Earthquake resistance frames with combination of rigid and

semi-rigid connections
Mohammad Razavi, Ali Abolmaali
Center for Structural Engineering Research and Simulation (CSER), Department of Civil Engineering, The University of Texas at Arlington, Arlington, TX 76019, United States
a b s t r a c t a r t i c l e i n f o
Article history:
Received 14 November 2013
Accepted 22 February 2014
Available online 18 March 2014
Keywords:
Hybrid steel frames
Semi-rigid connections
Earthquake engineering
SAC frames
Reliability analysis
Performance based design
The concept of the hybrid steel frame system, as it pertains to mixtures of fully-rigid and semi-rigid steel connec-
tions used in 20-story SAC frames, is presented herein. Several different patterns and locations of semi-rigid
connection replacements within the frame were examined in order to identify hybrid frames with the best
seismic performance. The effective connection stiffness was identied by performing a parametric study on the
initial stiffness of the semi-rigid connections. Then, the cyclic behavior of the connections with the most effective
stiffness was obtained, using nonlinear nite element analysis. Inelastic dynamic analyses were conductedonthe
proposed selected frames by subjecting them to Los Angeles earthquake records characterized as those with 2%
and 10%probabilities of exceedance infty years. The maximumstory drift for the hybrid frames was determined
and comparedwiththe life safety and collapse prevention performance limits, as recommended by FEMA356
[12]. The story drift andmember forces for the proposedhybrid frames were comparedwiththose of convention-
al SAC frames without semi-rigid connections. Finally, a reliability analysis, utilizing the collapse margin ratio
presented in FEMA P695, was performed to quantify and compare the collapse performance of the proposed
hybrid frames and conventional rigid frames.
Published by Elsevier Ltd.
1. Introduction
Seismic performance of structural systems has been at the
forefront of research for many years. Occurrences of more than 18
severe earthquakes, with magnitudes of more than 5.8 Richter in
the state of California in the 15-year period between 1979 and
1994, have been reported. The premature brittle failures of welded
connections after the Northridge earthquake of 1994 particularly
motivated researchers to look beyond the conventional design
philosophies
In the past fewdecades, several researchers have introduced new
design concepts and approaches to improve the seismic perfor-
mance of steel structures. These include, but are not limited to, the
introduction of more ductile connections and new lateral resistant
systems, including energy dissipating technologies such as base
isolators, frictional or visco-elastic dampers, and active control
elements.
An innovative seismic design method is the Performance Based
Plastic Design (PBPD) that was introduced and developed by Professor
Goel and his associates at the University of Michigan [14,17]. The
development of this method was in response to shortages in the current
seismic design codes on satisfying a performance objective in a direct
manner. In PBPD, the base shear is calculated by equating the work
done by pushing the structure to a predened target drift monotonically
to the work done by an equivalent elastic-perfectly plastic single degree
of freedom system. The idealized inelastic response spectra by [19] are
used in this study.
Development of the seismic eccentric braced frames (EBF), intro-
duced by Popov and Englehardt in 1988, was another attempt to
enhance the seismic behavior of steel frames. Well-designed EBFs,
constructed with shear links, provide high elastic stiffness and
strength under low to moderate ground motions, combined with
high ductility and energy dissipation capabilities in severe ground
motions.
Recently, Abolmaali et al. [4] introduced a newlateral resistant steel
moment frame referredto as a HybridSteel Frame. This system, which
was the foundation of the researchleading to the work that is being pre-
sented in this manuscript, is based on the concept of introducing energy
dissipating mechanisms in the structural frame systems by targeting
and replacing selected rigid connections withmore ductile energy dissi-
pating semi-rigid connections.
The proposed structural system enhances the seismic behavior of
the steel frames by taking advantage of the inelastic energy dissipative
semi-rigid connections, along with stiffer and higher strength rigid
connections. In the case of low intensity earthquakes, while rigid
Journal of Constructional Steel Research 98 (2014) 111
Corresponding author at: 425 NeddermanHall, 416 Yates St., Box 19308, Arlington, TX
76019, USA.
http://dx.doi.org/10.1016/j.jcsr.2014.02.006
0143-974X/Published by Elsevier Ltd.
Contents lists available at ScienceDirect
Journal of Constructional Steel Research
connections remain elastic, the semi-rigid connections absorb seismic en-
ergy and assist the systemin damping out the base motionaccelerationat
a faster rate. On the other hand, in the case of higher intensity earth-
quakes, this study shows that hybrid frames reduce the risk of collapse
when compared with its corresponding SMFs.
Current design codes have almost eliminated the partially restrained
connections in high seismic zones. However, there are several studies
that show high energy dissipating characteristics of the semi-rigid con-
nections with high stiffness and strength, as reported by Ackroyd and
Gerstle [5], Bjorhovde et al. [10], Astaneh et al. [6], Astaneh-Asl et al.
[8], Astaneh-Asl [7], Kukreti and Abolmaali [16], Abolmaali et al. [1],
Abolmaali et al. [2], and Abolmaali et al. [3].
Seismic behavior of SMFs with semi-rigid connections has been
investigated in several studies, both theoretically and experimentally
[18]. Excessive inter-story drift was a major concern for using semi-
rigid connections in steel frames. These studies showed that when
connection stiffness increases, base shear increases; however, inter-
story drift does not decrease proportionally.
Astaneh et al. [6] and Abolmaali et al. [3] studied the energy dissipa-
tion characteristics of different types of semi-rigid connections and
showed that they are capable of undergoing large inelastic rotation
(in excess of 0.05 rad), given that the connection is designed so that
the angle or plate yielding governs the behavior. In other words, if the
plate or angle thickness is relatively small compared to the bolt diame-
ter, bolt yielding and fractures are prevented, and plate yielding results
in a ductile connection behavior by undergoing large inelastic rotation.
In this research, The Los Angeles 20-story SAC frame is used as a
case study. SAC frame dimensions are presented in [15]. Selected
rigid connections were replaced by ductile semi-rigid connections.
A suit of 20 accelerograms [22], which was developed as a part of
the SAC project for the Los Angeles site, with different frequencies
was applied during inelastic dynamic analyses, and the results
were compared with the corresponding responses of the fully rigid
SAC frames.
A reliability-based analysis utilizing the Collapse Margin Ratio
(CMR), proposed in FEMAp695, was performed to quantify the collapse
performance of the Los Angeles SAC 20-story rigid frames and their
corresponding hybrid frames.
1.1. Nonlinear modeling
The concentrated plastic hinge model was adopted to introduce
nonlinear behavior in beams with rigid connections, beams with semi
rigid connections, and columns of structures. Beams with rigid connec-
tions and columns, as shown in Fig. 1, were modeled as compound
elements consisting of an elastic Bernoulli beam element at the middle
and conned by two plastic hinges and two end-zones that connect the
member to the rigid connections. The assumption of the formation of
plastic hinges at the two ends of beams and columns was adopted
based on the hypothesis that the failure mechanism is governed by
the seismic loading.
The semi-rigidbeamcompounds, as shown inFig. 2, were dened by
replacing the two stiff end zones with two non-linear moment-rotation
semi-rigid hinges in the rigid beam compounds. In this conguration,
plastic hinges and semi-rigid connections are both sources of nonlinear-
ity. However, since the plastic moment of semi-rigid connections is
usually much smaller than the plastic moment of beam sections, the
behavior of the beam compound is governed by the behavior of the
semi rigid connections. In fact, the moment demand in beams cannot
exceed the plastic moment of the semi-rigid connections; therefore, it
will not reach the plastic moment of the beam section.
The nonlinear behavior of plastic hinges is commonly expressed by
presenting their moment-rotation backbone-curves. In this study, the
backbone curve for different beam and column members were con-
structed based on the beam deterioration modeling guideline provided
in ATC-72 [9] as shown in Fig. 3.
The plastic hinge parameters for the beams used in this study are
summarized in Table 1.
Parameters used for modeling of semi-rigid connections will be
presented after the effective connection parameters are identied.
2. Selection of hybrid frame patterns
For a reasonable placement of semi-rigid connections in a hybrid
frame, it is necessary to investigate the local and global effects of adding
semi-rigid connections to the moment resistance frames' performance.
These effects were investigated as follows.
2.1. Moment redistribution/act as a fuse
Semi-rigid connections act as rotational springs; consequently, they
change the distribution of moments between beams and columns.
Moreover, the plastic moment of semi-rigid connections is generally
Fig. 1. Beam compound with stiff end zone and plastic hinges.
Fig. 2. Beam compound with semi-rigid connections and plastic hinges.
Pre-capping plastic rotation (
p
)
Post-capping rotation range (
pc
)
Ultimate rotation (
u
)
Effective yield strength and rotation (M
y
and
y
)
Capping strength and rotation, (M
c
and
c
)
Residual strength, M
r
Effective elastic stiffness, K
e
Post yield tangent stiffness, K
p
Effective post capping tangent stiffness, K
pc
Deterioration parameter ()
Fig. 3. Parameters of the monotonic backbone curve of the modied IbarraKrawinkler model.
After ATC72-10.
2 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
less than the plastic moment in their adjacent frame members (beams/
columns); thus, the moment cannot exceed the plastic moment of the
connection, and formation of plastic hinges in the adjacent structural
members is avoided.
A pushover analysis was performed on SAC 3-story rigid and hybrid
frames to investigate the effects of semi-rigid connections on moment
redistribution. Moment demands of the fully rigid 3-story frame and
its corresponding hybrid frame at roof drift of 3% are illustrated in Fig.
4-a and -b, respectively. In this study, a hybrid frame with a diagonal
pattern of semi-rigid connections was selected. Semi-rigid connections
are shown as rotational springs in Fig. 4-b. Moment demands of the
beams andcolumns adjacent tothe semi-rigid connections inthe hybrid
frame are noticeably less thanthe moment demands onthe correspond-
ing rigid frame members, as shown in Fig. 4.
2.2. Base shear reduction
In general, the stiffness of semi-rigid connections is smaller than the
rigid connections. Thus, implementing semi-rigid connections in a steel
frame reduces the overall stiffness of the frame and increases the
frame's fundamental period of the structure. Considering the shape of
the design spectra, as shown in Fig. 5, the softer frame experienced a
a) Moment in Fully rigid Frame (kip-in)
b) Moment in Hybrid Frame (kip-in)
-18364
25358
6994
-30083
25069
20028
-25042
-19342
7729
-25468
-5522
-3840.7
-1681
-12333
10882
-23215
10817
-10817
-12503
23441
12598
-8374
16737
-23450
-7733
23239
15505
1669
10826
12431
10724
21456
-10732
-30138 -24063 -29983 -23901
1602
-1659 1662
1669 -1605
Fig. 4. Bending moment demands at global drift of 3%.
Table 1
Parameters used for modeling plastic hinges in beams.
Section name y p pc My Mu Mr
W21X50 0.001272 0.031515 0.144247 6050 6655 2420 0.911012
W24X62 0.001123 0.027519 0.141778 8415 9256.5 3366 0.925926
W27X84 0.000974 0.022837 0.106695 13,420 14,762 5368 0.725497
W27X94 0.000967 0.022909 0.124601 15,290 16,819 6116 0.864289
W30X108 0.000881 0.021128 0.117753 19,030 20,933 7612 0.840001
W30X99 0.00089 0.020455 0.103769 17,160 18,876 6864 0.731785
3 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
reduced amount of acceleration. Fig. 5 shows that increasing the period
of the structures from t
1
to t
2
decreases the acceleration from Sa (t
1
) to
Sa (t
2
). However, since using semi-rigid connections will reduce the
system's stiffness, the stiffness of semi-rigid connections should be
determined to nd the minimum base shear, which still satises the
inter-story drift limit criteria. On the other hand, semi-rigid connections
may be used to shift the natural frequency of structure and avoid
resonance.
2.3. Shifting structures' period and changing their mode shapes
As explained in the previous section, implementing semi-rigid con-
nections in a SMRF reduces the stiffness of the frame; thus, the natural
periods of hybrid frames are higher than the natural period of corre-
sponding rigid frames. The natural mode-shapes of hybrid frames will
consequently differ from the fundamental mode-shapes of the corre-
sponding SMRF. The idea of the hybrid frame began with a simple as-
sumption that if a high-rise building is seen as a single cantilever
beam, theoretically, an indenite number of mode shapes is considered
for this system. The idealized simple model of rigid frame and its rst,
second, and third mode shapes are shown in Fig. 6.
Replacing rigid connections with exible semi-rigid connections at
certain story levels simulates spring development within the beam, as
shown in Fig. 7. These simplied models explain why the two cases of
rigidandhybridframes have a signicant difference inseismic behavior,
where the newly formed springs may help to reduce the frame's mode
shapes into two mode shapes, as shown in Fig. 7.
S
p
e
c
t
r
a
l

R
e
s
p
o
n
s
e
A
c
c
e
l
e
r
a
t
i
o
n
,

S
a

(
g
)
Period, T(sec)
SDS
t1 t2
Sa(t1)
Sa(t2)
Fig. 5. Schematic presentation of the Design Response Spectrum (ASCE 710).
a) Rigid Frame b) Simplified Model c) First Mode d) Second Mode e) Third Mode
Fig. 6. The simplied model of a rigid frame.
a) Hybrid Frame b) Simplified Model c) First Mode d) Second Mode
Fig. 7. The simplied model of a hybrid frame.
4 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
2.4. Increase energy dissipation
Introducing semi-rigid connections in SMRFs helps to reduce the
seismic demands in structures by dissipating seismic energy. In hybrid
frames, seismic energy dissipates in plastic hinges formed in frame
members (Beams/Columns) and in the semi-rigid connections. The
amount of the energy dissipated in each plastic hinge or semi-rigid
connection is found by calculating the area conned by the outer loop
of the moment-rotation hysteresis loop. Fig. 8 shows the locations of
plastic hinge formations, in SAC 20 story frame's members subjected
to LA35 ground motion, by lled circles.
2.5. Hybrid frame pattern selection
Based on the aforementioned effects of the semi-rigid connections
on the response of steel moment frames, three different approaches
were adopted for placement of semi-rigid connections in the 20-story
hybrid frame. Frames are designated as HSAC20, where the letter H
stands for Hybrid and SAC20 represents the 20 story SAC frame
designed for Los Angeles site.
The rst approach was based on introducing a rotational spring at
ve middle stories of the frame. This approach was implemented by re-
placing rigid connections with exible semi-rigid connections in stories
9 to 13 of the SAC 20 story rigid frame, as shown in Fig. 9(a). The newly
formed springs decoupled the earthquake acceleration into two mode
shapes, as is shown in Fig. 7. Although the performance of this system
was tested with initial linear models in Abolmaali et al. [4], in this
study, the pattern was evaluated using a comprehensive nonlinear
model analysis.
The second approach was based on utilizing the semi-rigid connec-
tions as an energy dissipative tool. As shown in Fig. 8, for a SAC 20-
story frame subjected to LA35 ground motion, plastic hinges formed in
beams and columns located in stories 1 through 5. On the other hand,
structural members of stories 6 through 20 remained elastic for the
most part and did not contribute to the inelastic energy dissipation.
The analysis result of the 20-story frame, under other maximum credi-
ble earthquake (MCE) ground motions scaledfor the Los Angles site, has
commonly followed the same pattern of plastic hinge propagation over
the height of frame. These results are presented in Razavi [20]. In the
second approach, semi-rigid connections were placed in the beams,
Fig. 8. Energy dissipated in plastic hinges of the SAC20 frame under LA35.
Rotational Spring
Approach
Energy Dissipation
Approach
Stability
Approach
a) HSAC20-1 b) HSAC20-2 c) HSAC20-3 d) HSAC20-4 e) HSAC20-5
Fig. 9. Hybrid models based on three proposed approaches.
5 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
which remained elastic. Thus, these connections contributed to the en-
ergy dissipation of the frame. This approach resulted in two patterns of
HSAC20-2 and HSAC20-3, as shown in Fig. 9(b) and (c), respectively.
The former pattern had more energy dissipative members; however,
the system was softer. Nevertheless, the HSAC20-2 frame experienced
less acceleration since it was softer and consequently had a larger
period.
The last approachwas based onmaintaining stability of the structure
under strong ground motions. Collapse occurs when plastic hinges form
in all columns of two different stories. Semi-rigid connections, as men-
tioned before, might be used as a fuse to protect their adjacent columns.
This approach aims to protect at least one column in each story level.
Using this approach, two patterns of HSAC20-4 and HSAC20-5, as
shown in Fig. 9(d) and (e), were proposed.
2.6. Evaluation of proposed patterns based on inter-story drift angle
Inter-story drift angle, which is expressed as inter-story drift (), di-
vided by the height of the story (h) is known as one of the best measures
of seismic performance at the story level of steel moment resistant frames
(SMF). The story drift is a global parameter since it is related to the global
drift angle (roof drift angle), which is dened as the roof displacement
divided by the height of the roof and consequently to the spectral
displacement demand. It is also a local parameter since it provides a
good estimation of member forces and deformation demands [15].
A major concern with using semi-rigid connections in steel frames is
that it may cause the inter-story drifts to increase beyond acceptable
limits. Although the use of hybrid frames causes a decrease in the initial
stiffness, the ground motions do not act similarly to static lateral loads
on frames. Ground motions exert forces to frames by introducing accel-
eration to stories' mass in story levels. Since semi-rigid connections shift
the period of structures, the amount of acceleration will reduce. On the
other hand, although the initial stiffness of a SMRF is more than the ini-
tial stiffness of its corresponding hybrid frame, the system stiffness of
the frame changes during an earthquake due to yielding in structural
members and nonlinear moment-rotation behavior of semi-rigid con-
nections. Consequently, the stiffness of the hybrid frame can be higher
than the stiffness of the corresponding moment frame during a ground
motion excitation. Thus, all frames are evaluated using an inelastic time
history analysis.
In order to determine the most effective pattern, the ve proposed
hybrid frames were modeled using a comprehensive nonlinear model
and were subjected to the SAC ground motions at the Los Angeles site.
SAC ground motions are categorized into two levels: Design Based
Earthquakes (DBE) and Maximum Credible Earthquakes (MCE), based
on their return period. The models were rst subjected to the set of 20
DBE records, LA01 to LA21, to be evaluated for the rst performance
objective, which was to satisfy the Life Safety (LS) performance under
the DBE hazard level. The acceptance criterion for this performance
objective is to maintain an average inter-story drift of less than 2.5%.
The models were then subjected to the set of 20 MCE records, LA21 to
LA40, to be evaluated for the second performance objective, which is
to satisfy the Collapse Prevention (CP) performance under MCE hazard
level. The acceptance criterion for this performance objective is to main-
tain an average inter-story drift of less than 5%.
0
2
4
6
8
10
12
14
16
18
20
0.0% 2.0% 4.0% 6.0% 8.0% 10.0% 12.0% 14.0% 16.0% 18.0% 20.0%
S
t
o
r
y
Drift
HSAC20-1
HSAC20-2
HSAC20-3
HSAC20-4
HSAC20-5
RIGID
C
o
l
l
a
p
s
e

P
r
e
v
e
n
t
i
o
n

L
i
m
i
t
Fig. 11. Average of story drift diagrams for various models of 20-story structure subjected
to L.A. MCE records.
0
5
10
15
20
1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% 5.0% 5.5% 6.0%
S
t
o
r
y
Maximum interstory drift ratio,
max
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
C
o
l
l
a
p
s
e

P
r
e
v
e
n
t
i
o
n

L
i
m
i
t
The percentages refer to the
stiffness of the beam with semi
rigid connection with respect to
the stiffness of the beam with
rigid connection
Fig. 12. Average story drift diagrams for HSAC20-4 with various connection stiffness
subjected to LA MCE records.
0
5
10
15
20
1.0% 2.0% 3.0% 4.0% 5.0% 6.0%
S
t
o
r
y
Maximum interstory drift ratio,
max
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
C
o
l
l
a
p
s
e

P
r
e
v
e
n
t
i
o
n

L
i
m
i
t
The percentages refer to the
stiffness of the beam with semi
rigid connection with respect to
the stiffness of the beam with
rigid connection
Fig. 13. Average story drift diagrams for HSAC20-5 with various connection stiffness
subjected to LA MCE records.
0
2
4
6
8
10
12
14
16
18
20
0.0% 1.0% 2.0% 3.0% 4.0% 5.0%
S
t
o
r
y
Drift
HSAC20-1
HSAC20-2
HSAC20-3
HSAC20-4
HSAC20-5
RIGID
L
i
f
e

S
a
f
e
t
y

L
i
m
i
t
Fig. 10. Average of story drift diagrams for various models of 20-story structure subjected
to L.A. DBE records.
6 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
Figs. 10 and 11 show the average of all drift results for the ve pro-
posed hybrid frames and the original SAC rigid frame subjected to DBE
and MCE records, respectively. In these graphs, each line shows an aver-
age of story drift ratios, which resulted from the dynamic inelastic time
history analysis of a frame subjected to 20 earthquake records. Thus,
each graph summarizes 120 nonlinear analysis drift results. Fig. 10
shows that all proposed models met the LS drift criteria, except the
HSAC20-1 model. However, HSAC20-4 and HSAC20-5 models have
the minimum amount of drift demand.
Similarly, drift results of the proposed hybrid frames subjected to the
MCE records conrmed the effectiveness of the semi-rigid connection
location pattern of HSAC20-4 and HSAC20-5 frames.
These observations led to selecting patterns used in HSAC20-4 and
HSAC20-5 models as the most effective patterns.
3. Sensitivity study on semi-rigid connections initial stiffness
To determine the effective connectionproperties, a parametric study
was performed on the initial stiffness of semi-rigid connections used in
the selected hybrid frames. These frames were subjected to the Los
Angeles MCE level records, and the average of results is presented for
each frame in Figs. 12 and 13.
As shown in Fig. 3, the beam model consisted of an elastic beam
column model, two plastic hinges, and two semi-rigid connections.
The exural rigidity of a beam is a function of the exural rigidity of
its components. The elastic beamcolumn member and the semi-rigid
connection can be considered as two rotational springs. The
compound's stiffness for this system of series springs is obtained by:
K
compound

1
1
K
Beam

1
K
Connection

K
Beam
K
Connection
K
Beam
K
Connection
: 1
Therefore, for connection stiffness equal to innity, the stiffness of
the compound is equal to the stiffness of the beam. On the other limit,
for a connection stiffness equal to zero, the compound's exural stiff-
ness equals to zero. Based on Eq. (1), the connections' initial stiffness
is back calculated for a total component stiffness of 10% to 100% of the
Column Beam
2
Top Angle Bolted
Seat Angle Bolted
15.6
3/4
d
4
Column
2
1/4
4
4
Weld
4
8.3
1/4
3/4
4
3/4
2
1/4
15
3
Web Angle
Fig. 14. Typical sketch of top- and seat-angle with double web angle connections.
Cyclic Load
Connection
Rigid End-Cap
Fig. 15. Top- and seat-angle with double web angle nite element model, and mesh properties.
7 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
stiffness of the beam. For an elastic beamcolumn member subjected
to double curvature bending, the initial stiffness of the beam equals to
6EI/L, where E is the modulus of elasticity of steel material, I is the
moment of inertia of the beam's cross section, and L is the length of
the beam. For the purpose of parametric study, the compound stiffness
is varied between 10 and 100% of the beams' stiffness. Then, the frames
are subjected to the MCE ground motions. Finally, the drift demands are
compared for selection of the best connections' initial stiffness.
Average drift diagrams for HSAC20-4 and HSAC20-5 frames with
various connection stiffness values subjected to LA MCE records are
illustrated in Figs. 12 and 13, respectively. These two graphs show that
in both semi-rigid patterns used for the 20-story frame, there is an
inverse relationship between connection stiffness and the maximum
drift. Further study on the frame drift demands subjected to single
earthquake records indicated that for ground motions with smaller ac-
celeration intensity, such as LA21, the greater connection stiffness
corresponded to the less drift demands. However, when frames were
subjected to a high intensity earthquake records, such as LA30, the
greater connection stiffness caused more drift demands. Story drift
ratio plots for individual ground motions are presented in Razavi [20].
Based on the presented study, a connection with stiffness of 30% to
50% of the stiffness of the beam was selected for implementation in
hybrid frames. Average story drift results support the hypothesis that
using connections with stiffness in this range reduces the probability
of collapse in MCE excitations; on the other hand, in the case of DBE
excitations, the system is stiff enough to meet the life safety criterion.
4. Semi-rigid connections
In general, boltedbolted or weldedbolted connections with slip
critical bolts, which are pre-tensioned to 70% of their minimum tensile
strength, are known as semi-rigidconnections. Static and dynamic char-
acteristics of semi-rigid connections are categorized by their moment
rotation (M-) curves and hysteresis loops, respectively.
For the purpose of selection of the type of semi-rigid connection, re-
sults of several experiments conducted by Abolmaali et al. [3] and Chen
and Kishi [11] were studied. Based on the effective initial stiffness and
ductility of the semi-rigid connections, top-and seat-angle with double
web angle connections were found to be the best match for the purpose
of this study. The hysteresis behavior of beamcolumnsteel connections
was predicted by means of a 3D non-linear nite element analysis
(FEA). Incremental nonlinear analysis takes into account material,
geometry, and contact nonlinearities in predicting moment-rotation
hysteresis loops. The results obtained from the nite element analyses
were validated by a series of full-scale structural tests performed by
Abolmaali et al. [3]. Modeling details are discussed in detail in Razavi
et al. [21] and Razavi [20]. Finally, semi-rigid connections used in this
study for different beam sizes were simulated. A bilinear curve was
tted to the moment-rotation hysteresis loop of each connection.
These curves were then used in global modeling of the hybrid frames.
A typical sketch of the top-and seat-angle with double web angle
connections selected for this study is shown in Fig. 14. A 3-D nite
element model of the top-and seat-angle with double web angle is
presented in Fig. 15. Fig. 16 shows the hysteresis loops for top-and
seat-angle with double web angle with W21X50 and the bilinear tted
curve.
The connections properties are also summarized in Table 2. Nota-
tions are explained in Fig. 3.
5. Force demands
To compare the shear and bending moments of the HSAC versus the
original rigid SAC frames, the ratios of the absolute value of maximum
forces for the HSAC were normalized with respect to the forces of the
SAC frame. Thus, a ratio of unity indicates that the HSAC forces are
equal to the rigid SAC forces, and the ratio of less than unity implies a
reduction in a given force for the hybrid frame (lower demand).
As an example, the beam's and column's moment ratios for the H-
SAC20-5 to the 20-story rigid frames subjected to two sets of ground
motions are plotted in Figs. 17 to 18. Each point in these graphs corre-
sponds to a normalized demand value, and the solid line shows the
-226
-169.5
-113
-56.5
0
56.5
113
169.5
226
-2000
-1500
-1000
-500
0
500
1000
1500
2000
-6%-5%-4%-3%-2%-1% 0% 1% 2% 3% 4% 5% 6%
M
o
m
e
n
t

(
k
i
p
.
i
n
)
Drift (%)
(
k
N
.
m
)
Fig. 16. FEMhysteresis loops for top- andseat-angle with double web angle with W21X50.
Table 2
Semi-rigid connections properties.
Beam Size
y
M
y
(kip-in)
u
M
u
(kip-in) K
e
K
p
W21X50 0.003 1032 0.05 1750 344,000 15,277
W24X62 0.003 1203 0.05 2000 401,000 16,957
W27X84 0.003 1411 0.05 2250 470,333 17,851
W30X99 0.003 1584 0.05 2450 528,000 18,426
W33X141 0.003 1920 0.05 2900 640,000 20,851
W36X150 0.003 2080 0.05 3300 693,333 25,957
0
5
10
15
20
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Ratio
LA21 LA22
LA23 LA24
LA25 LA26
LA27 LA28
LA29 LA30
LA31 LA32
LA33 LA34
LA35 LA36
LA37 LA38
LA39 LA40
Fig. 17. The ratios of the moment in beams for the 20-Story HSAC20-5/SAC subjected to MCE.
8 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
average in each story level. In addition, the average values of the mem-
bers' moment and shear ratios are tabulated in Table 3. On average, the
member forces are reduced between 6 and 16%.
6. Reliability analysis
This section focuses on evaluating and quantifying the collapse
performance of the Los Angeles SAC 20-story rigid frame and the pro-
posed hybrid frames by application of the FEMAP695 reliability analysis
methodology. An Incremental Dynamic Analysis (IDA) was performed
on the comprehensive nonlinear models of the rigid frame and its two
corresponding hybrid frames. The IDA was performed by applying the
20 records of the set of MCE, 2% probability of exceedance in 50 years,
ground motions. Finally, the collapse performance of the frames was
quantied and compared by computing the Collapse Margin Ratio
(CMR) value.
Incremental Dynamic Analysis (IDA), also known as Dynamic Push
Over (DPO), is an analysis method developed by Vamvatsikos and
Cornell [23] which aims at determining the global capacity of structures.
In this method, structures are subjected to one or more ground mo-
tion(s), each scaled to multiple levels of intensity Measure (IM). An IM
is a non-negative parameter, such as Peak Ground Acceleration (PGA)
or Spectral Acceleration (Sa), which represents the ground motion's
intensity. Then, the structure is analyzed under each scaled ground mo-
tion, utilizing a nonlinear dynamic analysis, and the Damage Measure
(DM) of interest is recorded. A DM is a measurable response of a struc-
ture such as ductility, global drift, or inter-story drift that is an output of
the nonlinear dynamic analysis under the prescribed seismic loading.
The smallest scale factor is selected to ensure an elastic response of
the structure, then the scale factor increases until the collapse limit
state is reached. The scale factor increment should be small enough to
capture the collapse point. Finally, a graph of DMs versus IMs is plotted
which also called IDA curve. In this study, the 5% damped rst mode
spectral acceleration (Sa (T1, 5%)) and the inter-story drift angle
(
max
) were selected as the IM and DM parameters, respectively.
A fragility function for collapse limit-state expresses the probability
of exceeding the limit state under a given ground motion with a certain
level of intensity. The fragility function is produced by using collapse
data from IDA results through a cumulative distribution function
(CDF) that expresses the probability of collapse as a function of ground
motion intensity (Ibarra et al., 2002). For this purpose, a lognormal dis-
tribution is tted to the collapse data obtained from IDA results. While
for a structure, the IDAis performed and the fragility curve is developed,
the median collapse intensity, SCT, should be identied. The lowest
intensity at which one-half of the records cause collapse is the median
collapse intensity, S
CT
[13]. In turn, this value is the spectral acceleration
corresponding to the 50% probability of collapse in the fragility curve.
The S
CT
value is a representative of the capacity of the structure. On
the other hand, the MCE intensity, S
MT
, which is dened as the median
5%-damped spectral acceleration of the MCE ground motions at the
fundamental period of the structure, is a representative of the plausible
demands applied to a structure.
To quantify the collapse performance of steel frames, FEMA P695
introduced the collapse margin ratio, CMR, which is the ratio of the
S
CT
to the S
MT
, as shown in Eq. (2).
Collapse Margin Ratio CMR
S
CT
S
MT
2
Indeed a bigger CMR corresponds to a less probability of collapse.
The collapse point of a frame for a particular ground motion is
dened as the lowest value of the following criteria: a) the point
where the slope of the IDA curve falls below 20% of the initial slope of
the curve, and b) the upper-bond inter-story drifts capacity of 10%.
The IDA curves and their corresponding fragility curves for different
20-story frames are shown in Fig. 19. Hollow circles in the IDA curves
correspond to the collapse points in structure. Moreover, each black
dot in the fragility curve corresponds to a cumulative probability of
collapse obtained from the IDA curves. The collapse margin ratio for
different SAC rigid frames and the proposed hybrid frames are shown
in Table 4. A larger CMR corresponds to a better performance.
Fig. 20(a) and (b) shows the nal deformed shape of SAC-20 story
rigid and the H-SAC20-5 frames subjected to LA36 ground motion, re-
spectively. The zigzag line in Fig. 20(b) was obtained by connecting
the locations of semi-rigid connections. As previously shown in Fig. 8,
formation of plastic hinges in stories 1 through 5 of the rigid frame
caused collapse in the original SAC frame. Although, the H-SAC20-5
has also shown excessive deformation, this structure did not collapse
when subjected to the identical earthquake record.
7. Conclusion
In this study, a new lateral resistant system called hybrid frame,
which is a combination of semi-rigid and fully rigid connections in
Table 3
Average hybrid/SAC demands response ratio.
Number of stories Model name Under DBE records Under MCE records
Beams' moment HSAC20-4 0.88 0.88
HSAC20-5 0.84 0.85
Beams' shear HSAC20-4 0.87 0.88
HSAC20-5 0.84 0.85
Columns' moment HSAC20-4 0.92 0.94
HSAC20-5 0.86 0.91
Columns' shear HSAC20-4 0.90 0.92
HSAC20-5 0.85 0.89
0
5
10
15
20
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
S
t
o
r
y
Ratio
LA21 LA22
LA23 LA24
LA25 LA26
LA27 LA28
LA29 LA30
LA31 LA32
LA33 LA34
LA35 LA36
LA37 LA38
LA39 LA40
Fig. 18. The ratios of the moment in columns for the 20-Story HSAC20-5/SAC subjected to MCE.
9 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
steel frames, was introduced to improve the performance of moment
resistant steel frames subjected to seismic excitations. The 20-story
SAC frame for the Los Angeles site was used as benchmark in this
study. Hybrid frames were designed by replacing selected fully-
rigid connections with more exible semi-rigid connections in the
SAC frame.
The effective semi-rigid connection patterns and the effective
connection stiffness were identied. Performance of the 20-story SAC
a) IDA Curves for Model SAC Rigid 20-Story b) Fragility Curve for Model SAC Rigid 20-Story
c) IDA Curves for Model HSAC20-4 d) Fragility Curve for Model HSAC20-4
e) IDA Curves for Model HSAC20-5 f) Fragility Curve for Model HSAC20-5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0% 5% 10% 15%
S
a
(
T
1
,

5
%
)

/

g
Maximum interstory drift ratio,
max
Maximum interstory drift ratio,
max
Maximum interstory drift ratio,
max
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
a
p
s
e
Sa(T1, 5%) / g
0
0.2
0.4
0.6
0.8
1
1.2
0% 5% 10% 15%
S
a
(
T
1
,

5
%
)

/

g
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
a
p
s
e
Sa(T1, 5%) / g
0
0.2
0.4
0.6
0.8
1
1.2
0% 5% 10% 15%
S
a
(
T
1
,

5
%
)

/

g
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
a
p
s
e
Sa(T1, 5%) / g
Fig. 19. IDA and fragility curves for different 20-story models.
10 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111
rigid and its corresponding hybrid frames were evaluated using the
reliability analysis presented in FEMA P695.
Comparison of performance of the 20-story SAC rigid frames with its
corresponding hybrid frames showed a superior performance for the
hybrid frame. Collapse Margin Ratio of the investigated hybrid frames
was between 20 and 33% better than the performance of the corre-
sponding rigid frame. This, in turn, reduced the probability of collapse
when the frame was subjected to severe ground motions.
The force demands in the beams and columns of the proposed 20-
story hybrid frames were reduced up to 16%. Considering the fact that
the structural members' (beams and columns) sections in the hybrid
frames were the same as those of the corresponding rigid frames, the
proposed frame system performed superior to conventional SMFs
when subjected to earthquake excitation.
References
[1] Abolmaali A, Kukreti AR, Razavi SH. Hysteresis behavior of semi-rigid double web
angle steel connections. J Constr Steel Res 2003;59:105782.
[2] Abolmaali A, Matthys JH, Farooqi M, Choi Y. Development of moment-rotation
model equations for ush end-plate connections. J Constr Steel Res 2005;61:
1595612.
[3] Abolmaali A, Kukreti AR, Motahari A, Ghassemieh M. Energy dissipation characteris-
tics of semi-rigid connection. J Constr Steel Res 2009;65:1927.
[4] Abolmaali A, Razavi M, Radulova D. On the concept of earthquake resistant hybrid
steel frames. J Constr Steel Res 2012;68:3442.
[5] Ackroyd MH, Gerstle KH. Behavior of type 2 steel frames. J Struct Div ASCE
1982;108(ST7):154158.
[6] Astaneh A, Nader MN, Malik L. Cyclic behavior of double angle connections. J Struct
Eng ASCE 1989;115(5):110118.
[7] Astaneh-Asl A. Seismic design of bolted steel moment-resisting frames. Moraga, CA
Steel Tips, Structural Steel Education Council; 1995 [82 pp.].
[8] Astaneh-Asl A, Nader MN, Harriott JD. Seismic behavior and design considerations in
semi-rigid frames. Proceedings of the 1991 national steel construction conference.
AISC; 1991.
[9] ATC-72. Guidelines on modeling and acceptance criteria for seismic design and anal-
ysis of tall buildings. Redwood City, CA Applied Technology Council; 2010 [ATC-72-
1].
[10] Bjorhovde R, Colson A, Brozzetti J. Classication systemfor beam-to-column connec-
tions. J Struct Eng ASCE 1990;116(11).
[11] Chen Wai-Fah, Kishi N. Semi-rigid steel beam-to-column connections: data base and
modeling. J Struct Eng 1989;115(1):10519.
[12] FEMA. Pre-standard and commentary for the seismic rehabilitation of buildings, re-
port FEMA-356. Washington, DC The SAC Joint Venture for the Federal Emergency
Management Agency; 2000.
[13] FEMA P695. Quantication of building seismic performance factors. Washington,
D.C. Federal Emergency Management Agency; 2009
[14] Goel Subhash C, et al. Performancebased plastic design (PBPD) method for
earthquakeresistant structures: an overview. The, structural design of tall and
special buildings, 19.12. , ; 2010. p. 11537.
[15] Gupta A, Krawinkler H. Seismic demands for performance evaluation of steel
moment resisting frame structures (SAC Task 5.4.3). Report no. 132. CA John A.
Blume Earthquake Engineering Center, Stanford University; 1999.
[16] Kukreti AR, Abolmaali A. Moment-rotation hysteresis behavior of top and seat angle
steel frame connections. J Struct Eng ASCE 1999;125(8):81020.
[17] Leelataviwat S, Goel SC, Stojadinovic' B. Toward performance-based seismic design
of structures. Earthq Spectra 1999;15(3):43561.
[18] Maison Bruce F, et al. Performance of PR moment frame buildings in UBC seismic
zones 3 and 4. J Struct Eng 2000;126(1):10816.
[19] Newmark NM, Hall WJ. Earthquake spectra and design. El Cerrito, California
Earthquake Engineering Research Institute; 1982.
[20] Razavi M. On the concept of earthquake resistant hybrid steel frames. A Dissertation
submitted as partial fulllment of the requirement for degree of Ph.D. Arlington,
Texas Department of Civil Engineering, University of Texas at Arlington; 2013.
[21] Razavi M, Kiamanesh R, Abolmaali A. Computational benchmarks in simulation of
cyclic performance of steel connections using three-dimensional nonlinear nite el-
ement method, i-manager's. J Struct Eng September November 2012;1(3):1525.
[22] Somerville Paul G, et al. Modication of empirical strong ground motion attenuation
relations to include the amplitude and duration effects of rupture directivity.
Seismol Res Lett 1997;68(1):199222.
[23] Vamvatsikos Dimitrios, Allin Cornell C. Incremental dynamic analysis. Earthq Eng
Struct Dyn 2002;31.3:491514.
Table 4
Collapse margin ratio for different frame models.
Structure type CMR
Rigid 1.48
HSAC20-4 1.80
HSAC20-5 1.99
a) Rigid SAC 20- Story b) H-SAC20-5
Fig. 20. The nal deformed shapes of the SAC20-story and HSAC20-5 frames subjected to
LA36 ground motion.
11 M. Razavi, A. Abolmaali / Journal of Constructional Steel Research 98 (2014) 111