Proceedings of fOllrth U.S.

National Conference on Earthquake Engineering 'pi,-_~

l\lay 20 24, !()C)O, Palm Springs, California (Volume 21 ~ ai~

SENSITIVITY OF CALCULATED IrJELASTIC SEISMIC RESPONSE

TO INPUT MOTION CHARACTERISTICS

A. J. Kapposl

ABSTRACT

The sensitivity of calculated inelastic response of reinforced con-

crete buildings to the characteristics of input ground motions is
studied, with reference to two typical medium-rise structures. The
method of normalizing the input motions to the same spectrum inte-
nsity is evaluated and it is found that differences in response
quantities calculated for various motions may amount up to 100%. A
modified spectrum intensity is introduced in the normalization pro-
cedure and is found to reduce variability in calculate response pa-
rameters. The results of the present study also indicate that ine-
lastic time-history analysis can be effectively used for design purposes.

INTRODUCTION

Although the seismic design of engineering structures has long been ba-
sed on the concept that a certain degree of inelastic behavior will take place
during the design earthquake, the analysis of these structures is still being
carried out using the traditional elastic approach. This was mainly due to the
difficulties involved in attempting to use inelastic analysis within the design
office environment. HOvlever, it is nO\'1 being recognized(2) that, \''iith presently
available computer codes, it is possible to perform inelastic analysis of rea-
sonably symmetric structures, to estimate the sequence of plastification in cri-
tical zones of structural elements and obtain an idea of the amount of ducti-
lity required in these zones. Modern design codes such as ATC 3-06[3), ANSI
A58.1[Z] and, under a number of restrictions, ACI 318-83[lJ consider inelastic
tlme-hlstory analysis of structures sUbjected to a suite of input base accele-
rograms (actually recorded and/or artificially generated) as a possible alter-
native to conventional design, particularly in the case of important structures.
Nevertheless, even in the case of common structures, such as bui ldings, there is no
one method of inelastic analysis that can be applied to all types of structures.
Furthermore, the calculated response has been found[8],[10) to be sensitive to
the characteristics of the ground motions considered, as well as to the mode-
ling assumptions made regarding the hysteretic behavior of the structural ele-
ments(6), [7].
The present study addresses the problem of sensitivity of the calculated
inelastic response of typical reinforced concrete(R/C) buildings to the chara-
cteristics of the time-histories of ground acceleration. The most commonly used

I
Lecturer, Department of Civil Engineering, University of Thessaloniki ,Thessa-
loniki, Greece.
method of normalizing a set of input motions to the same intensity is evalu:ltr>rj
and a modified technique is introduced.
INELASTIC RESPONSE OF TYPICAL RIC BUILDINGS TO VARIOUS Jr~PUT nOT lOW;
Structures considered and nodeling Assumptions u5ed
h/o RIC structures, a frame and a dual system, shown in Fig.l, 1"Il"!1r:' con-
sidered in the present study. Both structures are assumed to form part of a
symmetric nine-storey building, constructed in an area of medium seismic risk.
where the seismic loading in the one direction is carried by either type of
structure, located at 3.0m centers. The structures were designed to the plovi-
sions of the greek codes applicable in 1984, which did not include specific re-
quirements to ensure ductile behavior. Therefore the two structures represent
the type of construction typical, at least in Southern Europ9, jn the post-war
period up to the early seventies.
The inelastic seismic response of the structures was analysed with the
aid of DRAIN-2D/85 [6], which is an extension of the well-known DRAIN-2D pro-
gram. Standard point hinge modeling was used for RIC members, for which the bi-
linear version of the Takeda hysteresis model[9] was adopted. For the exterior
columns a bilinear hysteresis was assumed, but also taken into account was thl'
effect of fluctuations of axial load on member yield strength. Input data con-
cerning strength and stiffness of RIC elements were calculated using appropri-
ate stress-strain models for confined and unconfined concrete and for reinfor-
cing steel[6].
Motions considered and Normalization Technique used

A suite of five american and two greek records, listed in Table I, I'las se-
lected, based on the criterion of peaking velocity response spectra in the area
of fundamental periods of the structures studied (Tn=0.96 for the frame and
Tn=0.82 for the dual system, if a moderate amount of cracking is assumed).

The seven accelerograms were normalized to the same intensity according

to their spectrum intensity
2.5
51 =
J
0.1
Sv(T,~)dT ( 1)

\"lhere Sv(T,~) is the value of the spectral velocity corresponding to a !w\iorl

T and a damping ratio C Calculated values of SI and correspondinq no 1'111,' 1 ;7,'-
tion factors are given in Table 2. This technique is the IllOSt c0Il1I;10nly usprj nl1r>
and, according to Nishikawa et al.[8], it is better than alternative normaliza-
tion techniques, based on maximum or root mean square acceleration, for sttu-
ctures \'ii th a per i od longer than about 0.3 sec.
For the sake of economy in the parametric analysis and since cUl1lll1ulati','p
ductility demands \'iere not a main issue in the prpsent study. the rluratiot1 of
the records used \'laS limited to 10 sec. It hi'ls. f1nIIPver. to !H' noted t.hat tl,,'
differences in the nonnalization factors given in lable 2 11I1pl1 the full reU11'd',
are taken into account for the calculation of 51. do not exceed 3%.

26
Oi s~~~~~~QLB~~~Ll~
ShOl·/tl in Fig.2a are the envelopes of horizontal story displacements of the
frame structure studied, for five of the input motions of Table l(the remaining
tvlO curves are ommitted fOI' clarity of the figure). It is obvious that, although
the motions have been scaled to the same intensity, considerable differences
were found in the calculated displacements. For example the displacement at the
top, calculated for the Pacoima record, was found to be 63% larger than that
for the El Centro motion, while differences up to 74% were recorded in the other
stories. Although the distribution of displacements along the height of the
frame is similar for most of the motions, some discrepancies \"ere also detected;
note for eXdmple ttlat displacements calculated for the Taft N21E motion are
larger than those calculated for the Cal Tect1 Athenaeum motion in the lo\"er five
stories, while the opposite holds for the upper four stories.

The calculated envelopes of some other characteristic response quantities

(commonly used as seismic damage indices) are ShOlIll in Fig.2 for the three mo-
tions that corresponded to the largest discrepancies found in the parametric
study. The damage indices at the ground story (\',here column failures can occur
due to the presence of high axial loads \·,hich reduce the available ductility) are
higher for the Pacoima motion, IIhi1e the critical excitation I·lith respect to the
upper stories appears to be that of Cal Tech Athenaeum. Differences up to 75''10 in
the interstory drifts and up to 95~ in column ductility factors were calculated
in the present analysis. The increased ductility requirements in the upper sto-
ries stlOuld not be attributed solely to the effect of highel' modes, but also to
the poor design of these stories (resulting in strong beams-I'leak columns). HOI'I-
ever, the sensitivity of calculated inelastic response to input motion characte-
ristics vias also confirmed in the case of RIC structures designed to modern
seismic codes[6].

It has to be emphasized that a motion mayor may not be the critical one
for the structure under consideration, depending on the criterion used. For in-
stance, the El Centro motion is producing smaller displacements and rotational
ductility factors compared I·lith those for the Pacoima motion, but if cllml1lulative
ductility requirements (based on the sum of inelastic rotations in critical re-
gions) are considered, it is found that they are higher in the case of the E1
Centro motion (23/0 for the beams and 136"{, for the columns). As shol'm in Fig.3,
the response to Pacoima 516E is char'acter'ized by a large inelastic cycle follol'Jed
by small cycles, while multiple yielding in both directions is found to occur
in the case of El Centro SGGE, I'esulting in higher cummulative ductility requirements.

Envelopes of interstol'y drifts and ductility factors calculated for the

dual structure of Fig.lb subjected to three characteristic input motions Me
shown in Fig.4. As in the case of the frame structure, significant differences
were found in various response quantities, despite the normalization to the
same spectrum intensity. For example, the interstory drift at t.he ground stmV,
calculi.Jted for the Pacoima Illation, lias 'j7~ higher thdll tlLlt fat' the 1:1 Centl'o
motion and simi jar discrepancies l'lere found in the ottler response parameters.
These discrepancies are in general lower than those calculated for the frame
s t l' U ct u l' e, a ppa I' e nt I y due tot he 101'1 e I' 1eve I of i ne 1a s tic i t yin the dua 1 s t I' U -
cture. Besides the differences in magnitude of response rjllantities, the distri-
bution of them along the height of the dual structure \'ias different fol' lhe va-
rious motions; for instance, the ductility requirements at the ground story are
higher in the case of Pacoima SI6E, while the critical motion, as far as the

27
upper stories are concerned, is Thessa10niki N30E.

Given the nature of the problem, a statistical evaluation of the results

vias a 1so carr i ed out. ShOl'm in Fig, 5 are the average values of the main response
quantities of the frame structure, calculated for the seven input motions of
Table I, as well as ranges of variation defined by one standard deviation on
each side. It is noted that the range of variation is rather limited for all
response quantities. Values of coefficients of variation (x/a), shovlll in Table
3, do not exceed 22% for the displacements, 24% for interstory drifts, 27% for
column ductility factors and 26% for beam ductility factors. It should also be
pointed out that the distribution of variability along the height of the stru-
cture is quite uniform in the case of disp1acements(Fig.5a) and reasonably uni-
tOl'1lI (\'Iith the exception of some upper stories) for the other response quanti-
ties. Modern seismic codes, such as the Draft Eurocode No 8[5] introduce the
use of time domain non-linear dynamic analysis for a set of ground motion time-
histories, as an alternative to conventional elastic seismic design. The results
of the present analysis indicate the validity of the former methodology. How-
ever, more studies are required to establ ish an adequate fraction of the stan-
dard deviation to be used when determining the upper bound of seismic response
for design purposes.

SUGGESTED NORMALIZATION TECHNIQUE

In an attempt to reduce the variability of inelastic response due to dif-

ferences in ctlaracteristics of input motions, a modification of the normaliza-
tion technique based on spectrum intensity (Eq.l) vias introduced.The suggested
modification rel ies on the assumption that the inelastic seismic response of
the structure under consideration is dominated by the fundamental mode, which
is actually the case I~ith most 101'1- and medium-rise buildings. It is also assumed
that the fundamental natural period of vibration of the structure (Tn) is ac-
curately known; in actual RIC buildings which include masonry panels as parti-
tion elements, the estimation of Tn is usually not easy and, in addition, t.he
extend of cracking in both the load-carrying struct.ure ann the partit.ion vla11s
1S not knol~n \'Iith adequate accuracy. These 1imitat.ions should be taken int.o ac-
count. in the normal izat.ion procedure.

In t.he case of struct.ures where the previous assumptions hold, the limits
for ttl e cal cu1a t ion 0f the s pec t I' U min t. ens i t Y (E q. 1) may he appro PI' i a t (' 1yeo n-
densed, In the limiting case of a single-degree-of-freedom structure.subjected
to t\'IO earthquakes, EQl and EQ2, having the response spectra ShOl1ll in Fig.5, it
the accelerations of EQ2 are multiplied by the ratio of spectral velocities
(SvllSv21 corresponding to the fundamental period of the system, the elast.ic
response to t.his scaled accelerogram l'Iil1 be ident.ical t.o that resulting frolll
motion EQI. However, the inelastic response of the system to the accelerogralll
EQ2 x(Svl/Sv21 \'Iill not, in general, be the same as that to EQI, since Tn l'Ii11
be affected by changes in the stiffness of the system and the. ratios of spec-
tral values for the new period will be different from Sv1/Sv2. Therefore, in
the case of inelastic analysis a region of the velocity spectrum,starting from
Tn, has to be taken into account in the normalization procedure. Furthermore.
in the case of mu1ti-degree-of-freedom structures, in which, for reasons men-
tioned earl ier, the fundamental period of the structure is not accurately kno\lI\
a reg i on of the spectrum corresponding to peri ods lovler than Tn shou 1d a 1so he
considered. In the analytical model of the structure the assumptions concerning
element stiffnesses are crucial[7] and it is advisable to use uppet' and 10vlet'

28
bound values for inelastic analysis.
The suggested modified spectrum intensity (Slm) to be used in the norm,,-
lization technique, is defined by the relation
Tn+tZ
Sl m =
f Tn - t l
Sv(T,~)dT (2 )

where tl and t2 are appropriate limits (in sec) defining the width of the re-
gion of the velocity spectrum taken into account. For tl=Tn-O.l and tZ=Z.5-Tn,
Eq. Z co inc ide s I~ it h the s tan da I' d de fin i t ion 0 f H0 usn e I' . ssp ect I' U min ten sit y.

The critical point is, of course, the determination of appropriate values

for tl and t2. For the accelerograms of the present study, considered two at a
time, the way SIm varied vlith increasing values of tl and tz l'ias studied, in-
cluding the case of constant tl (to account fOl' the fact that Tn ahlays increa-
ses in the inelastic range)[4]. It I'las found that no consistent patte)'n of va-
riation of Slm existed; in particular, there are no limiting values of tz and
tl, beyond which the values of the ratios of Sl m, calculated for tl~O ground mo:-
t ions, tend to stab i 1i ze, The resu 1ts presented subsequent 1y \~ere obta i ned by
assuming, more or less arbitrarily, tFtZ=O.ZTn. The problem is still under stu-
dy and further outcome is expected to appear in the near future.

The new normalization factors for the seven accelerograms of Table 1, I~ere
calculated according to Eq, Z in the range of 0.96 sec ±2(),;;,(O.96 is the flln-
damental period of the frame structure). \·Jhen rounded to the first decimal point
(no further "accuracy" is l'larranted by the nature of the problem) these factors
differed from those of Table 2 in the following cases
for Taft 569E: 2.4 (difference of +20%)
- for Ca 1 Tee hAt hen aeum S9 mJ: 3. 2 (- 11 %)
- for Thessaloniki N30E: 2.2 (-15%)
- for Thessaloniki N60\4: 3.0 (+ 7%)

The frame structure of Fig.la was analysed for the above-mentioned mo-
tions and the new set of results vias statistically evaluated. The coefficients of
variation resulting when the suggested normalization technique is used, are
summarized in Table 3. As far as displacements are concerned, the maximum co-
efficient of variation was reduced by 12.5% and the average coefficient (for
all stories) was reduced by 12.3%. The maximum coefficients were reduced by
18.6% in the case of interstory drifts and by 10.3% and 3.9% in the case of
column and beam ductility factors, respectivily. Although the variability in
all response quantities was reduced, compared with the case of standard norma-
lization procedure, the reduction is clearly not drastic and the main conclu-
sions presented in the previous sections are still valid.

CONCLUSIONS

The present study confirmed that the calculated inelastic response of

RIC structures is sensitive to the characteristics of input ground motion. Dif-
ferences in main response Quantities up to about IOO'}, I~ere recorded for tI~o ty-
pical medium-rise structures, although the various input motions were normali-
zed to the same spectrum intensity.

29
 The suggested modification in the norm<1lization technique VirtS l()11llrJ In
reduce the variability of the response, comprtred I'lith that deri,/prj 1.lsillq 111111'
ner's spectrum intensity, The redtlctioll. hOI/ever l'I<1S Ilot drastic ,11ld fl/llllC'l
research is deemed necessary in this field,

On the other hand, the coefficients of vilriation of respollSP fll"liltitips

resulting from the statistical evaluiltioll of the ,rsults of thp PIPSI'Il! studv
wel'e lower than about 30% and their distl'ihution along the height of thp stru-
ctures I'/as quite uniform. This is an indication that inelastic time-history
analysis can be used effectively for design purposes, as suggested by some mo-
dern codes.

ACKrJO\·/LEDG Er·1ErHS

The present paper is largely based on the doctoral thesis presented by

the author to the Civil Engineering Department of thl" Univel'sity of lllC'c;<;,ll,'~
niki. The research l'las supervised by Professol' G.G.ppnel is and vias SI'I.'nSO\ prj
in part by the Onass is Foundat i on and by the Department of Research and 1e r h'H'
logy of the Greek Government.

REFERUHES

1. ACI Committee 318. "Building Cod", P",o'lile11lellts f(1"

Reinforced C"'1<I,,\'·(i"
318-83)", American Concrete Institute. Detroit. r·lich., 1983.

2. American tJational Standards Institute, "rlinilllLJm Design Loads fIJI' P,uilrli"l

and 0 the I' St r uc t ures ( AN SI A58 . 1- 198 Z) ", rJe 1"1 YOI' k. rJ. Y., 1982.

3. App lied Techno logy Counc iI, "Tenta t i ve Prov is ions for thp [)ev p 101'111"11 t ,,'
Seismic Regulations for Buildings", Rep. ATC 3-06, rJSF Publ,78-~\. \'}a","i"'1
ton, D.C., 1978.

4. C. Boskou, "Normalization Coefficients for Earthquake Response c;llf,(·tn",

Diploma thesis (supervised by A.J.Kappos) presented to the Univ.of Thesc;I
loniki, Greece, 1987.

5. Commission of the European Communities, "Eurocode No 8: Structur"~, in '".;

smic Regions-Design-Part 1"(~1ay 1988 Edition), Report EUR 12266 tiL [,11""'1
burg, 1989. .

6. A. J. Ka pp 0 s, " Eva 1ua t i on of the I ne I as tic Se i srn i c Be ha v i our 0 f r·lIllf i c; t " I ,

Reinforced Concrete Buildings", Ph.D. thesis presented to the University

of Thessaloniki, Greece, 1986.

7. A. J. Kappos, "Input Parameters for Inelastic Seismic Anfllysis of RIC F,',l-

me Structures", Proceed.8th European Conference on Earthquake Engineering.
Lisbon,Portugal, 1986, Vol.3, pp.6.1/33-40.

8. T. Nishika\'/a, S. Hayama and T. Seki, "Normalization Parameters of Maximum

Values of Earthquake t'lotion for Non-linear Response Analysis of StructlJrp'~
Proceed. of 8th World Conference on Earthquake Engineering, San Francisco,
Calif., 1984, Vol. IV, pp.83-90.

30
9. S. Otani and ~1. A. Sozen, "Behavior of ~lultistory Reinforced Concrete Fra-
mes during Earthquakes", Civil Engineering Studies, Structural Research
Series No 392, Univ. of Illinois, Urbana, Ill., Nov.1972.

10. G. H. Powell and D. G. ROI-I, "Influence of Design and Analysis Assumptions

on Computed Inelastic Response of Moderately Tall Frames", Rep. EERC-76/11,
Univ. of California, Berkeley, Calif., Apr.1976.

TABLE 1. Basic data for input motions used in the present study

Mognl- Feaklng ~'a>:lmum ~laxlmum

No Earthquake tude Slte Compo- Relllon,,· Acceler- Ve lOCH,
(M) nent ("ecl allon (mm/6eCI
(II )

Imperlal Va lley. 0.50-1 .20

1 6.7 EI Centro Site 500£ 0.348 :135
Callt .. 16-5-1~40 2. 00-3 40

Kern County. Tatt Llncoln

2 7.7 1121£ 0 .60-1 .40 0.156 157
Call! .. 21-7-1952 School Tunnel

Kern County. Tatt Llncoln a .40-0 .50

3 7.7 569£ 0.156 177
Ca llt .21-7-1952 School Tunnal 0 .75-2. 20
-
Kern CClunty. Cal Tech A tt,e- 0 .65-1 .1CJ
4 7.7 590\01 0.053 91
Callt .. 21-7-1952 naeurn. f'asoaena 2.20-3. 20

Son Fernando. 0.35-0 4"'

5 6 .4 Pacolma uam 516£ 1.170 1132
Ca I If .9-2-1971 0.85--1 .80

Thessa lOfllKl. CIty lio tel.

6 6.5 1130£ 0.80-1.30 0.142 127
Greece.20-6-1978 Thessalonlkl

Thc,,!;~ len I f: I. C 1 t Y Hotel. 0.40-0. 65

7 6.5 1160\01 0_154 167
Greece.20-6-197b fhessalonlkl 0.80-1 .05

"eglOng WI th 5v > 2/3 (5-.- 1 (5% d<lJl1ped gpectrwnl

TABLE 2. SI vallies (in mll1) calculated for the first 10 sec of the input
motions and cOl'lesponding normal ization factors I

Accelerogrem 51 n-O%l 51 n-O%l 51 n-10%1 Factor

El Centro SOOE: 2334 1479 1230 1 .00

Talt 1121E 1021 651 555 2. 24

Tatt S69E 1176 749 605 2. 00

Cal "I ech ,S90W1 I 6 ),1 414 338 3 .60
PaCOIma S16E 5676 3910 3316 0.37

ThessalonlKl 1l30E 721 555 484 2.61

Thc3S0JOIlIY.l 1100\01 698 51.7 457 2.78

Average at eases 1-5. and 1-10.

Second 10 gee

31
 TABLE 3. Coefficients of variation for maximum response of frame structure

Normallzotl0n Re~pon~e S tor y

Technique ~uontity 2 3 4 o

x 0.186 (j.185 0.191 0.204 0.207 O. 216 0.215 0.205 O. lfF )

nccording to hx 0.186 O. 19·1 0.211 (l ~32 O. ~37 O. 19;' [).13" 0. 101 r, nn-'

51 \JOe 0.230 0.239 0.210 0 .27;' O. 184 0.162 O. 150 0. 11.' 0.090

U.h O.21~ 0.174 0 21~ 0 ~5SJ L' . 251 O. 166 (I 168 O. 2:!6 0.100

x 0.1890.176 0.16 0 (l.171 0.170 C'.17~ 0.17 0 OJ 17<' u I,'

o.ccordlTl'J to !Ix 0.190 0.170 0.17·1 t, la' 0 IBo 0.17:' 0.13.' {) lJ" tJ 1.']

51_ 0.240 0 208 0.176 0.24'1 0.!96 0.141 0.L?3 0.1:.. 10.1;"

\JO~ 0.230 0.152 0.177 0.219 0.204 0.1" 0.122 0.240 O.Oo~

q q
I , I
/ / i

~"
8
// /.; /
/

I .. j

/
1/
/ ..../
,./
) 1(,,)
5 / / / ..;>') 5 //

4 /1
,( .J
j' _ H Cmtro ~[-IO 4 .r (
f ./ --- Taft H21Ea21 I )
I' // Pawlmo 51b[ -04 / /
1/.:.';' ThBsoteml\<; H~OE.2 b / (
Co! Trch At~""narurn 570.... 30
2 1///
~//
,.-;
2
;
,II /
II (oj I (b)
li,/h
o L-_-'- '--_..-.-_-'_
os 10 \%)

q Q

8 6

/) /)

4 4

(d)

o . ~ L _.~_---.l_. _ _ J.._.. _ .i

746 10 12

FIGURE 2. Maximum response quantities for the frallle stl'uctlHP subjpct,r>d tIl
various input motions: (a)Storey displacements. (b)Relative interstory drift\.
(c)Required column ductility factors and (d)Required beam ductility factors

32
 q 78 , ~o
(a) - (b)

25 ~6 11
II
20/70 7!J110
35/35 2'5/40
12 23 24

19 20 21

•
o
-, -.0/40 ::<1/55
25/75

18
~110
W 0
•
'"o 16 II N
I

.
"

'" ..,
0

'" !J I. 2'5/11'5
I' 25/110
C>

~0/40
10
40/65
11 12
<D

J B ~
juNu
2'5/65
,.')/ .. 5 ~0/IO
4 '5 6
- 1
:lOi9'5

"1'5'5 '55/'/0 <D ~.

L 0 ~~
I
- .; ~~ t
~--60 -+- 60--1 I---b_O~3_0+-()_O~
FIGURE 1. Geometric data of structures studied (section dimensions in em)

,,'"
lOO too

(b)
." ~~~ L-~~_~~---J -,,, ~~ ~_-L-_~~ J
lea -1 ~ ., C -2 , =, ,c '::c • '0 0 , ~ ., C ·l ~ 1 ~ s c: , )
~(~_FlQ lef])

FIFURE 3. Bending moment-plastic hinge rotation diagrams for the left beam
of the sixth story for: (a)EI Centro motion and (b)Pacoima motion

FIGURE 6. Determination ot the normaliza-

tion factor fOl' input motions in the case
£OJ of a single-degree-of-freedom system

33
 q q

~\~~\
o 0 r. P;l
~Ii)

\
.....

6 ~ £e C~'1tro saOE B B

I,.~
o---a Pacotma St6E
Tne"3!0'1'ld N30E
I>----'> - cnP\l""l~ " \,
r)j - - - watt ,'.. \

6 I> "I

"
I> '~"\~,,"
\
~ "

'\ }\ 1
5 " 5
"
~
4

2
4 "I ' ' "
tH
"I I .I

~
'",. ,
l>

(.
IV )}
(a) Ie)
6./h lJa p"
OL---'-----'---'----' 0 0
05 10(%) 3 5 I> 2 ~

FIGURE 4. Maximum response for the dual structure subjected to various input
" 5 I>

111(1-

tions: (a) Inters tory drifts, (b),(c)Ouctility factors for vertical members
Q q
I ~ (
1 (

8 x-C5 --I /
L-- ;(+(5 8
I
I /
I /
7 /
I
I /
I / I>
I> /
I /
I /
5 I / 5
1 I
1 /
I / 4
1 /
I /
/ 3
1 /
1 I
I I 2
2 1/
1I
1'1
II (a) (b)
x A1./tl
0 0
:xl 100 150 lmml 05 10(%)
9

" --
9

7
'" .....~ ::::::
-- ?
';
/
'/
B
~'_::::c::c=::=_=~ __ ,
/ /
/ / I I
I>
/
I
/
/ I> ~~~'.J ~./~~
-:: ..- r-~~ \~-~
/' ./
/
~ ....... ~
/-
(('
4 \ \ 4 ) ,)

3 (
(
\

I
\
I
3
,
\
/'
/'
/'

I
~
- ~

/ ( \ I
2 ,
\ \
\
\
,
/
\ )

\
\ , (e) (et)
\ \

1-19 Ilg
0 0
4 5 I> 3 4 5 6 8 Q ,0

FIGURE 5. Average response Quantities felt' the frarllE' structure c"lr"Iatprf Inl
the seven accelerograms of Table 1: (a)Story displacements, (b)Intelstory
drifts, (c)Column ductility factors and (d)Beam ductility factors