variations of the shear stress EarthdamJ, C.J. Costantino, N. Simos,
as viewed in a liquefaction assessment pro- Y.T. Gu, Technical Report NCEER-
cess, for example, a Rayleigh or Gausaian 91, Earthquake Research Center, City
probability density function p(r) can rap- University of New York.
resent the peak (or equivalently stress cy-
cle) distribution. Since the key to failure 3. Probabilistic Theory o Structural Dy.
is the number of induced stress cycles by namica, Y.K. Lin, Krieger, 1976.
the input one need to, in compliance with 4. Dynamics of Structures, R.W. Clough,
the statistical model, estimate the proba- J. Penzien, McGraw HiU,1975.
ble number of cycles. For a lightly damped 5. Liquefaction of Soils During Earth-
system (narrow-band) the ezpected equiva- quakes, Committee on Earthquake En-
lent cycles per unit time can be estimated gineering et al, National Academy
from Press, 1985.

6. A simplified Procedure for Evaluating

Neqv Weqv = (22) Soil Liquefaction Potential, H.B. Seed,
- 21r Weqt, _ -_y I.M.
1970. Idriss, Report No. EERC 70-9,

so the total number of cycles for the dura- 7. Analysis of Soil Liquefaction: Niiga.
tion T of the event are equivalently ta Earthquake, H.B. Seed, I.M. Idriss,
Journal of Soil Mechanics and Founda-
N - W,quT tions Division, pp 83-108, 1967.
27r 8. Evaluation of 5oil Liquefaction Porch-
and the number of cycles at a shear stress tial for Level Ground During
level 7"i Earthquakes, NUREG-0026, 1976.
9. Representation of Irregular Stress Time
w,_T
N (7"/) = 27r p (n) (23) Hiatoriea
Seriea in by Equivalent Analyses,
Liquefaction Uniform 5tresJ
Call-
For a particular soil medium for which fornia University, PB-252 635, 1975.
a relation between the number of stress 10. Cyclic Stress Conditions CauJing Liq-
cycles of a given intensity required to cause ue.faction of Sands, K.L. Lee, H.B.
failure exists, such as stress ratio vs. cycles Seed, Journal of Soil Mechanics and
in Figure 3, the failure potential can be Found. Die., No. SM1, pp.47-70, 1967.
viewed through the relation 11. Comparison o.f Dynamic Analyses .for
Saturated Sands, W.D.L. Finn, G.R.

N(r) dr (24) ASCE Geotech. Div., Vol.1 pp. 472-

Damage = _0°° Nreq (7") Martin,
491, 1978.M.K.W. Lee, Proc. of the

where N,eq(ri)is the mtmber of stress cy- 12. Compilation of Cyclic Triazial Lique-
cles that can lead to failure at the stress lee- [action Test Data, J.M. Ferrito, J.B.
el r. The above expression is similar to the Forest, G. Wu, Geotechnical Testing
well known Miner's Linear Failure criteria Journal, Vol 2, No.2, pp. 106-113,
and indicate failure when Damage > 1.0. 1979.
 9

13. The Generation and Dissipation of

Pore Water Pressures During Soil Liq-
uefaction, California University, PB-
252 648, 1975.

failure surface

bedrock
foundation layer

Figure I. A typical earthen dam cross

section

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United States
Government. Neither the United States Government nor any agency thereof, nor any of their
employees, makes any warranty, express or implied, or assumes any legal liability or responsi-
bility for the accuracy, completeness, or usefulness of any information, apparatus, product, or
process disclo._d, or represents that its use would not infringe privately owned rights. Refer-
ence herein to any specific commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recom-
mendation, or favoring by the United States Government or any agency thereof. The views
and opinions of authors expressed herein do not necessarily state or reflect those of the
United States Government or any agency thereof.
 I

10

".,. I , lllil l iillii I !

.. , ri,li
II,I}',..
iL11111111_-
i iiiiii!-,:i i .....
.....

1,5 "I_• .... _'" "',<1 i,l, :'%:L;_.IUII

" ';_ii I_ .....
1.4 ] : ',;1"1"1-I,."
_idl 1IT'btL I. " ....
" ,.3 _1....
_i . lllll
lll-'_._ "t>-,I,.l__lk!
I l'_"lll _'-I.:._. iiiii
: JI1H
11 I "' " t1"-----' '--'_1t4..0i_-.. ,rli.t4
II 1. I III11 I I 11IIIII "'_._,

!
II l . i , _11111I
illll I Ilillllll
II11111 Ii .,.iJii
ilIN
o.o :......:, i lllll I I l llitll ! lllti
;_.1111I _1Ii11111 I .....
0 2 4 6 8 10 II 14 16 18 20 ...,,_.

1000
O00-

_ 400-
soo- Figure 3. Typical strength data for

-
._ -200 01 sands
ii -4,00., I
-000-_
-TPO- 1
-tO001 ', i"
ll s,' lo , i
o _ i _,, 1, i'll ill Io

O.iO/ .....
0.161

o.oo-1--w,_'1
:_ -o.o54
°'°sl _.,.,,.ll
'
°"°1
-0.10
4
__
-0.151
-0.20_ ' I
o _ ; 5 "i5 IO
'! l
1, l,, 1'8Ill lo
'rL_; (..,.)

Figure i. Base excitation, induced

shear stress and temporal variation of
factor of safety along a failure surface
=rl