Professional Documents
Culture Documents
Earthquake
Earthquake
SUMMARY
A series of E-Defense shaking table tests are conducted on a large-scale test specimen that represents
a high-rise steel building. Two types of connections featuring the connection details commonly used in
1970s, in the early days of high-rise construction in Japan, are adopted: the field-welded connection
consisting of welded unreinforced flanges and a bolted web type, and the shop-welded connection in
which the flanges and web are all-welded to the column flange in the shop. To examine the seismic
capacity of a total of 24 beam-to-column connections of the specimen, particularly when it is subjected
to long-period ground motion characterized not so much by large amplitude as by very many cycles of
repeated loading, the test specimen is shaken repeatedly until the connections fractured. The test results
indicate that a few of the field-welded connections fractured from the bottom flange weld boundary in a
relatively small cumulative rotation primarily due to the difficulties in ensuring the welding and inspection
performance in the actual field welding. The shop-welded connections are able to sustain many cycles of
plastic rotation, with an averaged cumulative plastic rotation of 0.86 rad. Two shop-welded connections
exhibit ductile fractures but only after experiencing many cycles. The presence of RC floor slabs promotes
the strain concentration at the toe of the weld access hole in the bottom flange by at least twice compared
with the case without the slab, which had resulted in a decrease in the cumulative plastic rotation by
about 50%. Copyright 䉷 2010 John Wiley & Sons, Ltd.
KEY WORDS: high-rise building; long-period ground motion; beam-to-column connection; composite
action
1. INTRODUCTION
Periodic occurrences of large ocean-ridge earthquakes having a magnitude over eight along the
subduction zones in the southwest part of Japan have been documented in historical materials.
Because such earthquakes have return periods of 100-to-100 and 50 years, Japan will most probably
experience the next one by the middle of this century. Subduction zone earthquakes are known to
generate long-period ground motions on land, especially in basin areas where large cities, such as
Tokyo, Nagaya, and Osaka, are located. Their predominant periods range from several to 10 s, and
the durations of primary motion extend over several minutes [1, 2]. Long-period ground motions
tend to resonate high-rise buildings whose fundamental natural periods are a few seconds, and
accordingly the energy input to the high-rise buildings is expected to reach more than several times
what has been expected in Japanese seismic design [3]. According to some numerical analyses of
∗ Correspondence to: Yu-Lin Chung, Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto
611-0011, Japan.
† E-mail: y.l.chung77@gmail.com
existing high-rise buildings, long-period ground motions could demand significantly larger energy
dissipation than the design waves commonly used in the Japanese seismic design, especially to
beam-to-column connections, with anticipated cumulative deformations amounting to several to
10 times larger than those expected for the design waves [4].
The 1994 Northridge and the 1995 Hyogoken-Nanbu earthquakes tested the seismic performance
of steel moment frames, which have been widely used in seismic design in the U.S. and Japan. In
the earthquakes, severe damage was observed in the welded unreinforced flanges and bolted web
connections (WUF-B) [5–7]. A number of Japanese high-rise buildings built in the 1970s had also
adopted WUF-B connections, although connection details were similar but not exactly identical to
those damaged in the earthquakes.
Considering those circumstance, the writers decided to experimentally evaluate the seismic
performance of the high-rise buildings subjected to long-period ground motions. Several alternatives
were considered for the type of testing, i.e. a shaking table test, a hybrid online test, and a
quasi-static loading test. In all cases, the test specimen had to preserve a realistic scale since the
behavior involving serious damage such as fractures was to be simulated. A substructure-based
hybrid online test [8–13] was appealing to conduct such a test, but quasi-static loading was not
necessarily preferable as fracture behavior was known to be velocity-dependent. A real-time hybrid
online test [14–18] can allow for velocity-dependency, but the current state-of-the-practice has not
reached the level of reliable applications particularly for large-scale tests. A quasi-static test in an
in situ condition is notable, too, but to find a test specimen requires fortuity [19]. In reference
to those pros and cons, a shaking table test using a large-scale shaking table was adopted [20].
A substructure test method was employed, and a test specimen that represented a 21 story steel
high-rise building was shaken on a very large shaking table, dubbed the E-Defense shaking table
[21–23]. The test specimen, whose detail will be described later, consisted of a four-story frame
structure that featured full-scale steel members and reinforced concrete slabs, and substitute layers
that were placed on top of the frame and represented the upper stories. Two types of connections,
both popularly used for Japanese high-rise buildings constructed in late 1960s to 1970s, were
included in the frame: the field-welded connection, which is a type of WUF-B, and the shop-welded
connection, in which the beam web is welded rather than bolted, and the entire connection is
fabricated in the shop [7]. Three ground motions, including one motion used in Japanese seismic
design, one near fault motion, and two long-period ground motions, were applied sequentially with
increasing magnitudes. The largest inter-story drift, which was caused by the larger long-period
ground motion of the two, was about 2%. The cumulative story drift exhibited by that wave was
about 25 relative to the yield drift, which was 15 times as large as that caused by the design
wave. In the 56 beam-to-column connections of the test structure, the moments and rotations of
24 beam-to-column connections were recorded in the test.
After the shaking by the original set of ground motions, the test specimen was shaken repeatedly
using the larger of the two long-period ground motions until fractures occurred in multiple connec-
tions. The shaking aimed at quantifying the seismic capacity of the connections. This paper presents
the major results and findings obtained from the series of shaking conducted for the quantification
of the capacity. First, the test specimen, detail of the adopted beam-to-column connections, and
loading program are introduced. Second, the seismic responses of the test specimen generated by
the sequential loading is presented in terms of the maximum inter-story drift angles. Third, the
effect of RC floor slab on the overall connection behavior is investigated by comparing the results
obtained from the tests with those of bare-beam tests conducted in the past. Last, the cumulative
ductility capacity of these connections is characterized in reference to the test results.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 607
1450
1450
1500
3800
21850
3800
3800
4500
1150
6000 6000
Unit: mm 12000
Figure 1. E-Defense test specimen: (a) overview; (b) test specimen; and
(c) development of substitute layers.
a test method was developed that preserved the dimensions of structural members within the
loading capacity of the E-Defense shaking table. Figure 1(a) shows the overview of the test system.
The test specimen consists of a four-story, two-span by one-bay steel moment frame, and three
substitute layers are placed on top of the moment frame as shown in Figure 1(b). The substitute
layers, which consist of concrete slabs and rubber bearings, are arranged to represent the upper
portion of the prototype. The procedure to condense the upper portion and construct the substitute
layers is shown in Figure 1(c), and the detail is found in [24]. The test specimen has a plan
dimension of 12 m in the longitudinal direction and 8 m in the transverse direction, and is 21.9 m
in the total height. The first mode periods of the entire structure were 2.13 and 2.24 s in the
longitudinal and transverse directions. These values were obtained from a white-noise input with
the RMS of 0.6 m/s2 , 250 s in duration, and 0.2–20 Hz in frequency bandwidth. The detail of the
specimen and test setup is shown elsewhere [20, 24].
Figure 2 shows the two types of connection adopted in the test frame: the shop-welded connec-
tion and the field-welded connection. In Japanese high-rise buildings built in 1960s and early
1970s, shop-welded connections were often used due to the concerns about weld quality and weld
shrinkage. In such shop-welded connections, the beam is cut into halves in the mid span, and the
flanges and web of the half beam are all-welded to the column in the shop. Then, the shop-welded
assemblage that consists of a column and half beams is transported to the construction site. Finally,
the shop-welded half beams are bolted to each other at the mid span. With gradual development
of nondestructive testing, construction management, and accumulated experiences of welding, the
field-welded connection with welded flanges and bolted web gained popularity in the early 1970s.
Thus, these two types of connection were mixed in Japanese high-rise buildings constructed in
the early 1970s. With the advantages of better constructibility and lower cost, the field-welded
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
608 Y.-L. CHUNG ET AL.
Slab 7 35°
120
35
Column 35R
PL- 22
400
35
Upper side
600
35
H-400×200×8×13 35R
PL- 22 35
Column
H-600×200×9×19 Box-400×25 7 35°
(a) Bottom side
Slab Slab
120
120
735°
15
90
90
35
PL-22
560(@70×8) 7
560(@70×8)
Upper side
800
800
7 110
35° 70
150
150
PL-22
PL-9 PL-9
H-800×199×10×15 Bottom side
Box-400×25
(b)
connections were adopted for preference, but due to better quality control, the shop-welded connec-
tions were still used when the beam span was relatively short. Considering those situations at that
time, both types of connection were adopted in the test.
Figure 3 shows the typical plan of the specimen. In the longitudinal direction, a built-up wide
flange section of H 600×200×9×19 (the depth, width, web thickness, and flange thickness in
mm) was arranged with the shop-welded connection detail. In the transverse direction, a honey-
comb section of H 800×199×10×15, which was processed from a rolled section of H 596×
199×10×15, was arranged with the field-welded connection detail. The details of beams and
beam-to-column connections were chosen in reference to the design practice at the time of design.
Two types of web design were adopted in the field-welded connections (Figure 2(b)). One was a
then-typical design in which the web bolts were assumed to sustain only the beam shear force. The
other was a design in which the web bolts took both the shear force and partial bending moment
of the beam. The two connections turned out to have 9 and 13 bolts in the web, respectively.
The 9-bolt connection was arranged for the second and fourth floor beams and the 13-bolt design
was arranged for the third floor beams, respectively. In each of the second to fourth floors, eight
connections were chosen for detailed recording, which resulted in a total of 24 connections for
investigation of detailed hysteretic behavior.
The columns had a box section of 400×400×25 (the depth, width, and thickness in mm) with
inner diaphragms at the connection. The columns were designed to be stronger than beams by a
column-to-beam strength ratio of 1.5. RC floor slabs with a thickness of 120 mm were cast at every
floor. The first-story column bases were embedded in strong RC beam foundations for a depth of
0.9 m so that they were considered as fixed.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 609
Measured connection
Location of strain gauges
NR NL
H600 H400 Slab
B
WL ER 120
900 1600
h/3
H800
H400
H800
8000 h h/3
1600
h/3
WR EL
A Strain gauges
1000
H600 H400
SL SR Beam section
6000 6000 1000
1 2 3
the Japanese seismic design code [25]. Level 1 is for small to medium earthquakes with a PGA
ranging from 1.5 to 2 m/s2 , and the PGV is 0.25 m/s. Level 2 is for large earthquakes with the
PGA ranging from 3 to 4 m/s2 , and the PGV is 0.5 m/s. In Japanese seismic design, the 1940 El
Centro wave is commonly scaled to match the PGV of 0.25 and 0.5 m/s for levels 1 and 2. As
for long-period ground motions, two synthesized long-period ground motions, the Hog and San
waves, were adopted. The Hog wave was predicted at a Tokyo site, and a rupture of the Tokai
trough was supposed. The San wave was predicted at a Nagoya site, and simultaneous ruptures
of the Tokai and Tonankai troughs were supposed. Figure 4 shows the time histories and velocity
response spectra of the input waves. Waves were input in two horizontal directions. First, the level
1 El Centro wave was applied. Then, the level 2 El Centro wave (illustrated as El 0.5 m/s), Hog
wave, and the San wave were applied sequentially. After this loading, one-directional input of the
San wave, applied to the longitudinal direction, was applied two more times sequentially until a
few connections arranged in the longitudinal direction (shop-welded connections) fractured.
3. TEST RESULTS
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
610 Y.-L. CHUNG ET AL.
Longitudinal Transverse
El Centro-NS-0.5m /s El Centro-EW-0.5m /s
Acc. (m/s)
Acc. (m/s)
Hog-NS
Hog-EW
San-EW San-NS
0.1 0.1
0 100 200 300 0 100 200 300
(a) Sec Sec
Longitudinal Transverse
3 3
El Centro-NS-0.5m/s El Centro-EW-0.5m/s
Hog-NS Hog-EW
2 San-EW 2 San-NS
Sv(m/s)
Sv(m/s)
1 1
0 0
0 1 2 3 4 0 1 2 3 4
(b) Period (Sec) Period (Sec)
Figure 4. Time histories and pseudo velocity spectra of input waves: (a) time histories
and (b) pseudo velocity spectra.
20 20 EL 0.25m /s Level 1
EL 0.5m /s
Level 2
15 Hog
15
San-1
Story
Story
San-2 Level 3
10 10 San-3
5 5
0 0
0 0.005 0.01 0.015 0.02 0 0.005 0.01 0.015 0.02
(a) Drift (rad) (b) Drift (rad)
Figure 5. Maximum inter-story drift angles: (a) longitudinal direction and (b) transverse direction.
Figure 6. Time history of second story drift angle and instants of first fracture.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 611
Figure 7. Cumulative inter-story plastic drift angles: (a) cumulative inter-story plastic deformation ratio
and (b) definition of cumulative deformation.
longitudinal directions, respectively. The level 2 El Centro wave generated cumulative inter-story
plastic drift angles of 1% in both directions. It is notable that the long-period ground motion
generated 10 times or more the cumulative plastic deformation than what was expected by the design
wave.
After the 1994 Northridge earthquake, extensive analyses were conducted on the performance
of beam-to-column connections [5]. Based on numerous tests and analyses at that time, a standard
loading protocol for qualifying the performance of beam-to-column connections [26] was devel-
oped. According to the protocol, the loading history consists of six cycles each for drift angles
of 0.375, 0.5, and 0.75%, four cycles for drift angles of 1, and 2%, and two cycles each for drift
angles with an increment in drift angle of 1%. The tested connection is qualified if it sustains at
least 80% of the full plastic moment, b Mp , after the completion of at least one cycle of 4% drift
angle. In U.S. steel moment frames, the yield drift angle is commonly about 1%. With this yield
drift angle in mind, the cumulative inter-story plastic drift angle expected by the protocol reaches
52% at the end of loading with a drift angle of 4%. When looking into the cumulative inter-story
plastic drift angles obtained from the test, the first San wave generated about 20% in the cumulative
angle, and after two San waves, the cumulative inter-story plastic drift angle reached 53%. The
cumulative inter-story plastic drift angle stipulated by AISC, 52%, is close to this value. This
indicates that the criteria stipulated by AISC may be too stringent for the performance checking of
beam-to-column connections of high-rise buildings when they are subjected to long-period ground
motions.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
612 Y.-L. CHUNG ET AL.
Figure 8. Hysteretic behavior of field-welded connections (San-1): (a) second floor connections; (b) third
floor connections; and (c) fourth floor connections.
Figure 9. Hysteretic behavior of shop-welded connections (San-3): (a) second floor connections; (b) third
floor connections; and (c) fourth floor connections.
resistance recovered. None of the shop-welded connections fractured during the first and second
San wave loadings (San-1 and San-2), and all connections exhibited stable behavior. During the
third San wave loading (San-3), two shop-welded connections fractured (Figures 9(a) and (b)).
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 613
Column Web
Column
Flange
(a) (b)
Figure 10. Fractures in bottom flange: (a) field-welded connection and (b) shop-welded connection.
Two of the fracture surfaces, one from the field-welded connection and the other from the
shop-welded connection, are shown in Figure 10. In the field-welded connection, initial cracks
started from the toe of the weld access hole and grew to fracture along the weld boundary of the
bottom flange. In the shop-welded connection, cracks also started from the weld access hole and
grew to fracture in the flange metal.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
614 Y.-L. CHUNG ET AL.
(a)
(b)
Figure 12. Microscopic examination of unfractured cross-sections: (a) bottom flange test piece (third floor,
ER); (b) Section A; (c) Section B; and (d) detail of Section B.
Composite effects in steel beams and connections have been examined in the past (for example,
[28–33]). In what follows, the effects are characterized in terms of stiffness increase, strength
increase, and strain concentration in the bottom flange.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 615
2.5
Period (sec)
2.0
Longitudinal
Transverse
1.5
Initial Level 1 Level 2 Level 3
(El Centro) (El Centro) (San)
Loading
Figure 14. Moment–curvature relationship at beam section (1.6 m away from the column face): (a)
fourth-floor connection (NR) and (b) fourth-floor connection (EL).
on the average, with the coefficient of variation as 0.06 and 0.05, respectively. It is notable that
the stiffness values are very uniform. In four connections out of 24, another cross-section closer
to the column face and therefore yielding more significantly, located 0.9 m away from the column
face, was also installed with strain gauges. The stiffness values in the cross-section are plotted in
Figure 15(a), which indicates that they are also very close to those obtained for the cross-section
farther by 1.6 m. Figure 15(b) show similar plots, but this time for the stiffness after the level
1 El Centro wave. The stiffness values are again very uniform among the connections, with the
average and coefficient of variation as 1.45 and 0.05 (in the longitudinal direction) and 1.41 and
0.02 (in the transverse direction). Figure 16 summarizes the changes in stiffness after each loading.
It is notable that the averaged stiffness decreased rather significantly, originally from 1.9 down
to 1.1 after the San wave, and the growth of cracks present in the RC floor slabs was primarily
responsible for the reduction.
An associated numerical analysis was conducted to examine the correlation between the reduc-
tion in beam stiffness and the natural period. The numerical model adopted member-to-member
representation, whereas all members were assumed as elastic. First, the nominal geometric proper-
ties were assigned in the model, and the composite action with RC floor slabs was allowed for by
adopting a factor of 1.9 for the beam stiffness (in reference to Figure 15(a)). Young’s modulus of
the steel was then slightly adjusted so that the fundamental natural period in the initial stage would
match the experimentally obtained natural period. The same model was repeatedly analyzed for the
natural period, each time using a different factor for the beam stiffness in reference to Figure 16.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
616 Y.-L. CHUNG ET AL.
Stiffness ratio
(1.6m) (0.9m) (1.6m)
2 2
1 1
4F
WL 4F
3F
WL 3F
4F
NR 4F
3F
NR 3F
WR
WR
WR
WR
WR
WR
WL
WL
WL
WL
WL
WL
NR
NR
NR
WL
WL
NR
NR
NR
NR
SR
NR
ER
SR
SR
SR
ER
ER
ER
ER
ER
SR
SR
NL
EL
NL
NL
EL
SL
EL
NL
NL
NL
EL
EL
SL
SL
SL
SL
SL
EL
0 0
2F 3F 4F 2F 3F 4F 2F 3F 4F 2F 3F 4F
(a) Measured location (b) Measured location
Figure 15. Stiffness ratios of composite beams: (a) response before test and (b) response after
level 1 El Centro wave loading.
3
Stiffness ratio
1
Longitudinal
Transverse
0
Initial Level 1 Level 2 Level 3
(El Centro) (El Centro) (San)
Loading
Thus obtained natural periods are plotted in Figure 17, together with the corresponding periods
obtained experimentally. The correlation between the two is reasonable, except for the transverse
direction after the San wave. The discrepancy was because of the fractures of four connections
during the San wave. The observations in this section suggest that the natural period of steel frames
can change rather significantly although the steel itself is known to recover its original stiffness
even after yielding, and this is primarily due to the damage (cracks) induced in the RC floor slabs
and resultant reduction in composite action.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 617
2.5
Period (sec)
2.0
Experiment (Longitudinal)
Experiment (Transverse)
Numerical model (Longitudinal)
Figure 17. Comparison in natural period between test and numerical analysis in which
composite action is considered.
E-Defense specimen
2
M/bMp
1
WR
WR
WR
WL
WL
WL
NR
NR
NR
ER
SR
SR
SR
ER
ER
NL
NL
NL
EL
SL
SL
EL
EL
SL
0 Y K
2F 3F 4F 2F 3F 4F
Connections
by 5% than those with 9 bolts. As shown later, the cumulative plastic rotation of the fractured
connections remained nearly unchanged. Overall, the supplemental bolts had a minimal effect on
the improvement of performance.
The average strength of the fractured and unfractured field-welded connections are 1.1 and
1.3b Mp . For the shop-welded connections, the averaged values are 1.30 and 1.31, respectively. The
shop-welded connections show more stable performance, and although the connections fractured,
this occurred after they underwent stronger input.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
618 Y.-L. CHUNG ET AL.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 619
1.0
2rd cycle
0.5
WL
EL
EL
EL
WR
WR
WR
WL
WL
NR
NR
NR
ER
SR
SR
ER
ER
SR
NL
NL
NL
SL
SL
SL
Y K
0
2F 3F 4F 2F 3F 4F
(a) Connections
Unfractured Fractured
0.5
Field-welded Shop-welded
0.25
SR
SR
SR
WR
SL
WR
WR
SL
WL
WL
NR
ER
NR
SL
ER
NL
WL
NR
EL
EL
ER
NL
EL
NL
0
2F 3F 4F 2F 3F 4F
(b) Connections
Unfractured Fractured
0.5
Field-welded Shop-welded
0.25
WR
WR
WR
WL
WL
WL
NR
NR
NR
SR
ER
SR
ER
ER
SR
NL
NL
NL
SL
EL
EL
SL
SL
EL
0
2F 3F 4F 2F 3F 4F
(c) Connections
Figure 20. Cumulative plastic rotations of connections: (a) cumulative plastic rotation (up to fractures);
(b) cumulative plastic rotation (El Centro); and (c) cumulative plastic rotation (San).
even more vulnerable due to the promotion of tensile strain at that spot. When the weakest spot
is moved away from the connection and located in the beam section, the floor slab serves as a
restrainer to delay the beam instability.
Within the 12 field-welded and 12 shop-welded connections, the field-welded and shop-welded
connections fractured in four and two locations, respectively. The average cumulative plastic
rotations of those fractured connections were 0.16 and 0.86 rad, respectively. Considering these
8
numbers, 67% (= 12 ) of the field-welded connections can sustain the cumulative plastic rotation
over 0.16 rad, and 85% (= 10 12 ) of the shop-welded connections can sustain cumulative plastic
rotations over 0.86 rad. In Figures 20(b) and (c), the averaged cumulative rotations of the unfractured
connections (including both the field-welded and shop-welded connections) were 0.035 and 0.23 rad
under the level 2 El Centro and San waves, respectively. In comparison with the rotation demands
required for the two waves, the minimum capacity of the field-welded connections (0.16 rad)
fell in between. The long-period ground motion demanded a 1.5 times larger cumulative plastic
rotation than the minimum capacity, whereas it was about five times larger than the demand
required for the design wave. For the shop-welded connection, on the other hand, the minimum
capacity (0.86 rad) was four times larger than the demand required for the long-period ground
motion.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
620 Y.-L. CHUNG ET AL.
5. CONCLUSIONS
A series of shaking table tests were conducted on a full-scale test structure that represented
high-rise buildings. A large shaking table, called E-Defense, was used in the test. Two types
of connections having the connection details commonly used in the 1970s, the early days of
high-rise construction in Japan, were adopted: the field-welded connection consisting of welded
unreinforced flanges and a bolted web type and the shop-welded connection in which the flanges
and web are all-welded to the column flange in the shop. To assess the seismic capacity of a
total of 24 beam-to-column connections of the specimen, particularly when it is subjected to long-
period ground motion, characterized not so much by large amplitude but by very many cycles of
repeated loading, the test specimen was shaken repeatedly until the connections fractured. Based
on the test, the seismic capacity of those connections was examined in terms of the ductility
capacity and the effect of RC floor slabs. The major findings obtained from this study are as
follows:
The next series of full-scale shaking table test was recently completed for the identical test
specimen but this time with various retrofit methods. They included the addition of buckling-
restrained braces, addition of oil dampers, and strengthening of beam-to-column connections. As
for the last retrofit method, connections were strengthened by the attachment of a pair of horizontal
haunches (wing plates) at the bottom flange, attachment of a vertical haunch (a rectangular section)
beneath the bottom flange, and supplemental welds at the shear tab. The performance of the
retrofitted test specimen is a subject to report in succeeding papers.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
SEISMIC RESISTANCE CAPACITY OF BEAM–COLUMN CONNECTIONS 621
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support of the Ministry of Education, Culture, Sports, Science &
Technology (MEXT) to carry out the test presented in this paper. The authors thank Kunio Fukuyama and
Kouichi Kajiwara of National Research Institute for Earth Science and Disaster Prevention for designing
the test specimen and gathering the experimental data used herein. The authors would also like to thank
the reviewers for their valuable commons and suggestions.
REFERENCES
1. Kamae K, Kawabe H, Irikura K. Strong ground motion prediction for huge subduction earthquakes using a
characterized source model and several simulation techniques. Proceedings of the 13th World Conference on
Earthquake Engineering, Vancouver, BC, Canada, 2004; No. 655.
2. Kawabe H, Kaname K, Irikura K. Damage prediction of long-period structures during subduction earthquakes—
part 1: long-period ground motion prediction in the Osaka basin for future Nankai Earthquakes. Proceedings of
the 14th World Conference on Earthquake Engineering, Beijing, China, 2008; P291.
3. Japan Society of Civil Engineering, Architectural Institute of Japan. Report of Seismic Performance Improvement
of Civil, Architectural Structures Subjected to Long-period Ground Motions Generated by Subduction Zone, 2006
(in Japanese).
4. Suita K, Kitamura Y, Goto T, Iwata T, Kamae K. Seismic response of high-rise buildings constructed in 1970s
subjected to long-period ground motions—using predicted ground motions from hypothetical Nankai trough
earthquakes in Kansai area. Journal of Structural and Construction Engineering, Architectural Institute of Japan
2007; 611:055–061 (in Japanese with English abstract).
5. Federal Emergency Management Agency. Recommended seismic design criteria for new steel moment-frame
buildings, FEMA350, July 2000.
6. Bertero VV, Anderson JC, Krawinkler H. Performance of steel building structures during the Northridge earthquake.
Report No. UCB/EERC-94/09, EERC, University of California, Berkeley, CA, 1994.
7. Nakashima M, Roeder CW, Maruoka Y. Steel moment frames for earthquakes in United Stated and Japan. Journal
of Structural Engineering 2000; 126(8):861–868.
8. Wang T, Yoshitake N, Pan P, Lee TH, Nakashima M. Numerical characteristics of peer-to-peer (P2P) internet
online hybrid test system and its application to seismic simulation of SRC structure. Journal of Earthquake
Engineering and Structural Dynamics 2008; 37(2):265–282.
9. Wang T, McCormick J, Yoshitake N, Pan P, Murata Y, Nakashima M. Collapse simulation of a four-story steel
moment frame by a distributed online hybrid test. Journal of Earthquake Engineering and Structural Dynamics
2008; 37(6):955–974.
10. Pan P, Tada M. Hybrid formulation of operator splitting (OS) and Newmark- methods for collaborative structural
analysis (CSA). Journal of Earthquake Engineering and Structural Dynamics 2008; 37(7):1117–1133.
11. Yang TY, Stojadinovic B, Moehle J. Hybrid simulation of a zipper-braced steel frame under earthquake excitation.
Journal of Earthquake Engineering and Structural Dynamics 2009; 38(1):95–113.
12. Hung CC, El-Tawil S. A method for estimating specimen tangent stiffness for hybrid simulation. Journal of
Earthquake Engineering and Structural Dynamics 2009; 38(1):115–134.
13. Stavridis A, Shing PB. Hybrid testing and modeling of a suspended zipper steel frame. Journal of Earthquake
Engineering and Structural Dynamics 2010; 39(2):187–209.
14. Ahmadizadeh M, Mosqueda G, Reinhorn AM. Compensation of actuator delay and dynamics for real-time hybrid
structural simulation. Journal of Earthquake Engineering and Structural Dynamics 2008; 37(1):21–42.
15. Mercan O, Ricles JM. Stability analysis for real-time pseudodynamic and hybrid pseudodynamic testing with
multiple sources of delay. Journal of Earthquake Engineering and Structural Dynamics 2008; 37(10):1269–1293.
16. Bonnet PA, Williams MS, Blakeborough A. Evaluation of numerical time-integration schemes for real-time hybrid
testing. Journal of Earthquake Engineering and Structural Dynamics 2008; 37(13):1467–1490.
17. Chen C, Ricles JM, Marullo TM, Mercan O. Real-time hybrid testing using the unconditionally stable explicit
CR integration algorithm. Journal of Earthquake Engineering and Structural Dynamics 2009; 38(1):23–44.
18. Chen C, Ricles JM. Improving the inverse compensation method for real-time hybrid simulation through a dual
compensation scheme. Journal of Earthquake Engineering and Structural Dynamics 2009; 38(10):1237–1255.
19. Weng YT, Tsai KC, Chen PC, Chou CC, Chan YR, Jhuang SJ, Wang YY. Seismic performance evaluation of a
34-story steel building retrofitted with response modification elements. Journal of Earthquake Engineering and
Structural Dynamics 2009; 38(6):759–781.
20. Chung YL, Nagae T, Fukuyama K, Kajiwara K, Inoue T, Hitaka T, Nakashima M. Seismic resistance capacity
of high-rise buildings subjected to long-period ground motions—E-Defense shaking table test. Proceeding of the
14th World Conference on Earthquake Engineering, Beijing, China, 2008; No. 95.
21. Ogawa N, Ohtani K, Katayama T, Shibata H. Construction of a three-dimensional, large-scale shaking table and
development of core technology. Philosophical Transactions of the Royal Society of London 2001; 359:1725–1751.
22. Nakashima M. Roles of large structural testing for the advance of earthquake engineering. Proceedings of the
14th World Conference on Earthquake Engineering, Beijing, China, 2008; K009.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe
622 Y.-L. CHUNG ET AL.
23. Ji X, Kajiwara K, Nagae T, Enokida R, Nakashima M. A substructure shaking table test for reproduction of
earthquake responses of high-rise buildings. Journal of Earthquake Engineering and Structural Dynamics 2009;
38(12):1381–1399.
24. Chung YL, Nagae T, Hitaka T, Nakashima M. Seismic resistance capacity of high-rise buildings subjected
to long-period ground motions: E-Defense shaking table test. Journal of Structural Engineering 2010; 136(6):
637–644.
25. Nakashima M, Inoue K, Tada M. Classification of damage to steel buildings observed in the 1995 Hyogoken–
Nanbu earthquake. Journal of Engineering Structures 1998; 20(4–6):271–281.
26. American Institute of Steel Construction, Inc. Seismic provisions for structural steel buildings, March 2005.
27. Engelhardt MD, Husain AS. Cyclic loading performance of welded flange-bolted web connections. Journal of
Structural Engineering 1993; 119(12):3537–3550.
28. Lee SJ, Lu LW. Cyclic tests of full-scale composite joint subassemblages. Journal of Structural Engineering
1989; 115(8):1977–1998.
29. Hajjar JF, Leon RT, Gustafson MA, Shield CK. Seismic response of composite moment-resisting connections.
II: behavior. Journal of Structural Engineering 1998; 124(8):877–885.
30. Leon RT, Hajjar JF, Gustafson M. Seismic response of composite moment-resisting-resisting connections.
I: performance. Journal of Structural Engineering 1998; 124(8):868–876.
31. Civjan SA, Engelhardt MD, Gross JL. Slab effects in SMRF retrofit connection tests. Journal of Structural
Engineering 2001; 127(3):230–237.
32. Nakashima M, Matsumiya T, Suita K, Zhou F. Full-scale test of composite frame under large cyclic loading.
Journal of Structural Engineering 2007; 133(2):297–304.
33. Abe K. Test and simulation of complete collapse of structures. Joint Research Report of NIED and Kyoto
University, March 2009; 50 (in Japanese).
34. Yamada S, Kitamura Y, Suita K, Nakashima M. Experiment investigation on deformation capacity of beam-
to-column connections in early highrise buildings by fullscale tests. Journal of Structural and Construction
Engineering, Architectural Institute of Japan 2008; 623:119–126 (in Japanese with English abstract).
35. Kitamura Y, Hashida I, Suita K. Seismic performance and retrofit of beam-column connection for early high-rise
buildings, part 1: experimental method and hysteresis characteristics. Summaries of Technical Papers of Annual
Meeting, Architectural Institute of Japan, Hiroshima, Japan, 2008; 1001–1002 (in Japanese).
36. Kim YJ, Oh SH, Moon TS. Seismic behavior and retrofit of steel moment connections considering slab effects.
Journal of Engineering Structures 2004; 26(13):1993–2005.
37. Engelhardt MD, Fry GT, Jones SL, Venti M, Holliday SD. Experimental investigation of reduced beam
section connections with composite slabs. Proceedings; Fourth US–Japan Workshop on Steel Fracture Issues,
San Francisco, 28 February–1 March 2000.
38. Uang CM, Yu QS, Noel A, Gross J. Cyclic testing of steel moment connections rehabilitated with RBS or welded
haunch. Journal of Structural Engineering 2000; 126(1):057–068.
Copyright 䉷 2010 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2011; 40:605–622
DOI: 10.1002/eqe