You are on page 1of 18

INELASTIC EARTHQUAKE RESPONSE

OF ASYMMETRIC STRUCTURES
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

By Y. Bozorgnia 1 and W. K. Tso, 2 M. ASCE

ABSTRACT: The inelastic seismic response of a class of one-way torsionally un-


balanced structures is presented. The structural model consists of a single mass
supported by bilinear hysteretic elements. The yield strength of the model is a
function of the lateral period and follows a trend similar to the design strengths
suggested by design codes. The response parameters of interest are the resist-
ing element ductility demands and the edge displacement of the model. The
sensitivity of the response parameters to system parameters, such as structural
eccentricity, yield strength, uncoupled lateral period, and uncoupled torsional
to lateral frequency ratio, is examined for two types of ground motions. It is
found that the effect of asymmetry is most pronounced for stiff structures with
low yield strength. Unlike elastic response, the uncoupled torsional-to-lateral-
frequency ratio is not a sensitive system parameter that affects the inelastic
responses. Exceptionally large ductility demand can be expected on eccentric
stiff structures with low yield strength when exposed to ground motions with
large and long duration acceleration peaks as exemplified by the 1977 Romanian
earthquake record.

INTRODUCTION

Earthquake excitations induce both translational a n d rotational mo-


tions in eccentric buildings. Since an early investigation by Ayre (1), many
studies have been made on the elastic vibrations of such structures. Since
some degree of inelastic deformation can be expected in buildings u n d e r
moderate or strong earthquake ground motions, information on the in-
elastic response of eccentric buildings is important for seismic design
purposes. In spite of such needs, relatively few studies exist on the in-
elastic earthquake responses of asymmetrical structures (4-6,10-12). For
elastic response studies, the structural eccentricity a n d the uncoupled
torsional to lateral frequency ratio are two system parameters that strongly
influence elastic seismic responses. However, opinions differ regarding
the importance of these parameters on the inelastic torsional responses
(4,5,10). One reason for the diverse views is that different response pa-
rameters have been used to characterize the inelastic torsional effect.
Another reason is that different structural models are used in each of
the investigations. While only a few system parameters are n e e d e d to
completely define an elastic eccentric system, more detailed specifica-
tions on the system such as the location a n d force-deformation relation-
ship of each resisting element in the structure are n e e d e d in inelastic
'Post-Doctoral Fellow, Dept. of Civ. Engrg. and Engrg. Mech., McMaster Univ.,
Hamilton, ON, Canada.
2
Prof., Dept. of Civ. Engrg. and Engrg. Mech., McMaster Univ., Hamilton,
ON, Canada.
Note.—Discussion open until July 1, 1986. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manu-
script for this paper was submitted for review and possible publication on No-
vember 9, 1984. This paper is part of the Journal of Structural Engineering, Vol.
112, No. 2, February, 1986. ©ASCE, ISSN 0733-9445/86/0002-0383/$01.00. Paper
No. 20389.
383

J. Struct. Eng. 1986.112:383-400.


response studies. As a result, a number of systems can have the same
elastic response, but different inelastic responses. Alternatively, one can
view inelastic response studies as more model dependent than elastic
response systems.
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

Another difference in previous studies is the specification of the yield


resistance of the structure as a function of the structural periods. It was
determined, either based on a reduced elastic spectrum (4,10) or an in-
elastic spectrum (5) of the specific earthquake motion input. While such
a specification of the design strength is convenient to contrast the in-
elastic versus the elastic responses of the system under the same exci-
tation, the implication of the results obtained in these studies is not clear
for current building design practice. Building strength is usually guided
by building codes and the seismic coefficient in most building codes ap-
pears as a smooth spectrum. Therefore, the yield strength of the eccen-
tric model under study should be defined similarly to code-strength
specifications before the results obtained are interpreted for current de-
sign practice.
In this paper, a parametric study is carried out on the inelastic re-
sponse of a single mass, one-way, torsionally unbalanced system sub-
jected to horizontal earthquake excitations. The chosen system is simple
enough that its behavior can be characterized by a limited number of
parameters, yet has the essential characteristics of actual structures. It
is statically indeterminate, consisting of three resisting, bilinear, hyster-
etic elements to provide stiffness and strength in the direction of the
ground motions. Unlike in the previous studies, the variation of the yield
strength of the model with structural period is independent of the input
ground motion, and follows a smooth curve derived from the Newmark-
Hall smooth average elastic spectrum (7). In this manner, the yield strength
variation of the model is more closely related to current building prac-
tices. For a given lateral period, three levels of yield strength are con-
sidered, corresponding to the condition that the system is excited to the
threshold of yielding, to moderate inelastic deformation, and well into
the inelastic range, respectively.
Two categories of earthquake records are used as inputs. The first cat-
egory consists of records that have their elastic acceleration spectra sim-
ilar to the design spectrum recommended by Newmark and Hall, as typ-
ified by the 1940 El Centro and the 1952 Taft earthquake records. The
second category consists of records that show a few pronounced accel-
eration pulses and in general have acceleration spectra dissimilar to the
Newmark-Hall design spectrum, as exemplified by the 1965 Parkfield
and 1977 Romanian earthquake records. Using these records, the effect
of structural eccentricity, uncoupled lateral period, uncoupled torsional
to lateral frequency ratio, and the yield strength of the system on the
resisting element ductility demand, and on the building edge displace-
ment, are examined.
This paper aims first to clarify the sensitivity of the design parameters
such as ductility demands to the different system parameters, and, sec-
ond, to examine the consequence of eccentric structures designed ac-
cording to current code provisions when subjected to ground motions
dramatically differing in frequency content from the codified design
spectrum.
384

J. Struct. Eng. 1986.112:383-400.


ASSUMPTIONS AND FORMULATION

Consider a single-story structure consisting of a rigid deck of mass m


subjected to unidirectional horizontal ground motions. Resisting ele-
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

ments in the structure are assumed to be frames or walls having strength


and stiffness in their planes only. For simplicity, resistances of the mem-
bers that are not in the direction of the ground excitation are ignored.
The force-deformation relationship of each element is assumed to be of
the bilinear hysteretic type. The system eccentricity, e, is taken as the
distance between the center of mass, CM, of the deck and the initial
center of stiffness, CS, of the elements. It is assumed that eccentricity
exists only in a direction perpendicular to the ground motions. Thus the
structure is a one-way torsionally coupled system with two degrees of
freedom. The degrees of freedom will be taken as a lateral displacement
relative to the ground v(t), and a rotation 9(f), as shown in Fig. 1. Using
the initial center of stiffness as the point of reference, the undamped
equations of motion can be written as

m me m
(1)
me mp T me
in which p = the radius of gyration of the deck about a vertical axis
through the center of stiffness; vg = the ground acceleration; F = the
restoring lateral force; and T = the restoring torque.
Let Kv = the elastic lateral stiffness and K9 = the elastic torsional stiff-
ness of the system referred to the center of stiffness, CS. They are de-
fined by

K
" = 2 k°< and Ki =
2 %1 k
i °<- (2)

in which koi = the elastic stiffness of the z'th resisting element along the
Y-axis; and x, denotes the distance of the element measured from the

*- X

GROUND MOTION

FIG. 1.—Plan View of Single Mass Model

385

J. Struct. Eng. 1986.112:383-400.


center of stiffness. The uncoupled lateral and torsional frequencies, to0
and we, and the dimensionless eccentricity, e*, are given by
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

Kv I K„ „ e
a n d e
—; <oe=-i/—;; =- (3)
m \ mp p
The frequency ratio of (oe/a>B is denoted by fl. With the system param-
eters denned, the restoring force and torque can be expressed as follows:
In the elastic range, F = Kvv, T = K99, and Eq. 1 reduces to

a2
e*Cl
e*n
1 {SMeH£H -. <*>
in which 9 = flp8. Eq. 4 is the linear dynamic equilibrium equation,
which has been studied in detail (3,9).
For the inelastic response, it is assumed that the resisting elements
have identical yield deformation v0. Then the yield strength of the struc-
ture is given by
F0 = Kvv0 (5)
In other words, if the lateral force F0 is applied at the center of stiffness
of the eccentric structure, all elements will yield simultaneously. Eq. 1
can be written in the nondimensional form as

n2 e*n
e*n l • Mr*. "Uf"»^i <6>
v flp8 _. F F
in which !! = - ; $ = -!-; F* = —; and T* = • • (7)
v0 v0 F0 CipF0
The symbol a* in Eq. 6 represents a characteristic acceleration and is
used as a measure of the strength of the ground excitation. In this study,
the characteristic acceleration, a*, is related to the yield strength F0 by
ma*
Fo = (8)
T-
The parameter R in Eqs. 6 and 8 can be considered as a modification
factor to relate the yield strength F0 and the intensity of ground exci-
tation.
The deformation of the r'th resisting element in Y direction, vt, is ob-
tained by the following transformation:
Vi <&Xi

vQ ilp
The incremental force and displacement vectors of the system are related
by

*l K(f)
!AT*h K W A$
A* (10)

386

J. Struct. Eng. 1986.112:383-400.


1
wft)
in which K(t) = (na)
S(t) n2(t)
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

.dp a2
S fc((o
Coft) (lib)

2 *ftw
n2(t) = (He)
1
2 *••(')

2 *MO
S(t) = (lid)

and kj(t) is the tangent stiffness of the ;'th resisting element at time t.
Therefore, u>v(t) and 11(f) may be interpreted as the instantaneous lateral
frequency and torsional to lateral frequency ratio, respectively. S(t) de-
fines the current position of the stiffness center. The frequencies and the
position of the center of stiffness may be changed during the motion of
the system due to yielding and unloading of the elements. In the elastic
range of behavior K(t) reduces to the identity matrix; thus F * = U, T*
= <3>, and Eq. 6 reduces to a nondimensional form of Eq. 4. Comparing
Eqs. 4 and 6, it can be seen that the three parameters co9, O and e* that
are sufficient to define the elastic property of the system are not suffi-
cient to describe the system if yielding is allowed. The actual location
and force-deformation relationships of each element are needed for the
analysis. Thus, it is possible to have several different inelastic eccentric
models having the same set of co9, CL and e*, and thus the same overall
elastic responses, but having different inelastic responses.

SYSTEM CHARACTERISTICS

Since the inelastic response of a torsionally unbalanced system de-


pends on more parameters than those of the elastic response, a careful
specification of the model and its characteristics is necessary to interpret
the results. In this study, a statically indeterminate model is chosen,
consisting of three resisting elements providing stiffness and strength
in the direction of the ground motion> as shown in Fig. 2. Element 2 is
located at the center of mass of the rigid deck, while elements 1 and 3
are located at equal distance h from, and at opposite sides of element 2.
Elements 2 and 3 are identical, but element 1 has a higher elastic stiff-
ness and yield strength than the other two members. Thus, the system's
asymmetry is due to the uneven stiffness distribution, a common cause
of asymmetry in many actual buildings. The uncoupled lateral period Tv
= 2TT/O)0 of the model is considered to lie between 0.1-2.0 sec, a range
that covers the period of the fundamental mode of most buildings. The

387

J. Struct. Eng. 1986.112:383-400.


Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

FIG. 2.—Plan View of Model Considered

uncoupled frequency ratio Q, takes on values of 0.8, 1.0 and 1.2; 0 = 1


is included because of its importance for elastic response of torsionally
coupled systems.
To interpret the result, it is convenient to specify the eccentricity of
the system in terms of the eccentricity ratio e = e/r, where r = the radius
of gyration of the deck about a vertical axis through the center of mass.
e is related to e* by

Two classes of structural configurations are studied, corresponding to


eccentricity ratios e of zero and 0.25. The structural configuration with
e = 0 is referred to as the symmetrical system in which all three elements
are identical. For the eccentric configuration, e = 0.25 corresponds to
eccentricity values between 7-10% of the plan dimension for a uniform
rectangular deck; the exact value depends on the aspect ratio of the deck.
The force-deformation relationship for each element is assumed to be
bilinear hysteretic with the post-yield stiffness as 3% of the initial elastic
stiffness. Since it has been assumed that all the elements have identical
yield deformations, the yield strength of each member is proportional
to its initial elastic stiffness.
The characteristic acceleration, a*, is a measure of the strength of the
ground excitation. It may be taken as the peak ground acceleration (2),
the elastic pseudo-acceleration spectrum (PSA) of the ground motion
(4,10), or some other measure. Herein the characteristic acceleration a*
is taken as a smooth elastic PSA rather than a PSA of a specific ground
motion record, a* is assumed to be constant for Tv s 0.5 sec, and it varies
inversely to the period for Tv > 0.5 sec. When normalized with the peak
ground acceleration A, the a*/A ratio appears as shown in Fig. 3. The
flat portion of a*/A is taken to be 3.68, which is the median value of
the acceleration amplification factor for a 0.5% critically damped elastic
oscillator as recommended by Newmark and Hall (7). This relatively large
acceleration-amplification factor is consistent with the understanding that
the true viscous damping in the system is in fact small (it is ignored in
the computations). The energy dissipation is mainly provided by the
hysteretic damping of the elements when the system is excited into the
inelastic range.

388

J. Struct. Eng. 1986.112:383-400.


s
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

3
<

1
i i l l i J J I I L
0 1 2
T v (SEC)

FIG. 3.—Normalized Characteristic Acceleration Spectrum

The yield strength FQ is related to the characteristic acceleration by Eq.


8. With R assumed to be independent of the lateral period, the yield
strength F0 will have the same period variation as a*. In this way, the
yield strength of the model is prescribed to follow a trend similar to
many of the seismic code provisions. The modification factor R plays a
similar role to that of the "response modification factor" as described in
ATC-3 (8). For R = 1 and when the system is lightly damped, the sym-
metrical model is expected to just reach yield, if the spectrum of the
excitation is similar to a*. A larger value for R implies that the system
is designed with a lower yield level than the elastic demand, and the
resisting elements are expected to respond inelastically. For the eccentric
configuration, element 3 may be excited into the inelastic range more
than the other members due to the additional torsional response of the
system. In the present study, three yield resistances of the model are
considered, corresponding to R = 1, 3, and 5. R = 1 represents the class
of structures expected to respond elastically, while R = 3 and R = 5
represent the structures designed to resist the earthquake motion with
moderate and large inelastic deformations, respectively.

GROUND MOTION EXCITATIONS

In order to study the consequence of eccentric structures with yield-


strength design based on seismic code provisions and subjected to ground
motions of different characteristics, two categories of ground motions
are considered. The first category consists of records exhibiting highly
irregular acceleration motions and having the acceleration spectral shapes
similar to the shape of a*. Such records have been extensively used in
earthquake response studies. Typical examples of these are the records
of the 1940 El Centro, S00E component and the 1952 Taft, S69E com-
ponent. Shown in Fig. 4 are the normalized elastic pseudo-acceleration
spectra of these records for 0.5% critical damping. Plotted in the same
figure by dashed lines are the F0/(mA) of different classes of structures
corresponding to R = 1, 3, and 5. One can interpret the ground motion
spectra as the "strength demand" curves if the response is to remain
elastic. The dashed lines are the "strength supply" curves for the dif-

389

J. Struct. Eng. 1986.112:383-400.


EL CENTRO
I 1 1 1 1 1 1 1 1 1
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

- -
- •

-
R=l I \f vv
3

1 1 1 7i i
0.1 0.2 0.4 0.6 1.0 2.0 0.1 0.2 0.4 0.G . 1.0 2.0
LBTERBL PERIOD Tv (SEC) LATERAL PERIOD Tv (SEC)

FIG. 4.—Normalized Elastic Pseudo-Acceleration Spectra of El Centra and Taft


Records for 0.5% Critical Damping, and F0/(mA) Curves

- 0.3
o
- 0.2
0 0.1
H
£ 0.0 "~?~ •V*Y»
a:
u) -0.1
u "0.2
u
1 -0.3
6 8 10
TIME ( SEC )

FIG. 5.—Acceleration Records of 1966 Parkfield, Array No. 2 (N65E), and 1977
Romania (NS)

0.2 0.4 0.6 1.0 2.0 0.1 0.2 0.4 0.6 1.0 2.0
LBTERBL PERIOD Tv (SEC) LBTERBL PERIOD T» (SEC)

FIG. 6.—Normalized Elastic Pseudo-Acceleration Spectra of Parkfield and Ro-


mania Records for 0.5% Critical Damping, and F0/(mA) Curves

390

J. Struct. Eng. 1986.112:383-400.


ferent classes of structures. With a given period, inelastic action can be
expected if the strength demand exceeds the strength supply.
The other category consists of the records showing a few pronounced
pulses in their accelerograms and having acceleration spectra dissimilar
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

to a*. Examples of these are the records of the 1966 Parkfield, Array No.
2, N65E component and the 1977 Romania, NS component. The first 16
sec of the acceleration time histories of these records are shown in Fig.
5. The normalized response spectra of these records for 0.5% critical
damping, together with the strength supply curves corresponding to R
= 1, 3, and 5 are shown in Fig. 6. As is evident from this figure, the
spectra of both records are dissimilar to a*. In particular, the spectrum
of the Romania record has pronounced peaks in the period range of 1.0-
2.0 sec due to the long-duration acceleration pulses in the accelerogram.
While the inelastic responses of symmetric systems subjected to the near-
fault records containing such pulses have been investigated before (2),
the writers are not aware of any such study on asymmetrical systems.
The inclusion of these records as ground motion inputs in this present
study will therefore provide added insight to the consequence of eccen-
tric structures subjected to this type of ground motions.
In the present study these four earthquake records are used as the
input motions to the symmetric and eccentric systems. The duration of
the El Centra, Taft, Parkfield and Romania records are taken to be the
first 30, 30, 20, and 16 sec, respectively, and the records are normalized
to have the same peak acceleration. With this system properties and the
ground motions specified, Eq. 6 is solved numerically assuming linear
variation of the acceleration during each time increment. The time in-
crement is taken to be no more than l/30th of the smaller modal period
of the corresponding elastic, torsionally-coupled system.

SYSTEM RESPONSE

Before going into details of dynamic responses of the system, it is use-


ful to understand the behavior of the model under a monotonically-in-
creasing, static lateral load acting through the center of mass of the sys-
tem. Under such loading, element 3 yields first, and, consequently, the
center of stiffness moves further away from the center of mass. If the
elements were elastoplastic, the force of element 3 would remain con-
stant at its yield strength, and the force developed in element 1 also
would remain constant, being equal to that of element 3 in order to sat-
isfy the equilibrium of the moments. Since the yield strength of element
1 is greater than that of element 3, element 1 remains elastic. Thus, after
element 3 yields, element 1 will be stationary and the deck will rotate
about element 1. The ultimate load that leads to the yielding of element
2 is given by (1 — e/h)F0/ which is less than that of the corresponding
symmetric system. The system is incapable of resisting any additional
load beyond the ultimate load. Thus, one can expect asymmetry will
have a large effect on the lateral displacements of element 3 and also at
the right-hand edge of the system under the static load. This analysis
of the static behavior of eccentric system will be useful later to under-
stand the dynamic inelastic behavior of short-period systems under se-
vere earthquake loading.
391

J. Struct. Eng. 1986.112:383-400.


The displacement ductility ratio of the resisting elements under seis-
mic excitations, and the displacement at the edge of the slab are the two
main response parameters considered herein. The effects of the design
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

strength, i.e., the change of R value, the uncoupled lateral period, and
the frequency ratio (1 on these response parameters are examined. Four
situations are considered: (1) Symmetrical; (2) eccentric structures sub-
jected to the "El Centro-type" excitations; (3) symmetrical; and (4) ec-
centric structures subjected to the records with a few pronounced ac-
celeration pulses. Situation 1 has been studied extensively, and the be-
havior of the structure under this situation is fairly well understood. It
is included here as a reference. By comparing the results between 1 and
2, the influence of asymmetry on the design parameters can be assessed
when the structure is exposed to ground motions with frequency con-
tent similar to that assumed in the design spectrum. Situation 3 gives
the inelastic behavior of symmetrical buildings subjected to the near-
fault records, a situation similar to the study of Bertero et al. (2). Situ-
ation 4 gives the consequence of eccentric structures design according
to code-specified strength and subject to near-fault ground motions with
prominent acceleration impulses.
Ductility Demand.—Displacement ductility factor represents a mea-
sure of the degree of inelastic deformation developed in a member dur-
ing the ground shaking. Element 3 in the model is generally subjected
to larger lateral displacement than the other elements. Therefore, the
ductility ratio of this element is chosen as the ductility parameter of in-
terest. Fig. 7 shows the ductility ratio of element 3, plotted as a function
of the uncoupled lateral period T„ for different values of R, when the
system is excited by the El Centro and Taft records. In this figure the
solid lines are for the eccentric while the dashed lines are for the sym-

EL CENTRO
—I

0.1 0.8 0.4 0.6 1.0 8.0 0.1 0.8 0.4 0.G 1.0 8.0
LATERAL PERIOD Tv (SEC) LBTERBL PERIOD Tv !SEC)

FIG. 7.—Ductility Factor of Element 3: El Centro and Taft Excitations (CI = 1.2)

392

J. Struct. Eng. 1986.112:383-400.


metric systems. For symmetric systems with R = 1, the ductility factor
is about unity, indicating that the system is excited to near the yield
level. For the systems with lower yield resistances, the ductility demand
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

is about or less than three for R = 3; and less than five for R = 5 when
Tv > 0.4 sec. Higher ductility demand generally exists for stiffer struc-
tures with periods below 0.4 sec. This effect is particularly noticeable for
systems having R = 5.
Asymmetry generally results in an additional ductility demand on re-
sisting element 3. The effect of eccentricity can be seen by comparing
the solid and dashed lines for the same R value in Fig. 7. The effect of
eccentricity is particularly important for short period structures with low
yield strength. Ductility demand can be up to three times that of the
corresponding symmetrical system in this period range.
Consider the dashed lines in Fig. 8, which are for the symmetric sys-
tems subjected to the Parkfield and Romania records. The trends of these
differ substantially from the corresponding ones in Fig. 7. In the case of
R = 1, the ductility demand generally increases with increasing period.
This trend is particularly apparent for the responses associated with the
Romania record. As shown in Fig. 6, the strength demand is less than
the strength supply for this class of structures in the short period range.
Therefore, the system responds elastically to the excitation and a duc-
tility factor less than unity results. For periods longer than 0.5 sec the
strength demand tends to exceed the strength supply, resulting in a duc-
tility factor larger than unity. Because of the unusual spectrum associ-
ated with the Romania record, the strength demand far exceeds the
strength supply in the period range of 1.0-2.0 sec, causing a larger duc-
tility in this period range than for the shorter periods. A similar strength-
supply-and-demand argument can also be used to explain the behavior

_l I I ' l l ll i i I I I I I l I I I II l I |_
0.1 0.2 0.4 0.6 1.0 2.0 0.1 0.2 0.4 0.G 1.0 2.0
L A T E R A L P E R I O D T v 1SEC1 L A T E R A L P E R I O D T v (SEC)

FIG. 8.—Ductility Factor of Element 3: Parkfield and Romania Excitations (ft =


1.2)
393

J. Struct. Eng. 1986.112:383-400.


of the symmetric systems with strength corresponding to J? = 3. The
ductility factors of the systems associated with R = 5 are substantially
greater than those with R = 3 in the short-period range using the Ro-
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

manian ground motion record. The design yield force corresponding to


R = 5 is sufficiently low that even at short periods the strength demand

0.1 0.2 0.4 0.6 1.0 2.0 0.1 0.2 0.4 0.6 1.0 2.0
LATERAL PERIOD Tv (SEC) LATERAL PERIOD T, (SEC)

FIG. 9.—Effect of Frequency Ratio on Ductility Factor: El Centro and Taft Exci-
tations (e = 0.25)

0.1 0.2 0.4 0.6 1.0 2.0 0.1 0.2 0.4 0.6 1.0 2.0
LATERAL PERIOD Tv (SEC) LATERAL PERIOD Tv (SEC)

FIG. 10.—Effect of Frequency Ratio on Ductility Factor: Parkfield and Romania


Excitations (e = 0.25)

394

J. Struct. Eng. 1986.112:383-400.


exceeds the supply and the system is excited into the inelastic range.
Once yielded, the effective period of the system increases, and greater
imbalance between strength supply and demand, resulting in further
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

increase of responses. Thus, a large ductility demand is observed for the


systems with design strength greatly reduced from the elastic strength.
Comparing the solid and dashed lines in Fig. 8 shows that the effect
of eccentricity is large in the short-period range. Again, asymmetry can
increase the ductility demand by a factor up to three as compared to
that for the corresponding symmetric system. Since the ductility de-
mand for a short-period symmetrical system with a low design strength
is already substantial, the ductility demand for the corresponding ec-
centric system becomes very large.
The ductility factors of element 3 in the eccentric model for different
values of the uncoupled frequency ratio ft are shown in Figs. 9 and 10.
It can be seen that the element ductility demand is not very sensitive to
the frequency ratio ft. The system becomes torsionally more flexible when
ft decreases. On the other hand, decreasing ft implies the elements are
closer to each other. The former factor tends to increase the deck rota-
tion, while the latter reduces the effect of rotation on the element de-
formation. Thus for the structural model considered herein, ft becomes
a less critical parameter for the element ductility demand.
Edge Displacement.—Edge displacement of the deck may represent
the nonstructural damage potential. Displacements at other points, such
as center of mass and center of stiffness, are not considered herein as
the primary response parameters because of their limited practical in-
terest, even though they help interpret the behavior of the system. It
was found in previous studies (5,12) and confirmed in the present study
that the displacements at the center of mass and at the center of stiffness
are not very sensitive to the eccentricity. Herein maximum displacement
at the right edge, the edge parallel to the direction of the ground motion
and furthest away from the center of stiffness, is chosen as the displace-
ment measure. The distance between the initial center of stiffness and
the right edge, xr, depends on the geometry of the floor and the eccen-
tricity. For a uniform rectangular deck xr < e + rV3. In this study, the
displacement at xr = e + 2r is considered in order to cover possible non-
uniformity of the deck. Let vr and i^o denote absolute maximum dis-
placements at the right edge of the eccentric and the corresponding sym-
metric systems, respectively. Then the edge displacement ratio, vr/v^,
represents the effect of asymmetry on the lateral displacement at the
edge. Plotted in Figs. 11 and 12 is the edge displacement ratio as a func-
tion of the uncoupled lateral period for different values of R and ft, for
the Taft and Romania records. It can be seen that the effect of eccen-
tricity is generally more pronounced for the short period structures with
low yield resistances. Also, the effect of asymmetry on the edge dis-
placement is sensitive to the frequency ratio ft in the short period range
when R takes on a relatively high value. This sensitivity is more than
that for the element ductility because, unlike the element, the location
of the edge is fixed and independent of ft. Thus, asymmetry affects the
right edge displacement more than it affects the ductility ratio of element
3, especially for torsionally flexible structures. Depending on ft, the lat-

395

J. Struct. Eng. 1986.112:383-400.


TTT r
n
1.2
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

T- 1.0
0.8

_l I
0.1 0.2 0.4 0.6 1.0 2.0 0.2 0.4 0.6 , 1.0 2.0
LATERAL PERIOD Tv (SEC) LATERAL PERIOD Tv (SEC)

FIG. 11.—Edge Displacement Ratio: Taft Excitation

ROMANIA ( R = 3 ) ROMANIA I R=5

0.1 0.2 0.4 0.6 1.0 2.0 0.2 0.4 0.6 1.0 2.0
LATERAL PERIOD Tv (SEC) LATERAL PERIOD Tv (SEC)

FIG. 12.—Edge Displacement Ratio: Romania Excitation

eral displacement at the edge of the eccentric system can be two to six
times that of the corresponding symmetric system in the short period
range. Similar results, although not shown here, were obtained for the
other two ground motion records.
In order to gain insight into the behavior of the system, examples of
the response time histories are shown in Figs. 13 and 14. In these fig-
ures, U = the dimensionless displacement at the center of stiffness, 4>
= the dimensionless rotation of the deck, as defined in Eq. 7, and 8
denotes the ratio of the edge displacement v, to the yield deformation
v0. Shown in these figures are the responses of a short period system
(T„ = 0.2 sec) and a long period system (Tv =• 1.0 sec) both having low
yield strength (R = 5) and subjected to the Taft record as input excita-
tion. The translational and rotational motions are almost in phase when
the maxima occur for the short-period structure. As a result, the tor-
sional motions have a large effect on the edge displacement. On the
other hand, the translational and rotational responses are not in phase
for the long period structure; thus, they do not reinforce each other to
produce as large effect of asymmetry as in the former case.

396

J. Struct. Eng. 1986.112:383-400.


TUFT (R=5.T v =0.2.n = 1.2)
20

10
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

_, /^\rtu^/l,^.nWV«~lW(^A»v
u 0
\/»i»fV>W
l i

-10
1 1 1 1 1 I 1
-20

20

10

$ 0 A***
-10
i i i i i i i
-20

20

10 I
5 0

-10
! 1 1 1 1 1 1
G 8 10 12 14 1G
TIME ( SEC )

FIG. 13.—Dimensionless Response Time Histories for Short Period System; Dis-
placement U, Rotation $, and Edge Displacement 8: Taft Excitation (T„ = 0.2 sec,
e = 0.25)

TfiFT (R=5,T v =1.0,n=1.2)

U 0
^\ A / \ A r-^\ r\„/\ A/"V^
~ W W
-3
I I I I I -I I
-G

/N A A A / \ .
\/\/\v/\/N /\/VV

I I I I

. G

S 0 - -A A - A /"A "i ^ ^ A A
\/V\J\AJ\I\J\
i i i
G 8 10 12 14 16
TIME ( SEC )

FIG. 14.—Dimensionless Response Time Histories for Long Period System; Dis-
placement U, Rotation $, and Edge Displacement 8: Taft Excitation (Tv = 1.0 sec,
e = 0.25)

397

J. Struct. Eng. 1986.112:383-400.


CONCLUSIONS

Inelastic earthquake responses of a one-way, torsionally coupled sys-


Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

tem subjected to two types of ground motion excitations were studied


in this paper. The effects of eccentricity, yield strength, uncoupled tor-
sional-to-lateral frequency ratio, and uncoupled lateral period on the re-
sponse of the system were examined. The following conclusions can be
drawn based on this study:

1. The effect of asymmetry on the element ductility demand and on


the edge displacement is most pronounced for stiff systems with yield
strength considerably reduced from the elastic strength demand. The
ductility demand on the critical element in the eccentric model can be
up to about three times that for the corresponding symmetrical system.
Since the ductility demand for stiff symmetrical structures with low yield
resistances is already substantial, the ductility demands for the corre-
sponding eccentric systems become very large.
2. Asymmetry affects the right edge displacement more than it affects
the element ductility demand, especially for torsionally flexible struc-
tures.
3. The uncoupled torsional-to-lateral frequency ratio does not appear
to be a critical parameter for estimating ductility demand. The edge dis-
placement can be more sensitive to this ratio, especially for stiff struc-
tures with low yield levels.
4. Ground motions containing large and long duration pulses, such
as the 1977 Romanian NS component record can lead to special problems
for structures that are designed with a largely reduced strength from the
elastic strength demand. These types of ground motion do not have
spectra similar in shape to the codified design spectrum. As a result,
design strength based on the code provision may not match the strength
demand for this type of ground motion. It is shown that stiff eccentric
structures are particularly vulnerable to such high ductility demand.
Therefore, a design strength of stiff eccentric buildings should not be
reduced from the elastic strength demand.

ACKNOWLEDGMENTS

The writers wish to acknowledge the support of the Natural Sciences


and Engineering Research Council of Canada for the work presented in
this paper.

APPENDIX I.—REFERENCES

1. Ayre, R. S., "Interconnection of Translational and Torsional Vibrations in


Buildings," Bulletin of the Seismological Society of America, Vol. 28, No. 2, Apr.,
1938, pp. 89-130.
2. Bertero, V. V., Mahin, S. A., and Herrera, R. A., "Aseismic Design Impli-
cations of Near-Fault San Fernando Earthquake Records," Journal of Earth-
quake Engineering and Structural Dynamics, Vol. 6, No. 1, Jan.-Feb., 1978, pp.
31-42.
3. Dempsey, K. M., and Tso, W. IC, "An Alternative Path to Seismic Torsional
398

J. Struct. Eng. 1986.112:383-400.


Provisions," Journal of Soil Dynamics and Earthquake Engineering, Vol. 1, No.
1, Jan., 1982, pp. 3-10.
4. Irvine, H. M , and Kountouris, G. E., "Inelastic Seismic Response of a Tor-
sionally Unbalanced Single-Story Building Model," Report No. R79-31, Dept.
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

of Civ. Engrg., Mass. Inst, of Technology, Cambridge, MA, July, 1979.


5. Kan, C. L., and Chopra, A. K., "Torsional Coupling and Earthquake Re-'
sponse of Simple Elastic and Inelastic Systems," Journal of the Structural Di-
vision, ASCE, Vol. 107, No. ST8, Aug., 1981, pp. 1569-1588.
6. Kan, C. L., and Chopra, A. K., "Simple Model for Earthquake Response
Studies of Torsionally Coupled Buildings," Journal of the Engineering Mechan-
ics Division, ASCE, Vol. 107, No. EM5, Oct., 1981, pp. 935-951.
7. Newmark, N. M., and Hall, W. J., Earthquake Spectra and Design, Earthquake
Engrg. Research Inst., Berkeley, CA, 1982.
8. Tentative Provisions for the Development of Seismic Regulations for Buildings, ATC
3-06, Applied Technology Council, Palo Alto, CA, 1978.
9. Tso, W. K., and Dempsey, K, M., "Seismic Torsional Provisions for Dynamic
Eccentricity," Journal of Earthquake Engineering and Structural Dynamics, Vol.
8, No. 3, May-June, 1980, pp. 275-289.
10. Tso, W. K., and Sadek, A. W., "Inelastic Seismic Responses of Simple Ec-
centric Structures," Journal of Earthquake Engineering and Structural Dynamics,
Vol. 13, No. 2, 1985, pp. 255-269.
11. Tso, W. K., and Sadek, A. W., "Inelastic Response of Eccentric Buildings
Subjected to Bi-Directional Ground Motions," Proceedings of the Eighth World
Conference on Earthquake Engineering, San Francisco, CA, July, 1984, Vol. IV,
pp. 203-210.
12. Yamazaki, Y., "Inelastic Torsional Response of Structures Subjected to Earth-
quake Ground Motions," Report No. UCB/EERC-80/07, Earthquake Engrg.
Research Center, Univ. of California, Berkeley, CA, Apr., 1980.

APPENDIX II.—NOTATION

The following symbols are used in this paper:

A = peak ground acceleration,


a* = characteristic acceleration,
e,e,e* = eccentricity, and its dimensionless forms;
F,F* = restoring force, and its dimensionless form;
F0 = yield strength of symmetric system.
Kv = elastic lateral stiffness of the system,
Ke = elastic torsional stiffness of the system referred to center
of stiffness,
K(t) = tangent stiffness matrix,
koi,kj(t) = elastic and tangent stiffnesses of rth resisting element,
m = mass of deck,
R = modification factor,
r = radius of gyration about center of mass,
S(t) = current position of center of stiffness relative to its initial
position,
T,T* = restoring torque, a n d its dimensionless form;
Tv = uncoupled lateral period,
U = dimensionless lateral displacement at center of stiffness,
v = lateral displacement at center of stiffness,
v0 = yield deformation,
8 = dimensionless edge displacement,

399

J. Struct. Eng. 1986.112:383-400.


6 = rotation,
p = radius of gyration about center of stiffness,
3> = dimensionlesa rotation,
Downloaded from ascelibrary.org by UNIVERSITY OF REGINA LIBRARY on 09/29/13. Copyright ASCE. For personal use only; all rights reserved.

Cl,Cl(t) = elastic and instantaneous frequency ratios,


co„, coB(t) = elastic and instantaneous lateral frequencies, and
w9 = uncoupled elastic torsional frequency.

400

J. Struct. Eng. 1986.112:383-400.

You might also like