EARTHQUAKE-INDUCED PERMANENT DEFORMATIONS:

PROBABILISTIC A P P R O A C H
By M. K. Yegian,1 E. A. Marciano, 2 Members, ASCE,
and V. G. Ghahraman, 3 Associate Member, ASCE

ABSTRACT: A simple procedure for estimating earthquake-induced permanent de-

formations of earth dams, embankments and slopes is presented. The analytical
model is based on computed permanent deformations obtained by using New-
mark's sliding-block analysis and actual recorded acceleration time histories. The
model incorporates the effect of earthquake magnitude through the use of equiv-
alent uniform cycles of motion. Uncertainties in the model and the parameters used
are accounted for through the use of modeling error theory. Normalized plots and
a computer program are presented that provide the probability that the permanent
deformation of a critical sliding mass will exceed a specified value. The results
from this probabilistic procedure can be expressed in terms of damage probabilities
in the form of a seismic performance analysis matrix. The use of a seismic per-
formance analysis matrix in an overall seismic risk analysis for an earth dam,
embankment or a slopes is described in a companion paper.

INTRODUCTION

Earthquake-induced permanent deformations of embankments and earth

dams can be estimated by a number of approaches of varying degrees of
sophistication. On one hand, empirical procedures have been developed to
estimate permanent deformations (Ambraseys and Menu 1988; Constantinou
and Gazetas 1984; Franklin and Chang 1977; Lin and Whitman 1986; Mak-
disi and Seed 1978; Newmark 1965; Sarma 1975). On the other hand, two-
dimensional finite element methods have been used to evaluate the strain
potentials and permanent deformations within a dam subjected to seismic
excitations (Chaney 1979; Elgamal et al. 1987; Lee 1974; Paskalov 1984;
Prevost et al. 1985; Seed et al. 1975; Serff et al. 1976; Taniguchi et al.
1983). Regardless of the method used, there is always uncertainty in the
calculated permanent deformation of an earth dam due to uncertainty in the
input parameters and the analysis procedures. In a seismic risk analysis that
provides likelihoods of damage or failure of an earth dam, the uncertainty
in the predicted response of the dam needs to be considered properly.
In the companion paper (Yegian et al. 1991), the writers describe the use
of probabilistic estimates of permanent deformations in the calculation of the
overall seismic risk. This paper describes a procedure for calculating earth-
quake-induced permanent deformations of earth dams and slopes and their
likelihood of exceeding. The analytical model developed satisfies the fol-
lowing criteria.
'Prof., Dept. of Civ. Engrg., 420 Snell Engrg. Ctr., Northeastern Univ., Boston,
MA 02115.
2
Asst. Prof., Dept. of Civ. Engrg., Northeastern Univ., Boston, MA.
3
Grad. Student, Dept. of Civ. Engrg., Northeastern Univ., Boston, MA.
Note. Discussion open until June 1, 1991. Separate discussions should be sub-
mitted for the individual papers in this symposium. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on March
19, 1990. This paper is part of the Journal of Geotechnical Engineering, Vol. 117,
No. 1, January, 1991. ©ASCE, ISSN 0733-9410/91/0001-0035/$1.00 + $.15 per
page. Paper No. 25414.

35
 1. The seismological and geotechnical parameters used in the model are con-
sistent with output parameters of the seismic hazard analysis described in the
companion paper. This allows the incorporation of the model in the seismic risk
analysis described in the companion paper.
2. The model parameters can be evaluated using simple or sophisticated pro-
cedures as deemed necessary.
3. The permanent deformation model is simple enough to be conveniently
applied in design practice, yet it accounts for the pertinent seismic and material
parameters.
4. The model allows for the application of probability theory to estimate the
likelihood that the permanent deformation will exceed specified values.

PROCEDURE FOR CALCULATING PERMANENT DEFORMATIONS

Calculation of earthquake-induced permanent deformation can be made

using Newmark's sliding block model shown in Fig. 1. In this approach, a
rigid-plastic response is assumed, such that every time the acceleration of
the block representing a section of an earth dam exceeds a limiting yield
level, Ky, then a relative displacement is initiated. The total relative dis-
placement, Dr, represents the permanent deformation of the dam section.
An illustration of the motions of the block and the base for a triangular base
excitation is presented in Fig. 1. The derived mathematical formulations for
Dr, considering triangular, sinusoidal, and rectangular base motions, can be
found in Yegian et al. (1988). Based on these derivations for Dr, the fol-
lowing observations are made.
The expression for Dr is of the form

base acceleration, K a
,1

* • - time

) D r , permanent
displacement

yield

FIG. 1. Sliding Block Analysis for Triangular Base Motion

36
 RECTANGULAR

i i 11 i r r i i | i i i i i 1 i 1 1 | i i i i i i i i1 | i i i i i i i n i r r m r i t r
0.0 0.2 0.4 0.6 0.8 1.0
K y /K Q

FIG. 2. Normalized Permanent Deformation Functions for Simple Base Motions

Dr=f\^-)NcqKar (1)

in which / ( ) = a function that depends on the type of base motion consid-

ered; Ky = the yield acceleration; Ka = the peak acceleration of the base; T
= the period of the base motion; and Ncq = the number of equivalent uniform
cycles of base motion. The permanent deformation, Dr, can be normalized
with respect to the peak acceleration of the base, Ka, the number of equiv-
alent uniform cycles, 7Veq, and the square of the period, T, of the base mo-
tion, giving

D„ = (2)
KaNJT7- \K,
where D„, referred to as the normalized permanent deformation, is a function
of only Ky, Ka, and the type of base motion.
Fig. 2 shows plots of D„ versus Ky/Ka for each of the three simple base
motions considered (i.e., triangular, sinusoidal, and rectangular). It is clear
that the shape of the base motion has an important effect on the permanent
deformation, especially if Ky/Ka is close to 1.0. Recognizing that none of
these periodic motions properly simulates the random nature of earthquake-
induced ground motions, a more realistic determination of the function/( )
of Eq. 2 was made by considering actual earthquake records. Franklin and
Chang (1977) have published values of permanent deformations computed
with Newmark's sliding-block analysis using actual recorded acceleration

37
 o o o -.1 -o. </l trt t / l t/l ^ ir
Ut IO -~l ON Ul UJ 00 OO ON 00 Ui to Ui to NO CO - j ON OO ^1 Ui NO Ut CO Ul NO NO CO OO Ul Ui
to >— ^ 1 4^ -P- NO - J 4*. 4*-
(*> t4*.o A 4^
u» num ber
Caltec h file

to w J> H_1 1_1 (_1 i—• to t>J 7>

on - 4 00 on t/l W OO t*J (*J -P- t/» *4 4*. to
H_t l_1 tt/_
l
k1O—N t/l Ul
UJ U l NO -P- NO O 00 o
- 4 00 UJ UJ ON NO NO -P* OO 00 ON u> t o ON u» - 4 00 OJ NO - o : <i to NO Ul to Ul - 4 Ul 4^ to - 4 CO to Ul •—' .iy ^j»
o o o o o o o
Ul Ul o 4*. ON 00 Ul UJ ON h-* ui ON N) u> Lft 00 to - 0 O 00 U) to OO 4> © -J ON -4 *-
ON U l ON ^4 00 UJ -4 Ui ^1 CO ON - 4 NO - 4
w •-* *- o W ^ o CO •"*
i—' K) 1—I <*} ,__, ,_, t o to to to l_i H ^ to UJ
00 w
00 to )-_
4l - J OO H_l v_1 |_1 ,_,
4*- Ul 4^. i—' 00 ON Ui
o UJ - J 00 ON W too OO |-_
J
l N O h-
ON
l_1 H_1
NO -o. ON -J l_t
- 4 © N)
h _to
t — 1U l
u» 00 ^J •p> ON Ul I—1 NO 1—'
lN_
Ol U l

to |—L •-J tU ON I*) -p. 1/1 (*J w 00 t/l I>J -P- NO l/i no •P* to 1—' NO ON 00 to ^J -o NO 4^ 2 3 ^
88 O © 8 O © O O 8 O 8 © © © 8 © © © 8 © © © © 8 © © © © © © ©
o o o o o o o o o o o o o o o o o o
_ t— ^ H- h- ,_, i-t ,_, ,_! ^ t— !_, _ _ w
— UJ CO NO 00 UJ UJ NO Ul - 4 ( J l U l A NO
-p* s> t o U _ ,__,•—> 4^ M
UJ 4^ ON •-J CO U) CO -p* i—' t—* h_i 4^ U l Ut - 4 O UJ Ul OO to NO 4>. t o -p* 4*. Ui ON NO NO O
>—' o >—>
OO (O to --1 4i. <«J Of)
{,/! no NO t/l -4 t/l no to J> I—. 00 ~4 NJ OO NO
o o <o o l_l © © O © ©
S ii=
O O © © O © © © © O
o o o o o o o o o o o o o o o o o o o o o o o o o o o O o o ©!_> © o o o o
o o o o o o O o o o o o o o o o o o O o o o o o o O © o o o O © o o o o © © o O © O © O o © O © © O ©
i d 4^ to l_ <«J tfl !*J w (ij |_L <ti •p> t»J to to to 'O on -p. t/i to t/i to Ul CO •H
Ul Ul Ui Ui Ui Ui )—•
Ul Ul fj* U\ Ui Ul Ul Ul Ul © Ul Ui O © © Ul © © Ul Ui © Ui
o o o o o o o o o o Ui o Ui U i U i U l Ut LA L f t O o o o
I— t-* t— h-
OO Ji toh— O N OO OO H-
NO
,__,
to 00 ON K) to 4*.
l_1 U J Ul 00 Ln OO Lft NO
!_ OJ Ui
_
|k ^ ON 4^ OO to
l_t O J 00 J> ON - 4 to to - 4 4^ to ON
^_, A Ul to - 4 to _
l i ~4 ,_
I—* to - 4
w —' w -o o
1—k IO !
to - J to o o to t o i—• to 4> t/l K) U\ NO
•4^. l/l O t/l 4^ 1— W t/l 00 •p. OO t/l © Ul NO -4 - 4 -o to t/l NJ t/l ON -4 to
-J O U> UJ NO ON 2 OO ON 4>- 4 ^ -p* ON -P* t o -J -J NO to 00 (Jl UJ ON O ON Ul 4^ © NO 4>- - 4 NO to 4^ -p» OO © UJ 00 Ul Ul -P* to •-J © s a3: (
o -^ w o o u>
Q\ O N O N ON ON ON ON ON ON ON ON ON ON ON ON ON ON O N ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON Ul Ul Ul - 4 ON ON U i Ul ON ON ON ON - 4 ^4 •-4 - 4 ON ON
ON ON O N ON ON o\ ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON OS O N ON ON ON ON ON ON ON. ON ON ON UJ ON UJ U i Ul U I Ul - 4 - 4 - 4 - 4 -4 -4 3 s;
^ O
IO tO IO W 1/1 t*J Ji- 4i. w to •p' 4^- tO to ,_, •PK ^
ON N O OO NO ON DO to Ul to to O NO to NO OO to to to ON 1—' UJ UJ NO NO
NO NO to £ A NO O IO •^J 4^
,_! ON - 4 © 4*. NO NO
o o -o -o; o o m 1
OO O N O to IO UJ Ul O0 - J to 4* -o. O 4i. O NO Ln h U\ o 00 00 w ON NO NO 00 4^ © h- 4^. 00 ON 00 o 00 -p* 4^ © **
o Ui Ui O o UJ UJ 1*
~ ~ ^ - - ^
O O to _ _ NJ o o o o —_. O to - J to >- O o w h- o h- O OJ o to ^ © - o o Q © H- O © O -p* © © o -P* k- © to o © © © ,_*
ta- to 00 to t/l 4^ Ul to tfl W Irt t/l tA t»J CO -.1 trf W to t/l l_> (O to l_l to to O
j '^- ^
to -4 4^ o NO 41. NO t/i NO Ki 4>. -Pw ON © to ©
0\ NO - 4 U» to t*J Ji. 00 ON 00 t*J OO © t/l -o. t/i to
<o tO l/l to NO no too U t/i 00 to o
ON UI 4i. ON - J •1^ 4> t-ft K O
oN no Ui 00 NO -^t A t/l -4 NO **A t o t/l ON rn b ^
IO -t- NO ON NO 4^ 00 O t/i to 00 00 NO to NO 00 ON NO ON ON -J ON to to Ui NO ON ^.1 00 ON NO NO ON -Pv NO
o
t/l NO -p* 4*-
vO — to 4^. UJ NO -4 to 00 00 Ul ON 4^- 4i- - j to L/l ON
K (*> ON ON © S> - 4 OJ -o
00 to o
to UJ 4^ Ui NO NO Ul 4^ -4 - 4 -p* UJ UJ
f
— w *"" l-' II
h-
maliz

O O O O O O O O © © O © © © © © ©
Defor

O O o o O o o © o o o o o o o o o © o o o © o o O © o © © o © o
w v *
H_ (J H- i—> to iti ta^ UJ (*J t— K-> © t/l H— to 1—I H^ © 1_->
© 4*. t o i_. Of)
t/i •-4 00 t/i W ^A
o 00 to tA OJ to (J A 00 OO t>J Ul 00
o 0o0
-p* NO (*) 0^ © © 4* -P* © -p-
t/l <*J
o
t/l 4^. Ul - J NO t o !*J -o
t/l - J
o
t*J t/l o
to - 4 4^ 4^- no OO i->
—- 2 ^ 1 °-
1/1 -.1 - 0 O -^1 -O. C/l -p* on -P- to NO 4> tp to Nt t*J ^4 ^4 Ul t/i - 4 - 4 ^4 l/l 4^ NO Ul
ON ON Ui Ul to NO to - 4 00 00 £ Ui ON
o
- 4
-o t o
N O OO OO Ul U) •-J o n - J Ui NO OO Ut © ON
o
CO OO - 4 ON © UJ ON ON to •p. - 4
—
' -o o n
tion, D„

O O © o o o o o o O o o o o O O O o o O o O O O o © o © © o o o O o © © © o © © © o © © O © © © o © ©
Perman

O O O © © O © © © © O
^ (D
o o o O
o o O tu
o o o o o o ©
o
© ©
o o o
©
o o
88 8 8 © 8 ^, © —*
? - 4
o -o. 1 too o ON
to 00 o o o 8 o -.1 o o lt ioj
4^. Io
W 8 00 o
o O 4^. 8 © J> 8 o too Ui
*>- o o
t/l
o
00
IO 00 ^4 00
co w o
N> 4^. U> NO C/l A (/I to - 4 00 NO O ON 4> - 4 Ul ON no NO Ul
UJ -J *•• UJ - 4 00 N» NO h •P* t o 00 OO UJ
w
ON £ W ^1 ON J> 00 O M t-n •-4 U> O J NO
-o
U l ^1 ui © Ul NO 00 8 NO 00 (NJ - 4 to UJ UJ O ON ^4 Ul UJ
II
"'
 C
OQ S- H z UJto to to to to to to to to to to to
w 1^1 \A tA UJ <h> <>J UJ UJ 00 CO 00 00
B5 El o L/l UJ ON UJ UJ to
UJ CO ON U'I
4^ UJ
to -J -J ON On to to ON A -J to oo -J ON ON 00
B a n -I ^o NO
W A S o
° 8-g. ,^ h- ^ h- ^^ ^w ^ ^ ^ |_kto l__k
o +*• 00 f A (W NO 4^ Ut 00 •£> -J
to h-
on NO On UJ 00 -J UJ to 00 L/i NO 00 ON NO -4 -4 UJ NO un00
o ON 00 45*. -J UJ -J to NO .IV
u^ 4^
3 O on on O o ON w -0 O O OO -0 O NO JS. -0 -J UJ 4^ ON O NO NO UJ
o o o o o o O O o o o
1° "0
t—' >—> .. >_. i_* k-* 1—4 h-> UJ h-* w _
m I-- UJ
)_l t
00 NO 00 NO
1_1 UJ
1_L UJ
ro 1_1
Ut t—'
w ^4 on oo N> 00 ON NO 4^ O NO -J -J 4>
-~o l_i l-_Jl|_lw-J NO 4^ o
30 t/i t—' I—. 00
!»J Ut t*J to 4^ oo Ut
3 O 3 to -J -o •—'4ON*- ON
ON --1 Ut O -O Ut NO o i* un -fc.
o 5 O O s
^3 > o oo o o o o o o o o o o o o o O o o ON o O o o o o o O o o o
n £" h-* •_> H-
h- NO 00 H -
,__,,
to UJ H- l
__,,__, to_k ,ON
_,,_,NO oo
u> NO NO NO t — ' NO ON NO 00
_.to on NO -J -J o Ui to UJ
__
t-n tl— ' t -4 00 to
o •— CA UJ 4^
*-*4> K>00 00 -J ^-L
4^ UJ NO
NO
o
NO 00 --1 4^ p_> Ui NO a
O ON 00
o l-i o ,
_>-0 s
L f
so
-&.o o o on o o o o NO NO o o NO UJ ON o 00 to -J 4> 00 o o o o o o O o O
m
i—' oo -n o oo o o o o o o o O o o o o o o o o o o o O O H- o o o o H- O o o o O
SO O I—. (*i t J <>J «—I 1—1 C*J h_i
-t* t*J (a ro to UJ to 4i- t*J -F^ to UI m
oo tJj 3D Vi O o On on on On On t7) t/l LA o o U\ U\ o o o o o Ui Lft Ut Ut Ut Ut
S
o o o o un o o O Ui
_J h— h-. 1—> h-
> to 00 I-* 00 Ut •- H- k-
,__,,_,^ w W ~o to UJ o -fc- ON
,_,oo ,_,l_kUJ •1^ O N NO 00 L/1 to ON UJ
>-*,_ l_t _ ONUJ CO U*i NO •vj
o
(9)

u\ to 1—. -J I—1 Ut --1 1—> 00 i—> NO K->

K> S ON 4*. CJ1 UJ 4^ *—'oo
-J UJ On O s NO On -J NO U\ ON -4 -J ^ to UJ UJ
^^ UJ ON ON -t>
o -J O tj) 4^ o to -o A to
^ ^ ,_-
ai as ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON ON OS ON ON ON ON ON ON ON ON ON ON ON O N ON ON ON ON

oo o\ ON ON ON ON 0\ ON ON ON ON ON ON ON ON ON ON ON ON ON ON O N ON ON ON ON ON ON ON ON ON O N ON ON ON 3
£, 5
UJ On l/i UJ !*J UJ UJ w 4 ^ 4^ to (*J t»J <»J 4^ -b- 4^ UJ -.1 Ut •-1 UJ
UJ UJ to to
O to N) O NO NO so SO 00 to O) 00 00 •fi- NO NO LA UJ 4^ 00 to A Ln to O NO 4^ UJ O N
o o m
3 S» ON O O O NO as SD OO SO -o
<—* oo oo o o NO w UJ O UJ o O -J NO K- o> NO 00 NO NO UJ ON
o o o ~
K- -JO K- 4*. K- to •- V- O H- *-*. h- to — h- N- o _ o o ^ O _i o h- o o o O .- O O
£US O >- UJ N>
^
as C/l !*J W l_k --1 4^ to -J i—' t*J ON Ut ON (*J l_1 ON -4 Ji. M
On -J tAi f*J UJ o o Irt --1 00 --1 00 Ut
o -J A --I t/l
UJ -J too tlJ NO o (»J UJ -~A 4*. UJ ~-I !*J
<\ m
fJ-> t^I t/» UJ A (/I UJ
4± !*J
--I ~-l (A
UJ UJ
00 ON UJ U) OO or !*J
UJ -o -J
Ut ON
4^ * on on oo ON 4 ^ 4^ -o 4^ UJ Os to NO UJ--) UJ O NO NO 4^. CO U*i U^ 00 CO
& •*.
*-*
o H- o O H- o H- o o o O O o h- o o o O o o o o O O O o O o o o o o o o O
t*> (>J ^-L I*J t*J !*J l*J !*J to h- H— UJ O
O VO o Ut 4^ to to K - to H > o b-> to ON o to
c oo to -.1 oo '*> 1/1 4^ 00 oo Ui . NO l*J
l/i (>j tA 00
o 00 t/i ON
o—' JA NO ON NO o
O 4^ OO
^D H- !*J A -o
trt
00 VO NO
!*J tA !W (A 00 00
4^- m
NO NO
00 O N oo NO CO
t*J t/l UJ UJ
to u»
ON
to on h- 00 O N to ON -1^ u> UJ Ut -J 4^ O NO 00 NO UJ 00 UJ 00 h- 1
^^> o <_/> -1 on Ut o -b. 00 O
O u> n
d - m © o o o O o o o o o O o o o O o O O o o O o O o o O O o o o o o o o O
Oq _ O o o o o O o o o o a O r> <Ti <r> n n r> o o o <r> o o o
o Q 00 00 O NO -.] -.1 00 4^ ON to !*J to --] un Ut (A OO NO
f £ CL Ut l/> to
on on Ifl tw o
1/1 -.1 Of) to -4 UJ ON ON OO \A o
UJ NO on -4 o
00 Ut
Sr 3' 00 O -O 4*- O 00 U ) 00 i-rt 00 00 NO Ut ON 00 Ut NO UJ UJ 00 OO
g P OS -J to o 4*. to -J O O 4^ 4^
0> Oq -
 10 -3

RECTANGULAR -

FIG. 3. Normalized Permanent Deformations from 86 Actual Ground Motion Rec-

ords and Developed Function

median values of the plotted data points at Ky/Ka = 0.02, 0.1, and 0.5. The
advantage of using the median as opposed to the mean values is that the
effect of the few extreme points at the high and low ends, which are orders
of magnitude different than the next highest values, is de-emphasized. In
addition, D„ was assigned a very small value, 10~5, at Ky/Ka = 1 . 0 because
the deformation based on rigid-plastic model should be zero if Ky/Ka = 1.0.
This value was selected to obtain for f(Ky/Ka), in the range of 0.5 < Ky/
Ka < 1.0 where data are lacking, a shape consistent with that of the trian-
gular base motion. As can be observed in Fig. 3, the triangular base motion
appears to be more representative of actual earthquake motion than the si-
nusoidal or the rectangular base motions. The resulting polynomial curve is
shown in Fig. 3, and its mathematical expression is:

logD„
m. = 8 — I = 0 . 2 2 - 10.12 + 16.38( —

- -HI (3)

COMPARISON OF DEVELOPED FUNCTION WITH OTHER MODELS

In Fig. 3, the polynomial curve established for the calculation of seis-

mically induced permanent deformations is compared with the deformation
functions of the triangular, sinusoidal, and rectangular pulses. For Ky/Ka <
40
 z

<
s
DC
O

z
UJ
z
<
oc
UJ
CL

<
s
O
z

FIG. 4. Comparison of Developed Normalised Permanent Deformation Function

with Makdisi and Seed (1978) Data

0.1, these simple base motions yield significantly lower values of £>„ than
obtained based on the integration of recorded time histories. Conversely, for
Ky/Ka > 0.1, the sine and the rectangular base motions yield significantly
higher values of D„. For Ky/Ka > 0.1, the assumption of a triangular pulse
yields deformations that are generally in agreement with data from recorded
time histories and thus with the established function for permanent defor-
mations.
Fig. 4 shows values of D„ estimated based on the results of Makdisi and
Seed (1978) for magnitudes of 6.5, 7.5, and 8.25. These values of D„ were
obtained by normalizing the displacements computed by Makdisi and Seed
(1978) with respect to the peak acceleration of the potential sliding mass of
the dam, Ka, the square of the first mode fundamental period of the dam,
T, and the number of equivalent cycles of the ground motion, Ncq. The re-
sults obtained plot slightly below the developed function, probably because
the periods of the dams used for normalizing Makdisi and Seed (1978) data
are larger than the predominant periods of the computed motions of the dam
cross sections. Nevertheless, it is noted that data of Makdisi and Seed (1978)
plot within the range of data obtained from the integration of recorded time
histories.

UNCERTAINTY OF DEVELOPED FUNCTION, g(Ky/Ka)

Fig. 3 shows considerable scatter in the values of D„ computed using 86

seismic records for each of Ky/Ka = 0.02, 0.1, and 0.5. This observed scat-
41
 00] OOiOl 02 05 1 2 5 10 20 » 40 50 60 ?0 BO 90 95 9399 998 999 9999

PROBABILITY OF EXCEEDANCE

FIG. 5. Probability Plots of Normalized Permanent Deformations from 86 Actual

Ground Motion Records for K,/Ka = 0.02, 0.1, and 0.5

ter of the calculated values of D„ is primarily the result of the inherently

random and stochastic nature of seismic ground motions. That is to say, it
is not possible to completely define seismic ground motions, and thus their
effects, with a finite set of parameters. A similar conclusion was made by
Gazetas et al. (1981) when they normalized permanent displacements com-
puted from nine accelerograms recorded during the San Fernando 1971
earthquake. The nine records used were normalized to have common peak
acceleration and peak velocity. To account for the resulting uncertainty in
the predicted value of D„ given by the function of Eq. 3, statistical analysis
of the data shown in Fig. 3 was made. The log D„ values of all the data
were plotted on normal probability paper for Ky/Ka = 0.02, 0.1, and 0.5
and are shown in Fig. 5. The results show that for probability of exceeding
greater than 10%, D„ is approximately lognormally distributed as is indicated
by the nearly log-linear plots of the data on the normal probability paper.
These nearly log-linear plots resulted partly because the normalized displace-
ment, D„, was defined and computed as the product and quotient of several
random variables. The slope of each of the three normal probability plots
corresponding to the three values of Ky/Ka is about 0.45. Thus, the standard
deviation of log D„, (TiagD„, is 0.45. If the values of the parameters Ka, JVeq,
T, and Ky are known to be exactly equal to specific values, hereafter referred
to as ka u neq, t, and k^, then D„ can be simply treated as lognormal with
median d„, where log dn = g(ky/ka).
The lognormal distribution of D„ was used to calculate contours of prob-
abilities of D„ exceeding specified values d„. The resulting curves are shown
in Fig. 6. An example of the application of Fig. 6 follows. Considering the
following parameters: ka = 0.22g, ky = 0.07g, neq = 12 cycles, and t = 0.7
42
 K y /K G

FIG. 6. Probabilities of Exceeding D„ Considering Scatter in Data

s, what is the probability of Dr > 3 ft (0.91 m)? This corresponds to D„

exceeding the following value: D„ > 3/[0.22 X 32.2 X 12 X (0.7)2] =
0.072 for an acceleration ratio: ky/ka = 0.07g/0.22g = 0.32. Entering Fig.
6 with these two values yields a probability of about 10% of Dr exceeding
3 ft (0.91 m).
The aforementioned procedure considers the uncertainty in the deforma-
tion function introduced by the scatter in the data obtained from the inte-
gration of recorded time histories. However, prediction of permanent de-
formations using the developed function involves additional uncertainties
associated with the input parameters Ka, Ky, Neq, and T.
The uncertainty in Ka and N^, which are treated as ground motion param-
eters, can be taken into account through the seismic hazard analysis pro-
cedure described by the writers in the companion paper (Yegian et al. 1991).
The uncertainty in T and Ky must be considered in addition to the uncertainty
in the deformation function g(Ky/Ka), in order to make rational calculations
of the probability of permanent deformation exceeding a specified level. Us-
ing modeling error theory (Ang and Tang 1984), the uncertainty introduced
into Eq. 3 by the random nature of seismic ground motions can be repre-
sented by

logD„ = g(-A + 5(TlogD„ (4)

where S = the standard normal variate, which has a mean of zero and stan-
dard deviation of 1. Combining Eqs. 2 and 4 yields
43
 FIG. 7. Schematic Representation of Integrations Shown in Eq. 7

logD r = gl-A + SalogDn + 21og T + log (neqka) (5)

which defines a surface in a fictitious three-dimensional space above which

Dr exceeds a specified value dr, and below which Dr is less than dr. The
three-dimensional space is illustrated schematically in Fig. 7. The axes are
T, Ky/ka, and S. The procedure for determining the probability of Dr ex-
ceeding dr is to determine the probability that the values of T, Ky/ka, and
S will locate a point lying above the performance surface. To do this, a
suitable joint probability distribution is selected for T, Ky, and S, and then
it is triple-integrated within the region above the performance surface. The
region is defined by 0 < T < °o; 0 < ky < °°, and s^ =£ s < °°, where

log dr - log (Me,) - g\—) ~ 2 log t

(6)
"logD„

The integration is mathematically expressed as

P(Dr>dr\ka,ncq)= i l l fs(s)ds MW\Mk,)dk, (7)

Jk,=0 Ur=0 LJs
where P[Dr > dr\ka,ncq] = the probability of deformation Dr exceeding a
specified value dr given ka and neq; and fs(s), fT(f), fKy(ky) = the probability
44
density functions for the random variables, S, T, and Ky, respectively.
The integration was performed numerically in a computer program called
NIMPED developed for this purpose (Yegian et al. 1988). The variables T,
Ky, and S were assumed to be statistically independent and normally dis-
tributed. As the values of T and Ky are necessarily bounded by zero, their
distributions were truncated at zero and the probabilities of T and Ky equaling
zero were taken to be

P(T = 0) = <!>( ——J (8)

P(Ky = 0) = * P ^ J (9)
where \LT and u,^ = the means of T and Ky, respectively; and <& = the
cumulative distribution function of the standard normal variate.
The computer program was used to develop typical normalized plots that
provide the probabilities of exceeding of any specified deformations, dr. Figs.
8 and 9 present these plots for c.o.v. Ky of 0.5 and 0.2, respectively, and
c.o.v. T equal to l/2(c.o.v. Ky). Yegian et al. (1988) evaluated the rela-
tionship between c.o.v. T and c.o.v. Ky. The result suggests that, for both
cohesionless and cohesive soils, c.o.v. T can be considered to be on the
order of one-half of c.o.v. Ky for a flexible,dam on a rigid foundation.
The values of c.o.v. Ky that were used to generate the plots were chosen
based on values of coefficient of variations of shear strength parameters of
cohesionless and cohesive soils, which typically range between 10-15% and
30-50%, respectively (Harr 1977). For cases where there is a potential loss
of shear strength due to seismic excitation, c.o.v. Ky may be larger than 0.5.
In such cases, the computer program NIMPED can be used to get the prob-
abilities of exceeding.
To illustrate the use of this plot, the example presented earlier is consid-
ered again but this time with additional uncertainties in Ky and T. Assuming
that the coefficients of variation of Ky and T are 0.5 and 0.25, respectively,
Fig. 9 can be used to calculate the probability of permanent deformation
exceeding 3 ft (0.91 m), P(Dr > 3 ft).
The probability contour line, on which the point corresponding to Dn =
0.072 and ky/ka = 0.32 plots, defines the probability of deformation ex-
ceeding 3 ft (0.91 m). For the example problem, this probability is equal to
about 25%. Note that if the uncertainty in Ky and T were not considered,
the probability would be about 10%, as was demonstrated earlier and shown
by the plots in Fig. 6.
The plots of Figs. 6, 8, and 9 or the program NIMPED can be used to
determine the complementary cumulative distribution curve for Dr, condi-
tioned upon specified values of ka and neq. In Fig. 10 the curve corresponding
to c.o.v. Ky = 0.5 shows the resulting distribution for the example. This
curve can be used to determine the probabilities of the occurrence of dif-
ferent damage states. For example, if a permanent deformation greater than
3 ft (0.91 m) is considered catastrophic because of potential overtopping of
a dam, and if deformation of less than 1 ft (0.305 m) is considered incon-
sequential, then the following damage states with their corresponding prob-
abilities can be defined from Fig. 10 (c.o.v. Ky = 0.5): P(0) = P(minor or
no) = P[Dr < 1 ft (0.305 m)] = 0.57; P(H) = P(heavy) = P[l ft (0.305
45
 10 n

I
z
o
K
O
U.
UI
Q
I-
Z
UJ
z<
Ul
a.
Q
UJ
N
_J

<
2
a:
o

K y /K a

FIG. 8. Probability Contours of Normalized Permanent Deformation

0.0 0.2 0.4 0.6 0.8 1.0

Ky/K a

FIG. 9. Probability Contours of Normalized Permanent Deformation

46
 1.0 2.0 3.0 4.0 5.0

PERMANENT DEFORMATION, d r (ft)

FIG. 10. Complementary Cumulative Distributions of Permanent Deformation for

the Example (1 ft = 0.305 m)

m) > Dr < 3 ft (0.91 m)] = 0.18; and P(C) = P(catastrophic) = P[Dr >
3 ft (0.91 m)] = 0.25. The results of this probabilistic permanent defor-
mation analysis can be conveniently displayed in a damage probability ma-
trix. Table 2 shows a typical seismic performance analysis matrix involving
probability of exceedance of earthquake-induced permanent deformation. The
damage probabilities obtained from the aforementioned example are dis-
played as a single column in this matrix, for the peak ground acceleration,
A, in the range 0.15g < A < 0.2g, which corresponds to ka = 0.22g. In
order to perform an overall seismic risk analysis, these computations need

TABLE 2. Typical Damage Probability Matrix from Permanent Deformation Anal-

ysis
Number of Equivalent Cycles
Damage state 1-5 5-10 10-15 15-20 20-25
(D (2) (3) (4) (5) (6)
(a) O.Og =£ A < 0.15g
No/minor (O), < 1 ft
Heavy ( # ) , 1-3 ft
Catastrophic (C), > 3 ft
(b) 0.15g < A < 0.2g
No/minor ( 0 ) , < 1 ft 0.57
Heavy (H), 1-3 ft — — 0.18 —
Catastrophic (C), > 3 ft 0.25
(c) A a 0.20g
No/minor ( 0 ) , < 1 ft
Heavy (H), 1-3 ft — — —
-
Catastrophic (C), > 3 ft
Note: 1 ft = 0.305 m.

47
to be repeated for different combinations of «eq and A to fill in the entire
matrix. The use of the seismic performance analysis matrix in seismic risk
analysis for earth dams is described in the companion paper (Yegian et al.
1991).
The probabilistic procedure presented can also be used to evaluate the
effect of critical parameters and uncertainties upon the estimated probabili-
ties. For example, Fig. 10 presents conditional probabilities for different
levels of uncertainty in Ky and T. For the example considered, it is noted
that the uncertainties in Ky and T have little effect on damage probabilities
associated with less than 1 ft (0.305 m) of permanent deformation.

SUMMARY AND CONCLUSIONS

A simple procedure for calculating earthquake-induced permanent defor-

mation of earth dams and slopes is presented. The analytical model can pro-
vide estimates of the probability of the permanent deformation of a dam
exceeding a specified value given a specified seismic event characterized by
the average acceleration and the number of uniform cycles of motion of a
critical sliding mass. A computer program, NIMPED, is available that fa-
cilitates the probability calculations. Using this program, charts of exceeding
probabilities of permanent deformation are provided considering typical lev-
els of uncertainty in input parameters. These charts or the computer program
can be used to develop damage probability matrices for an earth dam. The
results from such a seismic performance analysis can be used in an overall
seismic risk analysis of an earth dam, as described in the companion paper
(Yegian et al. 1991). In addition, the probabilistic procedure presented can
help identify the important parameters and uncertainties that require special
considerations in the evaluation of permanent deformation.
The model developed for calculating permanent deformation is based on
the assumption that the soil deforms as a rigid-plastic mass. Also the pre-
dominant periods of most of the records used are less than 0.5 s. Improve-
ments in the model can be made by considering elastoplastic soil response
and earthquake records with larger periods associated with soil deposits.

ACKNOWLEDGMENTS

The research described in this report was sponsored by the National Sci-
ence Foundation through Grant No. DFR-84-12124 for research on Inte-
grated Seismic Risk Analysis for Earth Dams. The writers gratefully ac-
knowledge this support.

APPENDIX I. REFERENCES

Ambraseys, N. N., and Menu, J. M. (1988). "Earthquake-induced ground displace-

ments." Earthquake Engrg. and Struct. Dynamics, 16, 985-1006.
Ang, A., and Tang, W. H. (1984). Probability concepts in engineering planning
and design. Vol. II, John Wiley and Sons, New York, N.Y.
Asturias, R. W., and Dobry, R. (1982). "The equivalent number of cycles of re-
corded accelerograms for soil liquefaction studies." Report No. CE-82-5, Rens-
selaer Polytech. Inst., Troy, N.Y.
Chaney, R. C. (1979). "Earthquake induced deformations in earth dams." Proc.

48
 Second U.S. Nat. Conf. on Earthquake Engrg., Earthquake Engrg. Res. Inst.,
633-642.
Chang, F. K. (1978). "Catalogue of strong motion earthquake records, Volume I,
Western United States, 1933-1971." State-of-the-Art for Assessing Earthquake
Hazards in the United States, Rept. 9, Misc. Paper No. S-73-1, U.S. Army Engr.
Waterways Experiment Station, Vicksburg, Miss.
Constantinou, M., and Gazetas, G. (1984). "Probabilistic seismic sliding deforma-
tions of earth dams and slopes." Proc. Fourth ASCE Specialty Conf. on Proba-
bilistic Mech. and Struct. Reliability, ASCE, 318-321.
Cook, R. D. (1981). Concepts and applications of finite element analysis. John Wiley
and Sons, New York, N.Y.
Elgamal, A. W., Abdel-Ghaffar, A. M., and Prevost, J. H. (1987). "2-D elasto-
plastic seismic shear response of earth dams: Application." J. Geotech. Engrg.
Div., ASCE, 113(5), 702-719.
Franklin, A. G., and Chang, F. K. (1977). "Permanent displacements of earth em-
bankments by Newmark sliding block analysis." Earthquake Resistance of Earth
and Rock-Fill Dams, Rept. 5, Misc. Paper No. S-71-17, U.S. Army Engr. Water-
ways Experiment Station, Vicksburg, Miss.
Gazetas, G., DebChaudhury, A., and Gasparini, D. A. (1981). "Random vibration
analysis for the seismic response of earth dams." Geotechnique, 31(2), 267-277.
Harr, M. (1977). Mechanics of particulate media. McGraw-Hill, New York, N.Y.
Lee, K. L. (1974). "Seismic permanent deformations in earth dams." Rept. No.
UCLA-ENG-7497, U.C.L.A., Los Angeles, Calif.
Lin, J. S., and Whitman, R. V. (1986). "Earthquake induced displacements of slid-
ing blocks." J. Geotech. Engrg. Div., ASCE, 112(1), 44-59.
Makdisi, F. I., and Seed, H. B. (1978). "Simplified procedure for estimating dam
and embankment earthquake-induced deformations." J. Geotech. Engrg. Div., ASCE,
104(7), 849-867.
Newmark, N. M. (1965). "Effects of earthquakes on dams and embankments." Geo-
technique, 15(2), 139-160.
Paskalov, T. A. (1984). "Permanent displacement estimation on embankment dams
due to earthquake excitations." Proc. 8th World Conf. on Earthquake Engrg.,
Earthquake Engrg. Res. Inst., 327-334.
Prevost, J. H., Abdel-Ghaffar, A. M., and Lacy, S. J. (1985). "Nonlinear dynamic
analysis of an earth dam." J. Geotech. Engrg. Div., ASCE, 111(7), 882-897.
Sarma< S. K. (1975). "Seismic stability of earth dams and embankments." Geo-
technique, 25(4), 743-761.
Seed, H. B., Lee, K. L., Idriss, I. M., and Makdisi, F. I. (1975). "The slides in
the San Fernando dams during the earthquake of February 9, 1971." J. Geotech.
Engrg. Div., ASCE, 101(7), 651-688.
Serff, N., Seed, H. B., Makdisi, F. L, and Chang, C. Y. (1976). "Earthquake in-
duced deformation of earth dams." Report No. EERC 76-4, Univ. of California,
Berkeley, Calif.
Taniguchi, E., Whitman, R. V., and Marr, W. A. (1983). "Prediction of earthquake-
induced deformation of earth dams." Soils and Found., 23(4), 126-132.
Yegian, M. K., Marciano, E. A., and Ghahraman, V. G. (1988). "Integrated seismic
risk analysis for earth dams." Report No. 88-15, Northeastern Univ., Boston, Mass.
Yegian, M. K., Marciano, E. A., and Ghahraman, V. G. (1991). "Seismic risk
analysis for earth dams." J. Geotech. Engrg., ASCE, 117(1), 18-34.

APPENDIX II. NOTATION

The following symbols are used in this paper:

A = peak ground acceleration;

c.o.v. Ky = coefficient of variation of Ky;'
c.o.v. T = coefficient of variation of T;
D„ = normalized permanent deformation;

49
 Dr = permanent deformation;
dn — median value of normalized permanent deformation;
dr = specified permanent deformation;
fT(t) = probability distribution function of T;
g = gravitational constant;
Ka — average acceleration of a critical sliding mass;
Ky = yield acceleration of a critical sliding mass;
iVeq = number of equivalent uniform cycles;
S = standard normal variate;
T = period of motion of a critical sliding mass;
\iKy - mean value of Ky;
u, r = mean value of T;
°"iogD„ = standard deviation of log D„; and
<I> = cumulative distribution function for the standard normal var-
iate.

50