Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2002; 31:1131–1150 (DOI: 10.1002/eqe.152)

Seismic response of self-centring hysteretic


SDOF systems

Constantin Christopoulos, Andre Filiatrault∗;† and Bryan Folz


Department of Structural Engineering; University of California at San Diego; 9500 Gilman Drive, Mail Code
0085; La Jolla; CA 92093; U.S.A.

SUMMARY
The seismic response of single-degree-of-freedom (SDOF) systems incorporating ag-shaped hysteretic
structural behaviour, with self-centring capability, is investigated numerically. For a SDOF system with
a given initial period and strength level, the ag-shaped hysteretic behaviour is fully dened by a post-
yielding stiness parameter and an energy-dissipation parameter. A comprehensive parametric study
was conducted to determine the inuence of these parameters on SDOF structural response, in terms of
displacement ductility, absolute acceleration and absorbed energy. This parametric study was conducted
using an ensemble of 20 historical earthquake records corresponding to ordinary ground motions having
a probability of exceedence of 10% in 50 years, in California. The responses of the ag-shaped hysteretic
SDOF systems are compared against the responses of similar bilinear elasto-plastic hysteretic SDOF
systems. In this study the elasto-plastic hysteretic SDOF systems are assigned parameters representative
of steel moment resisting frames (MRFs) with post-Northridge welded beam-to-column connections. In
turn, the ag-shaped hysteretic SDOF systems are representative of steel MRFs with newly proposed
post-tensioned energy-dissipating connections. Building structures with initial periods ranging from 0.1
to 2:0 s and having various strength levels are considered. It is shown that a ag-shaped hysteretic SDOF
system of equal or lesser strength can always be found to match or better the response of an elasto-
plastic hysteretic SDOF system in terms of displacement ductility and without incurring any residual
drift from the seismic event. Copyright ? 2002 John Wiley & Sons, Ltd.

KEY WORDS: hysteretic models; SDOF systems; non-linear analysis; self-centring systems; seismic
response

1. INTRODUCTION

Following the unexpected failures of beam-to-column connections in more than one hundred
steel moment-resisting frames (MRFs) during the 1994 Northridge, California earthquake, a
comprehensive research program—the SAC Joint Venture—was initiated in the United States
to investigate and remediate the causes of these failures [1]. It was concluded from the
∗ Correspondence to: Andre Filiatrault, Department of Structural Engineering, University of California at San Diego,
9500 Gilman Drive, Mail Code 0085, La Jolla, CA 92093, U.S.A.
† E-mail: aliatrault@ucsd.edu

Received 23 April 2001


Revised 8 September 2001
Copyright ? 2002 John Wiley & Sons, Ltd. Accepted 10 September 2001
1132 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Figure 1. Concept of PTED steel connection: (a) steel frame with PTED connections; (b) deformed
conguration of exterior PTED connection.

investigation phase of this project that the major cause of these failures was the unexpectedly
low rotational capacity of beam-to-column welded connections. In the remediation phase of
the SAC project, studies on new construction led to a better understanding of the cyclic
behaviour of welded and bolted steel moment-resisting connections and to the development
of more stringent welding practices [2]. However, even with these enhanced requirements,
inelastic deformations as well as residual drifts are expected to occur in steel MRFs under
seismic loading.
In parallel with the post-Northridge steel research, moment-resisting connections using post-
tensioning concepts were developed for precast concrete construction [3]. A series of innova-
tive beam-to-column connections [4], combining self-centring characteristics as well as energy
dissipation, were proposed. It was demonstrated that the performance of these connections was
excellent under simulated seismic loading. The most signicant characteristic of these connec-
tions was their capacity to ensure small residual drifts, through self-centring capabilities, even
when signicant inelastic transient deformations were mobilized during the seismic response.
Recently, this post-tensioning technology has been extended to steel MRFs [5; 6]. Exper-
imental and numerical results obtained from these studies show that these post-tensioned
connections are capable of achieving stiness and strength characteristics comparable to tradi-
tional welded moment-resisting connections. In addition, these connections can be tailored to
provide a specied amount of energy dissipation. This structural behaviour can be achieved
without introducing inelastic deformations in the beam or column and without residual drift.
The concept for the particular post-tensioned energy dissipating (PTED) connection devel-
oped by the authors [6] is illustrated in Figure 1. This PTED connection incorporates high
strength steel post-tensioned (PT) bars designed to remain elastic during the seismic response,
and conned energy-dissipating (ED) bars designed to yield both in tension and compression.
Figure 2 shows a moment-rotation relationship obtained experimentally from a large scale
PTED connection [6]. The self-centring capacity and energy dissipation characteristics of the
connection are evident. Figure 3 shows an idealization of the ag-shaped hysteretic behaviour
of a PTED connection, which is easily amenable to numerical modelling. The overall response

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1133

Figure 2. Experimental moment–rotation curve of PTED connection (after Reference [6]).

Figure 3. Idealized hysteretic behaviour of the PTED beam-to-column connection: (a) contribution
of PT Bars; (b) contribution of ED bars; and (c) moment–rotation relationship of PTED connection.

of the connection can be decomposed into the non-linear elastic contribution from the PT-
bars and the bilinear elasto-plastic hysteretic contribution from the ED-bars. It is of interest to
note that this ag-shaped hysteretic response has also been achieved using specialized energy
dissipating dampers or materials [7; 8].
To date, there is limited information on the non-linear dynamic response of hysteretic self-
centring systems under seismic loading. The main objective of this paper is to shed some light
on this issue by investigating numerically the inelastic response of single degree-of-freedom
(SDOF) ag-shaped hysteretic systems, under code prescribed levels of seismic input. For
this purpose, an ensemble of 20 historical records representative of ordinary ground motions
having a probability of exceedence of 10% in 50 years in California was considered. Although
the parametric study that is conducted focuses on structural systems that display a ag-shaped
hysteretic response similar in form to that developed by a PTED framing system, the results
obtained can also be applicable in part to the other self-centring hysteretic systems cited above.
Finally, to assess the advantages and disadvantages of using this new type of framing system,
the dynamic responses of SDOF systems exhibiting ag-shaped hysteresis are compared to
the responses of SDOF systems exhibiting bilinear elasto-plastic hysteresis, typical of post-
Northridge steel MRFs incorporating welded fully restrained moment-resisting connections.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1134 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

2. ANALYSIS PROCEDURE

2.1. Hysteretic models

Two hysteretic models are considered in this study: a bilinear elasto-plastic model and a ag-
shaped model. The bilinear elasto-plastic hysteretic model is representative of the behaviour
of steel MRFs incorporating post-Northridge welded beam-to-column fully restrained moment-
resisting connections. It is assumed that following the new recommendations on welded beam-
to-column connections [2; 9], a large number of plastic rotation cycles can be achieved without
any fracture of the welds. A post-yielding stiness of 0.02 of the initial stiness is also
assumed. The idealized hysteretic force–displacement relationship of a system incorporating
these types of welded connections is shown in Figure 4(a). Note that this idealization is an
upper bound of the actual response of steel MRFs considering that strength degradation is
expected under cyclic loading.
The ag-shaped hysteretic model considered is representative of the behaviour of steel MRFs
incorporating PTED connections both at all beam-to-column connections and at the base of
each column. Figure 4(b) shows the idealized hysteretic force–displacement relationship of a
system incorporating these PTED connections. Associated with this hysteretic model are two
independent response parameters  and . In this study the post-yielding stiness coecient
, expressed as a fraction of the initial stiness, ranges in value from 0.02 to 0.35. The
coecient  reects the energy dissipation capacity of the system. A lower bound of  = 0:0
produces a piecewise non-linear elastic system. An upper bound of  = 1:0 is required to
ensure the self-centring capability of the hysteretic model.

2.2. Normalized equations of motion

The equation of motion of a SDOF system under seismic input is given by

mx + cẋ + F(x) = −mxg (1)

Figure 4. Idealized pseudo force–displacement relationships: (a) system incorporating welded


connections; and (b) system incorporating PTED connections.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1135

where m is the mass of the system, c is the viscous damping coecient and F(x) is the
non-linear restoring force dened by the hysteretic model of the system. The displacement,
velocity and acceleration of the system, relative to the ground are denoted by x, ẋ and x ,
respectively. The ground acceleration is designated by xg .
Two key parameters that can be used in dening the dynamic response of a non-linear
SDOF system are the initial period T0 and the strength ratio :


T0 = 2 m=k 0 (2)

Fy
= (3)
mg

where k 0 is the initial stiness of the system, Fy is the yield force and g is the acceleration
of gravity.
Using Equation (2), Equation (1) can be rewritten as
   2
2 2
x + 20 ẋ + ˜ = −xg
f(x) (4)
T0 T0
with 0 denoting the initial fraction of critical damping of the system,
c
0 = √ (5)
2 k0m

and f(x)
˜ representing the non-linear pseudo-restoring of the system:

˜ = F(x)
f(x) (6)
k0
The yield displacement xy of the system is given by
Fy
xy = = f˜y (7)
k0
as shown in Figure 4. Using Equations (2) and (3), the yield displacement xy can be expressed
in terms of the key parameters of the system T0 and :
T02 g
xy = (8)
42
With this formulation, for a specied level of critical damping 0 , initial period T0 and strength
level  the SDOF is completely dened for the case when the restoring force–displacement
relationship is bilinear elasto-plastic (Figure 4(a)) and requires only the additional parameters
 and  to be assigned if the restoring force–displacement relationship exhibits a ag-shaped
hysteresis (Figure 4(b)).
For the time-history dynamic analyses performed in this study, the normalized non-linear
equation of motion given by Equation (4) is integrated using the Newmark constant average
acceleration scheme. The analyses were continued for 10 s of zero ground acceleration at the

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1136 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

end of each record to allow the system to oscillate under viscously damped free vibrations in
order to return to rest.

2.3. Energy balance

The energy balance at time t for the normalized equation of motion can be written as

Ek (t) + Ed (t) + Es (t) = Ein (9)

where Ek (t); Ed (t), Es (t) and Ein are the kinetic energy at time t, the energy dissipated by
viscous damping up to time t, the strain energy at time t (recoverable elastic and dissipated
hysteretic) and the relative seismic input energy, respectively. These energy quantities can be
dened as follows:

Ek (t) = 12 ẋ(t)2
   x(t)
2
Ed (t) = 20 ẋ(t) d x (10)
T0 0
 2  x(t)
2
Es (t) = ˜ dx
f(t)
T0 0
 x(t)
Ein = − x g (t) d x
0

Equations (9) and (10) determine how the seismic input energy is distributed in the system
over time, as well as allowing for a check on the accuracy of the time integration scheme.

2.4. System response indices

The inelastic response of SDOF systems under seismic input can be characterized in part by
the following normalized non-dimensional response indices:

(i) The maximum displacement ductility  :


max06t6tD |x(t)|
 = (11)
xy
where tD is the total duration of the seismic input.
In performance-based earthquake engineering, the maximum inelastic displacement is
one of the primary response indices to determine both the structural and non-structural
damage to buildings under seismic loading [10; 11].
(ii) The normalized absolute maximum acceleration amax :
max06t6tD |x (t) + x g (t)|
amax = (12)
g

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1137

This index is a measure of the damage potential to acceleration-sensitive non-structural


elements, as well as an indicator of potential injury to occupants during an earthquake
event. In addition, this response index is a direct indicator of the force level induced
into the system by the seismic input.
(iii) The normalized maximum absorbed energy Eabs :
max06t6tD |Es (t)|
Eabs = (13)
xy mg
This index is a measure of potential structural damage including duration eects.
(iv) The normalized residual displacement xres :
|x(tD )|
xres = (14)
xy
This index is an indicator of the structural damage sustained after an earthquake and of
the extent of repair costs. Residual displacements are only computed for the bilinear
elasto-plastic hysteretic model (see Figure 4(a)). The ag-shaped hysteretic model
(see Figure 4(b)), by virtue of its self-centring capabilities, does not have residual
displacements.

2.5. Ground motions considered in parametric study

An ensemble of 20 historical strong ground motion records from California representative of


ordinary earthquakes having a probability of exceedence of 10% in 50 years are used in this
study [12]. These records are free of any forward directivity eects (near-fault eects). All
records were recorded on soil types C or D, and were generated by earthquakes of moment
magnitude Mw ranging from 6:7 to 7:3. The hypocentral distance for these records range
between 13 and 25 km. Table I gives further details on the characteristics of these earthquake
records.
Following the method proposed in NEHERP Provisions for the seismic rehabilitation
of buildings [13], a 5% damped design elastic acceleration response spectrum for a seis-
mic zone 4 and a soil type C or D was constructed and used as the target spectrum.
Each of the 20 earthquake records was then scaled to minimize the square of the error
between its 5% damped response spectrum and the target NEHERP spectrum at ve pe-
riod values: T = 0:1; 0:25; 0:5; 1:0 and 2:0 s. The resulting scaling factors are listed in
Table I.
The mean and the envelopes of the maximum and minimum spectral values of the 20 scaled
records along with the NEHERP target spectrum are shown in Figure 5. A good match is
obtained between the mean spectral values and the target spectrum in the range of periods
of interest (0:1–2:0 s). However, the envelopes of maximum and minimum spectral values
indicate the large variability that exists between the records. Table I also lists the scaled peak
ground accelerations (PGA), and scaled peak ground velocities (PGV). The mean value of
the PGA of the 20 scaled records is 0:43 g, which is very close to the eective peak accel-
eration Ca = 0:40 g specied in the NEHERP provisions for a seismic zone 4 and soil types
C and D.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1138

Table I. Characteristics of ground motions considered (after Reference [12]).

Earthquake event Year Mw Station Rclosest Soil type Duration Scaling Scaled Scaled
(km) (NEHRP) (s) factor PGA (g) PGV (cm=s)
Superstition Hills 1987 6.7 Brawley 18.2 D 22.0 2.7 0.313 46.44
Superstition Hills 1987 6.7 El Centro Imp. Co. Cent. 13.9 D 40.0 1.9 0.490 77.71
Superstition Hills 1987 6.7 Plaster City 21.0 D 22.2 2.2 0.409 45.32

Copyright ? 2002 John Wiley & Sons, Ltd.


Northridge 1994 6.7 Beverly Hills 14145 Mulhol 19.6 C 30.0 0.9 0.374 53.10
Northridge 1994 6.7 Canoga Park—Topanga Can 15.8 D 25.0 1.2 0.427 38.52
Northridge 1994 6.7 Glendale—Las Palmas 25.4 D 30.0 1.1 0.393 13.53
Northridge 1994 6.7 LA—Hollywood Stor FF 25.5 D 40.0 1.9 0.439 34.77
Northridge 1994 6.7 LA—N Faring Rd 23.9 D 30.0 2.2 0.601 34.76
Northridge 1994 6.7 N. Hollywood—Coldwater Can 14.6 C 21.9 1.7 0.461 37.74
Northridge 1994 6.7 Sunland—Mt Gleason Ave 17.7 C 30.0 2.2 0.345 31.90
Loma Prieta 1989 6.9 Capitola 14.5 D 40.0 0.9 0.476 32.85
Loma Prieta 1989 6.9 Gilroy Array # 3 14.4 D 39.9 0.7 0.386 24.99
Loma Prieta 1989 6.9 Gilroy Array # 4 16.1 D 40.0 1.3 0.542 50.44
Loma Prieta 1989 6.9 Gilroy Array # 7 24.2 D 40.0 2.0 0.452 32.80
Loma Prieta 1989 6.9 Hollister Di. Array 25.8 D 39.6 1.3 0.363 46.28
Loma Prieta 1989 6.9 Saratoga—W Valley Coll. 13.7 C 40.0 1.4 0.465 86.10
Cape Mendocino 1992 7.1 Fortuna Fortuna Blvd 23.6 C 44.0 3.8 0.441 114.00
Cape Mendocino 1992 7.1 Rio Dell Overpass—FF 18.5 C 36.0 1.2 0.462 52.68
C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Landers 1992 7.3 Desert Hot Springs 23.3 C 50.0 2.7 0.416 56.43
Landers 1992 7.3 Yermo Fire Station 24.9 D 44.0 2.2 0.334 65.34

Earthquake Engng Struct. Dyn. 2002; 31:1131–1150


SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1139

Figure 5. Elastic response spectra of 20 scaled accelerograms.

3. PARAMETRIC STUDY

3.1. Range of key system and hysteretic parameters

The parametric study presented herein focuses on the seismic response of steel MRFs ranging
in number of stories from one to twenty. From the seismic provisions of the 1997 edition of
the uniform building code (UBC) [14], the natural period range of these structures can be
estimated by the equation:
T0 = Ct hn3=4 (15)
where Ct = 0:0853 for steel MRFs, and where hn is the height of the building in meters.
Using Equation (15) and assuming a storey height of 3:4 m, the range of periods T0 for a
single storey and for a 20 storey building, respectively, is:
0:2 s6T0 62:0 s (16)
The strength factor  which corresponds to the ratio Vy =W in the 1997 UBC [14] is dened
as
Vy Cv I
= = (17)
W RT0
where Vy is the design base shear, W is the weight of the structure, I , taken as 1, is the
importance factor, Cv is computed as 0:64 for a Zone 4 with soil type D and R is the force
reduction factor ranging from 4.5 to 8.5 for an ordinary steel MRF and a special steel MRF,
respectively. A lower bound for Equation (17) for seismic Zone 4 is
Vy
=¿0:11Ca I (18)
W
where Ca is computed as 0.44 for a seismic Zone 4 and soil type D.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1140 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Table II. SDOF system values used in parametric study.


T0 (s)  ∗ ∗
1.00 0.05 0.02 0.00
0.25 0.10 0.10 0.30
0.50 0.20 0.20 0.60
1.00 0.30 0.35 1.00
1.50 0.50
2.00 1.00

For ag-shaped hysteretic model only.

Substituting the lower and upper period bounds into Equation (17) and verifying Equation
(18), the range of strength factors  is found to be

0:05660:71 (19)

In addition, the ag-shaped hysteretic model requires the specication of the parameters 
and  to completely dene the system. Table II lists the complete set of parameters T0 , ,
 and  considered in this parametric study. These values result in 576 dierent ag-shaped
hysteretic systems. The resulting force–deection relationships of the ag-shaped hysteretic
systems are illustrated qualitatively in Table III for the specied range of values of  and
. Throughout this study the fraction of critical damping 0 is taken as 0.05 for all SDOF
systems.

3.2. Non-linear dynamic response of ag-shaped hysteretic systems

Mean values over the ensemble of earthquakes of the displacement ductility  are shown
in Figure 6 for all ag-shaped hysteretic systems considered. For all values of  and , the
mean displacement ductility generally increases for decreasing values of initial period T0 and
decreasing values of strength ratio . The mean displacement ductility is reduced in all cases
for increasing values of  and . This reduction of  with increasing values of  and  is
more signicant for low period structures (T0 61:0 s) and for structures with lower strength
ratios (60:3) where the values of  are also the largest.
Mean values over the ensemble of earthquakes of the maximum absolute acceleration amax
are shown in Figure 7. Note that when  = 1:0 the SDOF oscillator responds in the elas-
tic range for most of the earthquake records considered and for all values of  and .
For this case, the plot of the mean maximum absolute acceleration versus period tends
towards the elastic response spectrum. The mean maximum absolute acceleration is insen-
sitive to the value of  as seen in Figure 7. When  is increased, the accelerations of
systems with lower values of  are increased. For increasing values of ; amax for all
values of  tends towards the elastic response spectrum. For small values of initial pe-
riod (T0 60:5 s); amax remains high even when the strength ratio  is reduced. This is due
to the combination of non-zero post-yielding stiness  with large values of displacement
ductility.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1141

Table III. Qualitative force–deection relationships of ag-shaped hysteretic systems for all values
of  and  considered in parametric study.
Energy-dissipation coecient, 
Post-yielding stiness,  0.0 0.30 0.60 1.0

0:02

0:10

0:20

0:35

Mean values over the ensemble of earthquakes of the absorbed energy E abs are shown
in Figure 8. In general, the energy absorbed increases for decreasing initial period and for
decreasing strength ratios. This trend is similar to that observed for the displacement ductility
(see Figure 6). However for smaller values of , the absorbed energy (Figure 8) does not
increase as much as the displacement ductility (Figure 6).
The mean absorbed energy E abs is insensitive to increasing values of , but highly dependent
on the value of . In general, the mean absorbed energy E abs doubles when the value of 
changes from 0 to 1. This increased absorbed energy indicates a higher amount of hysteretic
damping but also larger cumulative inelastic excursions in the system.

3.3. Non-linear dynamic response of bilinear elasto-plastic hysteretic systems

Figure 9 collectively shows mean values, over the ensemble of earthquakes, of displacement
ductility  , normalized maximum absolute acceleration amax , normalized absorbed energy E abs
and normalized residual displacement xres for the bilinear elasto-plastic hysteretic systems.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1142 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Figure 6. Mean displacement ductility for ag-shaped hysteretic systems.

Similar to the ag-shaped hysteretic systems, the mean displacement ductility  increases
for decreasing values of initial period T0 and decreasing values of the strength ratio , as
shown in Figure 9(a).
Mean maximum absolute accelerations amax for the case where the strength ratio is taken
as 1 as shown in Figure 9(b), are similar to those of the elastic response spectrum. The
maximum accelerations decrease for decreasing values of the strength ratio. Similar to the ag-
shaped hysteretic systems, the accelerations for systems with short initial periods (T0 = 0:1s) do
not decrease when the strength ratio is decreased (60:3). This is also due to the combination
of the post-yielding stiness of the bilinear elasto-plastic hysteretic systems,  = 0:02, and large
displacement ductility values for these systems.
The mean absorbed energy E abs , as shown in Figure 9(c), increases for decreasing values
of initial period T0 and decreasing values of strength ratio  similarly to the displacement
ductility. As noted for the ag-shaped hysteretic systems, for lower strength systems, lowering

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1143

Figure 7. Mean normalized maximum accelerations for ag-shaped hysteretic systems.

the strength ratio  causes a larger increase in the displacement ductility than in the absorbed
energy.
The residual displacements shown in Figure 9(d) increase for decreasing values of initial
period T0 and decreasing values of strength ratio . For the highest strength ratio ( = 1:0),
there are no residual displacements at the end of the earthquake. For lower strength values
(60:3), residual displacements are more pronounced and more dependent upon the initial
period.

3.4. Comparative response of the ag-shaped and elasto-plastic hysteretic systems

The response of ag-shaped hysteretic and bilinear elasto-plastic hysteretic systems are qual-
itatively very similar as seen by comparing Figures 6–8 with Figure 9. The following three
observations can be made on the comparative response of these two types of hysteretic

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1144 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Figure 8. Mean normalized absorbed energy for ag-shaped hysteretic systems.

systems:
(i) For each bilinear elasto-plastic system, there is at least one ag-shaped hysteretic
system of similar initial period and strength ratio that can achieve equal or smaller
displacement ductility. In general, the intermediate values of  and  are sucient to
achieve this.
(ii) The maximum absolute accelerations are similar between these two hysteretic models
for low values of . For larger values of , maximum accelerations are larger for the
ag-shaped hysteretic systems, especially for systems with lower strength ratios.
(iii) The energy absorbed is in general signicantly larger for the bilinear elasto-plastic
hysteretic systems than the ag-shaped hysteretic systems, especially for low values
of .

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1145

Figure 9. Response of bilinear elasto-plastic hysteretic systems: (a) displacement ductility; (b) normal-
ized maximum accelerations; (c) normalized absorbed energy; and (d) normalized residual displacements.

To further compare the response of these two hysteretic models, three particular systems
are considered. The rst system is characterized by an initial natural period T0 = 0:25 s and
a strength ratio  = 0:5, representing a one-storey steel MRF structure. The second system
is characterized by an initial natural period T0 = 1:0 s and a strength ratio  = 0:1, represent-
ing a 7-storey steel MRF structure. The last system considered is characterized by an initial
natural period T0 = 2:0 s and a strength ratio  = 0:05, representing a 20-storey steel MRF
structure.
For each of these structural congurations, an elasto-plastic hysteretic system (EP) and
three ag-shaped hysteretic systems (FS) are dened and considered for comparative pur-
poses. The FS systems have the same initial period T0 and strength ratio  as the corre-
sponding EP system. As listed in Table IV, each FS system has a dierent combination
of post-yielding stiness coecient  and energy dissipating coecient . The mean re-
sponse values over the ensemble of earthquakes for these systems are presented in Table
IV. The maximum absolute accelerations are increased for increasing values of . The en-
ergy absorbed, also as discussed earlier, is larger for the elasto-plastic hysteretic systems and
is also increased for the ag-shaped hysteretic systems for larger values of . Finally, un-
like the bilinear elasto-plastic systems, the ag-shaped hysteretic systems sustain no residual
displacements.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1146 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Table IV. Comparative response of three systems exhibiting bilinear elasto-plastic hysteresis (EP)
and ag-shaped hysteresis (FS).

 amax Eabs xres


System 1: T0 = 0:25 s;  = 0:50
EP 2.28 0.55 2.39 0.61
FS ( = 0:10;  = 1:0) 2.10 0.56 2.46 0.00
FS ( = 0:20;  = 0:6) 2.26 0.64 1.92 0.00
FS ( = 0:35;  = 0:3) 2.27 0.73 1.48 0.00

System 2: T0 = 1:0 s;  = 0:10


EP 6.19 0.13 2.71 1.10
FS ( = 0:02;  = 1:0) 6.07 0.13 2.27 0.00
FS ( = 0:20;  = 0:6) 6.45 0.22 2.00 0.00
FS ( = 0:35;  = 0:6) 6.19 0.29 1.98 0.00

System 3: T0 = 2:0 s;  = 0:05


EP 4.41 0.06 0.83 1.25
FS ( = 0:02;  = 1:0) 4.78 0.06 0.70 0.00
FS ( = 0:20;  = 0:6) 4.50 0.09 0.55 0.00
FS ( = 0:35;  = 1:0) 4.21 0.11 0.69 0.00

Figure 10. Loma Prieta record (Hollister dierential array) scaled at 130%: (a) accelerogram; and
(b) elastic response spectrum for 5% damping for accelerogram and for ensemble of earthquake records.

3.5. Examples of time-history analyses

To this point, results have been given in terms of non-dimensional mean value response
quantities over the ensemble of 20 earthquake records. To further compare the two hysteretic
models, and to provide some insight in their response over time, a series of ve structural
systems were subjected to the 1989 Loma Prieta earthquake recorded at the Hollister Dier-
ential Array and scaled at 130% of its amplitude (see Table I). The scaled accelerogram for
this record is presented in Figure 10(a). As shown in Figure 10(b), the 5% damped elastic
response spectrum of the scaled record is in good agreement with the mean spectrum for
the ensemble of 20 records used in the parametric study. All systems considered have an
initial period of T0 = 1:0 s and a mass of 4:0 kN s2 = mm. The resulting initial stiness k 0 is

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1147

Table V. Response of SDOF systems with T0 = 1:0 s, m = 4 kN s2 = mm and k0 = 157:9 kN= mm under
130% of Loma Prieta record.
System    y Fy  (max) A (max) Eabs  (res)
(mm) (kN) (mm) (g) (kN mm) (mm)

EP 0.02 — 0.10 24.85 3924 124.70 0.13 2:0 × 106 22.95


FS1 0.25 0.30 0.10 24.85 3924 110.72 0.20 7:4 × 105 0.00
FS2 0.25 0.30 0.07 17.40 2747 117.88 0.17 8:3 × 105 0.00
FS3 0.15 0.50 0.10 24.85 3924 110.00 0.16 1:1 × 106 0.00
FS4 0.15 0.50 0.07 17.40 2747 115.22 0.14 1:1 × 106 0.00

157:9 kN= mm. The structural system so dened is a representation of a 7-storey steel MRF.
As shown in Figure 10, the peak ground acceleration is 0:36 g while the spectral acceleration
at a period of 1:0 s is 0:72 g.
The rst system designated by EP incorporates the elasto-plastic hysteretic model. The four
other systems designated by FS1 through FS4 utilize the ag-shaped hysteretic model. The
yield force for systems EP, FS1 and FS3 were set equal to 3924 kN which corresponds to
a strength ratio  = 0:1. The yield force of systems FS2 and FS4 were set equal to 2747 kN
which is equal to 70% of the yield force of systems EP, FS1 and FS3 and corresponds
to a strength ratio  = 0:07. As noted earlier, intermediate values of  and  result in dis-
placement ductility values for the ag-shaped hysteretic model that are similar to systems
with large values of  combined with low values of  and vice-versa. For systems FS1
and FS2,  and  were set to 0.25 and 0.30, respectively. For systems FS3 and FS4,
 and  were set to 0.15 and 0.50, respectively. The dening parameters for these ve
systems are summarized in Table V. Response values of maximum relative displacement
max , maximum absolute acceleration Amax , absorbed energy Eabs as well as residual dis-
placement res obtained from dynamic time-history analyses are presented in Table V. It is
noted that these response values obtained for the 130% Loma Prieta earthquake record are
in close agreement with the mean values obtained over the ensemble of earthquakes and
follow similar trends as discussed earlier. In all cases, all four ag-shaped hysteretic sys-
tems achieve smaller maximum displacements than the elasto-plastic system. Systems FS1
and FS2 have greater maximum accelerations and lower absorbed energy than systems FS3
and FS4.
Figure 11 shows the time-histories of displacement, acceleration, absorbed energy for the
EP and FS4 systems along with their force–displacement responses. Note that the elasto-plastic
system deforms inelastically primarily in one direction, while the FS4 system has a similar
amount of inelastic excursions in both directions. For the elasto-plastic system, the one-sided
inelastic deformations will in fact be accentuated by P-delta eects [15] that have not been
taken into account in this study. The FS4 system, with a strength ratio equal to 0.70 of the EP
system achieves a smaller maximum displacement, while the maximum absolute accelerations
are similar. The energy absorbed is considerably smaller for the FS4 system. Finally, unlike
the EP system that sustains a residual displacement of 23 mm, the FS4 system returns to its
initial zero position after the end of the earthquake.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1148 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

Figure 11. Comparative response of EP and FS4 systems under 130% of Loma Prieta record (Hollister
dierential array): (a) displacement time-history; (b) acceleration time-history; (c) absorbed energy
time-history; and (d) force–displacement response.

4. CONCLUSIONS

The seismic response of SDOF systems incorporating either a bilinear elasto-plastic hysteretic
model or a ag-shaped hysteretic model have been investigated and compared through time-
history dynamic analyses. All systems were subjected to an ensemble of 20 historical records
representative of ordinary ground motions having a probability of exceedence of 10% in 50
years in California. For a SDOF system with a given initial period and strength level, the
incorporated ag-shaped hysteretic model is dened through a post-yielding parameter  and
energy-dissipation parameter . These two independent parameters allow for exibility in
tailoring the response of this type of SDOF system. A comprehensive parametric study was
conducted to determine the inuence of these parameters on SDOF structural response, in
terms of displacement ductility, absolute acceleration and absorbed energy. It was found that
reduced displacement ductility in systems with short initial periods and low strength levels
was most eectively achieved by increasing the value of  as opposed to increasing the value

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
SELF-CENTRING HYSTERETIC SDOF SYSTEMS 1149

of . The reverse is true for long period, high strength systems: increasing  is more eective
than increasing .
It was also shown that the seismic response of ag-shaped hysteretic systems was qual-
itatively similar to the elasto-plastic hysteretic systems. In addition, by adjusting the values
of  and , a ag-shaped hysteretic system could be made to quantitatively match or better
the response of an elasto-plastic system in terms of displacement ductility. Values of  and
 to achieve this were not unique. In general, this match in performance can be realized
using intermediate values of  and . Such values are physically achievable using the newly
proposed post-tensioned energy-dissipating (PTED) connections.
With respect to absolute acceleration, the ag-shaped hysteretic system tends to produce
higher values than the comparable elasto-plastic system. The greatest dierence is seen with
higher values of . Also, the absorbed energy by the ag-shape hysteretic system is always less
than the comparable elasto-plastic system. However, the importance of this response index for
steel structures incorporating PTED connections is minimal since cumulative damage is limited
to the replaceable energy-dissipating bars within the beam-to-column and base connections.
Finally, residual drifts occurred in all of the elasto-plastic hysteretic SDOF systems
considered. Residual drifts were largest in systems with low strength and short periods. In all
of the ag-shaped hysteretic SDOF systems there was no residual drift due to the self-centring
capability of the force–displacement model.

ACKNOWLEDGEMENT

The nancial assistance of the Dean’s oce of the Irwin and Joan Jacobs School of Engineering at
the University of California, San Diego in support of this study is greatly appreciated.

REFERENCES

1. SAC. Proceedings of the invitational workshop on steel seismic issues. SAC Report No. 94-01, Sacramento,
CA, 1994.
2. FEMA. Recommended seismic design criteria for new steel moment-frame buildings. FEMA No. 350. Federal
Emergency Management Agency, Washington, DC, 2000.
3. Priestley MJN. Seismic design philosophy for precast concrete frames. Structural Engineering International
1996; 6(1):25– 31.
4. Stanton JF, Stone WC, Cheok GS. A hybrid reinforced frame for seismic regions. PCI Journal 1997; 42(2):
20 – 32.
5. Ricles MJ, Sause R, Garlock MM, Zhao C. Postensioned seismic-resistant connections for steel frames. Journal
of Structural Engineering, ASCE 2001; 127(2):113 –121.
6. Christopoulos C, Filiatrault A, Uang C-M, Folz B. Post-tensioned energy dissipating connections for moment-
resisting steel frames. Journal of Structural Engineering, ASCE 2002, accepted.
7. Witting PR, Cozzarelli FA. Shape memory structural dampers: material properties, design and seismic testing.
NCEER Report No. 92-0013. National Center for Earthquake Engineering Research, Bualo, NY, 1992.
8. Filiatrault A, Tremblay R, Kar R. Performance Evaluation of Friction Spring Seismic Damper. Journal of
Structural Engineering, ASCE 2000; 123(4):491– 499.
9. Gross JL, Engelhardt MD, Uang C-M, Kasai K, Iwankiw NR. Modication of Existing Welded Steel Moment
Frame Connections for Seismic Resistance. American Institute of Steel Construction, 1999.
10. Moehle JP. Displacement-based design of RC structures subjected to earthquakes. Earthquake Spectra 1992;
8(3):403 – 428.
11. Priestley MJN. Displacement-based seismic assessment of existing reinforced concrete buildings. Bulletin of the
New Zealand National Society for Earthquake Engineering 1996; 29(4):256 –272.
12. Krawinkler H, Parisi F, Ibarra L, Ayoub A, Medina AR. Development of a testing protocol for wood frame
structures. CUREE Report No. W-02, Richmond, CA, 2000.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150
1150 C. CHRISTOPOULOS, A. FILIATRAULT AND B. FOLZ

13. Federal Emergency Management Agency. NEHERP provisions for the rehabilitation of buildings. FEMA No.
273 (Guidelines) and 274 (Commentary), Washington, DC, l997.
14. International Conference of Building Ocials. Uniform Building Code, vol. 2, Whittier, CA, 1997.
15. MacRae GA, Kawashima K. Post-earthquake residual displacements of bilinear oscillators. Earthquake
Engineering and Structural Dynamics 1997; 26:701–716.

Copyright ? 2002 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2002; 31:1131–1150

You might also like