You are on page 1of 22

Seismic Performance of Ductile Steel

Moment-Resisting Frames Subjected to


Multiple Strong Ground Motions
Randy Tenderan,a) Takanori Ishida,a) Yu Jiao,b) and Satoshi Yamadaa)

This study evaluates the seismic performance of steel moment-resisting


frames (SMRFs) under multiple strong ground motions. The cumulative damage
of beam members is used as the main damage index. A cumulative damage for-
mula is generated based on the experimental results of the steel beam-to-column
connection test considering the ductile fracture. Local buckling of members is not
considered in this study. Six SMRF models with two parameters (the number of
stories and the strength of the column base) are analyzed by conducting an inelas-
tic response analysis. Three different ground motion intensities (peak ground
velocity = 0.5 m/s (design level), 0.75 m/s, and 1.0 m/s), each with five repeated
excitations are used in the inelastic response analysis to simulate the occurrence
of multiple strong ground motions. Stable behavior with a linear increment in
cumulative damage is found in most cases, especially when the ground motion
intensity is equal to the design level. However, when the intensity is greater than
the design level, both ductile fracture and weak story collapse are observed in
several cases. [DOI: 10.1193/111217EQS235M]

INTRODUCTION
During an earthquake event, multiple shakings are commonly observed rather than just a
single main shock. As noted by Li and Ellingwood (2007), in several events worldwide, the
main shock was found to be followed by a considerably large aftershock, e.g., in Italy (Friuli
1976, Umbria-Marche 1997), Greece (1986, 1988), Turkey (1992), and Mexico (1993, 1994,
1995). A recent event in Japan, the 2016 Kumamoto Earthquake, consisted of two strong
shakings: the foreshock on April 14 and the main shock on April 16. Both shocks had a
seismic intensity of 7 according to the Japan Meteorological Agency (JMA) scale (or
approximately equivalent to intensity X–XII on the Mercalli scale) in the most damaged
area, Mashiki Town. That shock event led structural engineers and researchers to ask
how safely the buildings are designed using current seismic codes in response to seismic
sequence events. The current Japanese seismic design code ensures that a building will
be able to sustain one severe earthquake without collapse; however, its performance
under multiple severe earthquakes must be analyzed further.
Some researchers have focused on analyzing the influence of the main shock–aftershock
sequences or repeated earthquakes on the structural performance. However, such studies

a)
Tokyo Institute of Technology, Mail Box J2-21, 4259 Nagatsuta-cho, Midori-ku, Yokohama 226-8503, Japan;
Email: ishida.t.ae@m.titech.ac.jp (T. I.)
b)
Tokyo City University, 1-28-1 Tamazutsumi, Setagaya-ku, Tokyo 158-8557, Japan

289
Earthquake Spectra, Volume 35, No. 1, pages 289–310, February 2019; © 2019, Earthquake Engineering Research Institute
290 TENDERAN ET AL.

typically use single-degree-of-freedom models (Iancovici and Ionică 2007, Hatzigeorgiou


and Beskos 2009, Hatzigeorgiou 2010, Zhai et al. 2013, Zhai et al. 2014, Zhai et al.
2016, and Kojima and Takewaki 2016). To the best of the authors’ knowledge, only a
few researchers have analyzed a multi-degree-of-freedom system of steel moment-resisting
frames (SMRFs; Lee and Foutch 2004, Fragiacomo et al. 2004, Li and Ellingwood 2007,
Ruiz-García and Negrete-Manriquez 2011, Loulelis et al. 2012). Most of the researchers
mentioned have used the inelastic ductility demand (μ), behavior factor (q), force reduction
factor (R), maximum roof displacement, or maximum interstory drift ratio to measure seismic
performance. Those parameters are mostly related to design and do not directly represent the
actual damage state of the structure. In addition, none of these researchers have focused on
identifying the seismic behavior of SMRFs under multiple earthquakes.
The seismic design of SMRFs commonly adopts a plastic design concept, especially for
the ultimate state. In other words, during severe earthquakes, damage is allowed to occur, and
the seismic energy will be dissipated by damaging the designated members. In the case of
multiple strong shakings, more seismic energy must certainly be dissipated, which in turn
further damages the structural members. Loulelis et al. (2012) have attempted to quantify
the damage index (DI) using the Park and Ang (1985) and Krawinkler and Zohrei (1983)
equations. The Park and Ang (1985) DI is defined by considering the combination of the
maximum displacement and the total energy dissipated; however, that equation was
originally defined for a reinforced concrete structure; thus it might not be suitable for a
steel structure. Meanwhile, the Krawinkler and Zohrei (1983) DI considers the low-
cycle fatigue phenomenon that occurs in a steel member. In this study, a similar DI is
used. However, the number of cycle to failure–beam’s rotation formula was generated
from a recent beam-to-column connection test that focused on the ductile fracture failure
mode. In that test, the specimens were designed to fail via fracture by preventing the occur-
rence of local buckling, and the specimens were also made using recently developed detail-
ing, materials, and welding conditions, which had been improved in the past decades to
prevent brittle fracture.
In this study, six SMRF models are analyzed by conducting a series of inelastic
response analyses that simulate the case of multiple earthquakes. The models consist of
two parameters: the number of stories and the strength of the column base. The number
of stories is taken as one of the parameters because models with a different number of
stories are expected to have different natural periods and characteristic dissipated energy
distributions. In addition, the variation in the strength of the column base mainly deter-
mines whether slip behavior occurs. To the best of the authors’ knowledge, the effect
of the slip of the column base on the performance of SMRFs under multiple earthquakes
has not yet been discussed by researchers. The main purpose of this study is to evaluate the
seismic performance, including the collapse mechanism and beam cumulative damage of a
ductile SMRF. Note that the DI of the member is represented by the cumulative damage at
the beam end since the application of the strong-column–weak-beam concept in the design
results in the beams becoming the weakest members of the SMRFs. In addition, the struc-
tures analyzed in this study are ductile SMRFs, i.e., the structures are constructed using
highly ductile members, and any deterioration effects, e.g., from local buckling, are not
considered.
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 291

Figure 1. Test specimen.

STEEL BEAM-TO-COLUMN CONNECTION TEST UNDER CYCLIC LOADING


CONSIDERING DUCTILE FRACTURE FAILURE MODE
In this experiment, eight beam specimens with the same structural and connection details
were tested. The beam and connection were designed to ensure the failure mode would be
ductile fracture. Various cyclic loading histories were used as the main parameter of the test.
Moreover, the test results were used to generate a formula for calculating the beam’s damage
via ductile fracture.
The test specimen and connection details are shown in Figure 1. The beam has a wide
flange section of 600 mm (depth)  200 mm (width)  11 mm (web thickness)  17 mm
(flange thickness) SN490B steel grade (nominal yield strength is 325 N∕mm2 and nominal
tensile strength is 490 N∕mm2 ). The beam is connected to the column using an end plate–
type connection. The beam–to–end plate connection used full-penetration welds on the
flange and fillet welds on both sides of the web with weld access holes that conform to
JASS 6 (Architectural Institute of Japan 1996). Stiffeners are attached 120 mm, 220 mm,
and 320 mm from the end plate to prevent local buckling. Thus the failure will be controlled
by fracture. Tensile coupon tests of the flange and the web were conducted using JIS-1A
testing samples (Japanese Industrial Standard Committee 2011), and the material test results
are shown in Table 1.
The test setup is shown in Figure 2. A half-span beam with cantilever loading was used
to represent the double curvature bending characteristic of the beam under lateral/seismic
loading. The specimen was rotated 90° so that the beam was standing vertically and con-
nected to the column lying horizontally through the end plate using high strength steel bars
(nominal yield strength is 1,080 N/mm2). An oil jack was attached at the beam’s free end
to load it, and a screw jack was attached at the other end to receive the reaction force.

Table 1. Material test results

Section Yield strength (N∕mm2 ) Tensile strength (N∕mm2 ) Elongation (%)


Flange 349 520 29
Web 373 522 26
292 TENDERAN ET AL.

Figure 2. Test setup.

To prevent out-of-plane and torsional deformation of the beam, lateral supports were
attached at two positions: near the loading point and near the beam connection.
Cyclic loading histories used in the test are shown in Figure 3. Those loading histories
could typically be divided into three groups: constant, variable, and random displacement
amplitude loading. For constant amplitude loading, four peak-to-peak amplitudes were
used: 0.015 rad, 0.02 rad, 0.025 rad, and 0.03 rad. For variable and random amplitude load-
ing, one set was determined for each loading history, and this set would be repeated until
fracture occurred. The peak-to-peak rotation amplitude of variable loading Set A consisted of
two cycles of 0.015 rad, two cycles of 0.02 rad, and two cycles of 0.025 rad, while Set B
consisted of six cycles of 0.015 rad, four cycles of 0.02 rad, and two cycles of 0.03 rad.
The random loading set was generated from a response analysis of a four-story four-span
SMRF excited by the 2011 Tohoku Earthquake Iwaki Record East-West (EW) component
and the 1968 Tokachi-Oki Earthquake Hachinohe Record EW component. In that response
analysis, the rotation response of the first-story beam of each excitation was converted into a
random loading history named IWAKI and HACHI, respectively. To convert the rotation
response into loading history, the rotation response was first scaled by multiplying it by
a certain factor so that the maximum peak-to-peak amplitude was limited to 0.03 rad to con-
trol the loading to be in a similar amplitude range. Next, to make the loading process more
efficient, all cycles with a peak-to-peak amplitude less than 0.01 rad were removed because
the contribution of those cycles to the beam damage is relatively small.
Figure 4 shows the load-deformation relationship of each specimen. In addition, the cal-
culated value of fully plastic moment (M p ) and elastic stiffness (K e ) are shown in each graph,
as indicated by the horizontal and diagonal dashed line, respectively. The development of the
cracks of all specimens differ according to the rotation amplitude. Here, the specimens
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 293

Figure 3. Loading history.

subjected to constant amplitude loading are shown as an example. Figure 5 shows the state of
the specimen when the fracture occurred. Meanwhile, Figure 6 shows the progression of the
crack along the flange. The crack was studied by human observation. In all specimens,
the first crack was formed at the weld access hole, and then another crack was formed at
the flange weld toe. However, the progression is different according to the rotation amplitude.
In the case of peak-to-peak rotation amplitude of 0.03 rad, the cracks that started from the
weld toe and weld access hole connected to one another, causing a fracture in the flange.
Meanwhile, in the case of amplitude of 0.025 rad and 0.02 rad, both cracks that started from
the weld toe and the weld access hole are largely developed, but in the end, the crack that
originated from the weld toe caused the fracture. In the case of amplitude of 0.015 rad, only a
slight crack developed from the weld access hole, and the flange fractured from the crack that
originated from the weld toe. In Table 2, the number of cycles to fracture of the constant
amplitude case and the number of sets to fracture for the variable and random cases are
shown. The number of cycles to failure in constant amplitude case means the number of
cycles needed to reach an ultimate state, which was defined as the cycle when either
most of the flange section has been fractured or the peak moment has reduced to 90%
of the original maximum moment. Meanwhile, the number of sets to fracture corresponds
to the number of single loading sets performed until fracture occurred.
294 TENDERAN ET AL.

Figure 4. Load-deformation relationship.

By using the test results of constant amplitude loading cases, the N f  Δθb relationship
could be obtained, as shown in Figure 7 and Equation 1. Here, N f is the number of cycles to
failure, and Δθb is the peak-to-peak amplitude rotation angle of the beam:

N f ¼ 4.8  103  Δθ2.6


EQ-TARGET;temp:intralink-;e1;41;123

b (1)
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 295

Figure 5. Final state of fracture at beam flange under constant peak-to-peak rotation amplitude
loading: (a) 0.03 rad; (b) 0.025 rad; and (c) 0.015 rad.

Figure 6. Development of crack at beam flange under constant peak-to-peak rotation amplitude
loading: (a) 0.03 rad; (b) 0.025 rad and 0.02 rad; and (c) 0.015 rad.

Combining the N f  Δθb relationship with Miner’s Rule (Equation 2) establishes a


method to evaluate the beam cumulative damage to fracture. In Equation 2, D expresses
cumulative damage, and when D equals 1.0, the beam is considered fractured; ni is the num-
ber of cycles with peak-to-peak amplitude rotation angle-i (Δθbi ); and N f i is calculated
296 TENDERAN ET AL.

Table 2. Number of cycles and sets to fracture

Constant amplitude Number of cycles Variable/random Number of sets


case to failure amplitude case to fracture
0.015 rad 266 A 20
0.02 rad 126 B 10
0.025 rad 70 IWAKI 7
0.03 rad 44 HACHI 18

using Equation 1. In the case of random loading, the number of cycles (ni ) and corresponding
peak-to-peak amplitude rotation angle (Δθbi ) are counted using the rain-flow counting
method (Endo et al. 1974):
X ni
D¼ (2)
Nf i
EQ-TARGET;temp:intralink-;e2;41;476

To verify the reliability of this method, variable and random loading test results are used.
From each set of hysteresis data, the number of cycles is counted using the rain-flow counting
method, and then, the cumulative damage is calculated using Equations 1 and 2. The
verification result is shown in Figure 8. The maximum error between the experimental
and calculation results is 6%, which is for the IWAKI case, and the average error is approxi-
mately 4%. This result verifies the reliability of the method for calculating the cumulative
damage caused by random recorded input ground motions. In addition, Sato et al. (2017)
have evaluated the deformation capacity of beam-end connection under various steel grades,
beam sizes, shear span ratios, and connection details. It was found that the number of cycles
to failure−maximum rotation amplitude (N f  θmax ) relationship, in which θmax is defined as
half of Δθb , is not sensitive to the steel grade, beam size, shear span ratio, and connection
detail. Thus, the accuracy of this method to evaluate a model with different steel material and
constructional detail is assumed to be acceptable.

Figure 7. N f  Δθb relationship.


SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 297

Figure 8. Verification of the beam’s damage evaluation method.

Figure 9. Infinite uniform plane frame model.

ANALYTICAL MODEL
The infinite uniform plane frame model is used as the analytical model with a typical
story height of 3.5 m and span of 7 m, as illustrated in Figure 9. This plane frame model
assumes that an infinite number of plane frames with uniform size exist throughout the plane,
so that simply taking one bay of the frame could represent the whole response of the frame
because the response will be uniform in every bay. The weight per story per span is assumed
to be 400 kN. Columns and panel zones have square hollow sections with BCR295 grade
steel (nominal yield strength of 295 N∕mm2 and nominal tensile strength of 400 N∕mm2 ),
beams have wide flange sections with SN400 grade steel (nominal yield strength of
235 N∕mm2 and nominal tensile strength of 400 N∕mm2 ), and the column base is an
exposed-type of column base. The PΔ effect is considered in the analysis, and the damping
factor is assumed to be equal to 2% Rayleigh damping for the first and second mode.
298 TENDERAN ET AL.

Six types of SMRFs, named 3_13, 6_13, 9_13, 3_07, 6_07, and 9_07, are analyzed. These
frames consist of two main parameters: the number of stories and the ratio of the moment
capacities of the column base and the first story column. The first index of the
models’ names represents the number of stories. The last two indices indicate the moment
capacity ratio of column base to first story column, which is equal to 1.3 in the case of a
strong-type column base and 0.7 in the case of a weak-type column base. The first and second
mode natural periods of all models are shown in Table 3. Moreover, the members’ size and
column-to-beam moment ratio (M pc ∕M pg ) of all models are shown in Tables 4–6. The models
are designed using a plastic design method for severe earthquakes based on the Japanese
seismic design code. In the code, a severe earthquake approximately corresponds to
PGV (peak ground velocity) = 0.5 m/s. In addition, note that the strong-column–weak-beam
concept is applied and, according to the Building Standard Law of Japan (Building Center of
Japan 2013), the column-to-beam moment ratio (M pc ∕M pg ) should be greater than or equal
to 1.5.
The hysteresis loop characteristic of the members is based on the decomposition of
the skeleton part, Bauschinger part, and elastic unloading part, as illustrated in Figure 10
(Kato et al. 1973, Akiyama and Takahashi 1990). The skeleton part corresponds to the
load-deformation relationship under monotonic loading (Kato et al. 1973, Akiyama and
Takahashi 1990). Moreover, the Bauschinger part is modeled using the simplified model
proposed by Akiyama and Takahashi (1990). Last, the elastic unloading part is the unloading

Table 3. First and second mode natural periods of the models

Model T 1 (sec.) T 2 (sec.)

3_07 0.816 0.244


3_13 0.925 0.277
6_07 1.128 0.371
6_13 1.135 0.369
9_07 1.368 0.490
9_13 1.367 0.491

Table 4. Member sizes and column-to-beam moment ratio of the three-story model

3_07 3_13

Story Column Beam M pc ∕M pg Column Beam M pc ∕M pg

3 □□ − 375  13 H − 527  226 – □□ − 353  12 H − 493  211 –


 8  13  8  12
2 □□ − 375  13 H − 527  226 1.56 □□ − 353  12 H − 493  211 1.58
 8  13  8  12
1 □□ − 375  13 H − 527  226 1.52 □□ − 353  12 H − 493  211 1.53
 8  13  8  12
Note: □□, square hollow sections.
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 299

Table 5. Member sizes and column-to-beam moment ratio of the six-story model

6_07 6_13

Story Column Beam M pc ∕M pg Column Beam M pc ∕M pg


6 □□ − 420  14 H − 594  255 – □□ − 425  14 H − 602  258 –
 9  14  10  14
5 □□ − 420  14 H − 594  255 1.54 □□ − 425  14 H − 602  258 1.54
 9  14  10  14
4 □□ − 420  14 H − 594  255 1.51 □□ − 425  14 H − 602  258 1.51
 9  14  10  14
3 □□ − 477  16 H − 594  255 1.85 □□ − 470  16 H − 602  258 1.76
 9  14  10  14
2 □□ − 477  16 H − 664  284 1.56 □□ − 470  16 H − 653  280 1.56
 11  16  10  16
1 □□ − 477  16 H − 664  284 1.52 □□ − 470  16 H − 653  280 1.52
 11  16  10  16
Note: □□, square hollow sections.

Table 6. Member sizes and column-to-beam moment ratio of the nine-story model

9_07 9_13

Story Column Beam M pc ∕M pg Column Beam M pc ∕M pg

9 □□ − 436  15 H − 618  265 – □□ − 436  15 H − 618  265 –


 10  15  10  15
8 □□ − 436  15 H − 618  265 1.53 □□ − 436  15 H − 618  265 1.53
 10  15  10  15
7 □□ − 436  15 H − 618  265 1.51 □□ − 436  15 H − 618  265 1.51
 10  15  10  15
6 □□ − 507  17 H − 618  265 1.93 □□ − 509  17 H − 618  265 1.94
 10  15  10  15
5 □□ − 507  17 H − 710  304 1.54 □□ − 509  17 H − 713  306 1.54
 11  17  11  17
4 □□ − 507  17 H − 710  304 1.52 □□ − 509  17 H − 713  306 1.52
 11  17  11  17
3 □□ − 540  18 H − 710  304 1.65 □□ − 538  18 H − 713  306 1.63
 11  17  11  17
2 □□ − 540  18 H − 743  318 1.56 □□ − 538  18 H − 740  317 1.56
 12  18  12  18
1 □□ − 540  18 H − 743  318 1.52 □□ − 538  18 H − 740  317 1.52
 12  18  12  18
Note: □□, square hollow sections.
300 TENDERAN ET AL.

Figure 10. Decomposition of hysteresis loops.

Figure 11. Trilinear model of the beam and column.

part of the loops that is characterized by the same stiffness as the original elastic stiffness of
the member. The skeleton curve models of the columns and beams are shown in Figure 11, a
trilinear model considering the strain hardening effect of the members. Note that even though
local buckling is likely to occur in the column section, in the present analysis, that type of
deterioration effect is not considered, and the column is assumed to be strong enough to
maintain its restoring force without any deterioration. For the exposed-type column base,
the hysteresis model is shown in Figure 12. The model uses a form of the slip-type
model that considers the effect of the moment resistance from axial force (M n ) and strain
hardening of the anchor bolt (K p ; Yamada et al. 1997).

INPUT GROUND MOTION


For the input ground motion in the inelastic analysis, ten ground motion records are used,
as listed in Table 7. All the records are scaled based on the PGV value to control their input
energy. Three intensities are considered: PGV = 0.5 m/s (design level), 0.75 m/s, and 1.0 m/s.
The velocity response spectra of all records at the design level and the first mode natural
period of all models are plotted together in Figure 13. In every analysis, five repeated
excitations of the same input wave and same intensity are considered to simulate multiple
earthquake sequences. The reason for using the same wave and same intensity is the
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 301

Figure 12. Slip-type model of the exposed column base.

simplicity in presenting the analytical results, because the trend can be clearly shown.
Meanwhile, the limitation of the number of excitations (No. Exc.) to five excitations is
because the occurrence of more than five strong ground motions in a short period of
time is considered an extremely rare case. To simulate the real condition of the seismic
sequences between excitations, 30 sec. of zero acceleration was added to the original
input ground acceleration. In other words, during that 30 sec., the structure will be freely
vibrated until its response becomes constant (i.e., the structure stops vibrating).

ANALYTICAL RESULTS
The inelastic response is analyzed using the six models and ten input ground motions
with three different ground motion intensities, as mentioned in the previous sections, thus

Table 7. Ground motion records used in the analysis

No. Earthquake Mw Seismic record Component PGAa (m∕s2 ) PGV (m/s)


1 1940 Imperial Valley 6.95 El Centro NS 3.42 0.38
2 1952 Kern County 7.36 Taft EW 1.76 0.18
3 1968 Tokachi-Oki 8.2 Hachinohe EW 1.81 0.37
4 1989 Loma Prieta 6.93 Gilroy Array #3 90° 3.61 0.45
5 1994 Northridge 6.69 Newhall NS 5.78 0.97
6 Olive View NS 8.26 1.29
7 1995 Kobe 6.9 JMA Kobe NS 8.21 0.89
8 1999 Chi Chi 7.6–7.7 TCU129 EW 9.81 0.55
9 2011 Tohoku 9 JMA Sendai NS 4.10 0.54
10 2016 Kumamoto (Apr. 16) 7.3 KiK-net Mashiki EW 11.57 1.42
a
PGA, peak ground acceleration.
302 TENDERAN ET AL.

Figure 13. Velocity response spectra of the input ground motion at the design level (h ¼ 5%).

resulting in a total of 180 cases. However, because of space limitations, not all the results are
shown here. Instead, summaries and examples are presented.
The DI of the structure is represented by the cumulative damage of the critical beam (D),
which was calculated by the previously described method. The critical beam is the beam
element that has the largest cumulative damage; this beam also commonly has the largest
amount of dissipated energy. The analytical result is mainly presented in two major parts: the
response of the structure caused by ground motion excitations with an intensity equal to the
design level (PGV of 0.5 m/s) and greater than the design level (PGV of 0.75 m/s and
1.0 m/s).

GROUND MOTION INTENSITY: EQUAL TO DESIGN LEVEL (PGV 0.5 M/S)


Under design-level excitations (PGV of 0.5 m/s), the structures show a stable behavior,
even after five repeated strong ground shakings. Figure 14 shows one of the examples of
the response of the ductile SMRFs excited by multiple ground accelerations: it is a 6_13
model that is excited by the design level of the JMA Sendai North-South (NS) record. In
Figure 14a, the increment in both the maximum and residual story drift angle (SDA) is
relatively small, so the structure can still maintain its stability. Moreover, Figure 14b
shows the cumulative damage of critical beam (D) versus No. Exc. the hysteresis response
of the critical beam in some stages of excitation. Clearly, the cumulative damage increases
almost constantly, i.e., the relation between D and No. Exc. is approximately linear. In
terms of the hysteresis response of the beam, at first excitation, the skeleton part could
be observed, i.e., the beam’s rotation increases from elastic to inelastic condition. However,
usually, if starting from the third excitation, the skeleton could hardly be observed, and the
energy was predominantly dissipated by the Bauschinger part; thus, the beam’s rotation is
almost constant, as is the increment in the cumulative damage. Note that the constant incre-
ment in the cumulative damage is also related to the usage of the same input wave of the
five excitations.
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 303

Figure 14. PGV 0.5 m/s JMA Sendai NS excitation of the 6_13 model: (a) SDA; and (b) critical
beam’s cumulative damage.

In summary, the increment in maximum SDA is relatively small on average, i.e., approxi-
mately 0.005 rad, 0.0035 rad, and 0.003 rad for the three-, six-, and nine-story models,
respectively. Moreover, the residual SDA is less than 0.01 at the fifth excitation for all
cases. Regarding cumulative damage of the critical beam, it constantly increases in every
excitation. At the fifth excitation, the cumulative damage of all cases ranges from 2.3%
to 38.6%, and on average, the cumulative damages for three-, six-, and nine-story models
are 17.1%, 9.6%, and 8.3%, respectively.
Figure 15 shows a comparison of the critical beam’s cumulative damage for different
input ground motions. The cumulative damage results of the strong column base model
as excited by various input ground motions are shown as the examples. It can be seen
that the cumulative damage of the model with the same natural period is varied depending

Figure 15. Comparison of the cumulative damage of the strong column base model for different
input ground motions.
304 TENDERAN ET AL.

on the characteristic of the input ground motion. However, the increment in the cumulative
damage still shows the same linear trend and stable behavior, regardless of the input ground
motion.
A comparison of the critical beam’s cumulative damage related to the two parameters
of the models is shown in Figure 16. Figure 16a shows a comparison of the number of
stories; one of the examples of the analytical results, the strong column base model excited
by the design level of the JMA Sendai NS record, is shown. The three-story building
clearly has the largest cumulative damage, followed by six-story and then nine-story;
the main reason for this result is the concentration of dissipated energy. As shown in
Figure 17, in the three-story building, the energy is mostly dissipated by the first
story; however, for six- and nine-story buildings, the middle stories also absorbed a
fair amount of energy. The influence of the type of column base on the cumulative damage
of critical beam is shown in Figure 16b, illustrating the response of the three-story model
excited by the JMA Kobe NS record as an example. In the weak column base model, the

Figure 16. Comparison of the cumulative damage by the number of stories and the column base
type: (a) PGV 0.5 m/s JMA Sendai NS excitation of the strong column base model; and (b) PGV
0.5 m/s JMA Kobe NS excitation of the three-story model.

Figure 17. Distribution of dissipated energy of PGV 0.5 m/s JMA Sendai NS exc. of the strong
column base type model: (a) three-story; (b) six-story; and (c) nine-story.
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 305

column base is expected to slip, thus increasing the beam’s rotation in every excitation.
However, Figure 16b shows that the cumulative damage of the model using strong and
weak column bases are almost equal, and in all cases, the difference between the cumu-
lative damage at the fifth excitation of those two types is less than 8%, which is relatively
small. However, no definite trend is observed that indicates, e.g., whether the weak
column base model always suffered more damage or vice versa; the damage differs
for each combination of ground motion records and number of stories. Figure 18
shows the hysteresis response of the critical beam and the column base of the same
case shown in Figure 16b. At the first excitation, slip occurs; however, at the next excita-
tion, the column base’s maximum rotation angle does not significantly increase. In other
words, even though a weak column base is used, as long as the upper structure is ductile, a
similar stable behavior could be achieved.

GROUND MOTION INTENSITY: GREATER THAN DESIGN LEVEL


(0.75 M/S AND 1.0 M/S)
Under multiple excitations with PGV of 0.75 m/s and 1.0 m/s, stable behavior is still
observed in most cases, as shown in Figure 14. However, in several cases, collapse occurred
within the five repeated excitations. At PGV of 0.75 m/s, collapse occurred in 3 out of a total
of 60 cases, i.e., Hachinohe EW excitation of the 3_07 and TCU129 EW and Hachinohe EW
excitations of the 3_13 models. Moreover, at PGV of 1.0 m/s, collapse occurred in 12 of a

Figure 18. Hysteresis response of PGV 0.5 m/s JMA Kobe NS exc. of 3_07 model: (a) column
base; and (b) critical beam.
306 TENDERAN ET AL.

total of 60 cases, i.e., Gilroy Array #3 90 deg., Hachinohe EW, JMA Sendai NS, and Kik-net
Mashiki EW excitations of 3_07 model; Gilroy Array #3 90 deg., Hachinohe EW, JMA
Sendai NS, and Taft EW excitations of 3_13 model; Hachinohe EW excitation of 6_07
model; Hachinohe EW excitation of 6_13 model; Taft EW excitation of 9_07 model;
and Taft EW excitation of 9_13 model. The collapse cases can be divided into two categories:
local collapse via fracture of the critical beam and weak story collapse. The critical beam
is estimated to fracture when the cumulative damage (D) is equal to or greater than 1.0.
Moreover, weak story collapse is assumed to occur when the maximum SDA reaches
10% or higher. The average cumulative damages of the critical beam of the uncollapsed
cases at fifth excitation are 37.2%, 21.5%, and 17.7% for the three-, six-, and nine-story
models, respectively, with PGV of 0.75 m/s and 61.3%, 39.3%, and 30.1% for the
three-, six-, and nine-story models, respectively, with PGV of 1.0 m/s.
Figure 19 shows an example where the critical beam is estimated to be fractured, i.e.,
the 6_13 model excited by the Hachinohe EW record with PGV of 1.0 m/s. Figure 19a
shows that the maximum and residual SDA of the building continue to increase signifi-
cantly as No. Exc. increases until the structure starts to lose its stability. This result is
demonstrated by the relationship between D versus No. Exc. shown in Figure 19b, which
is no longer linear; for comparison, the corresponding results with PGV of 0.5 m/s are
shown in the same graph. Figure 19c shows the hysteresis response of the critical

Figure 19. PGV 1.0 m/s Hachinohe EW excitation of the 6_13 model: (a) SDA; (b) comparison
of cumulative damage; and (c) critical beam’s hysteresis response.
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 307

beam at several stages of excitation, revealing that the skeleton part is observed at
every stage. In other words, in every stage of excitation, the beam’s rotation angle becomes
increasingly large so the cumulative damage increases exponentially instead of constantly.
Figure 20 shows an example where weak story collapse occurred, i.e., the 3_07 model
excited by the Hachinohe EW record with PGV of 0.75 m/s. Figure 20a illustrates the
collapse mechanism of the first story. At first excitation, the building remained stable,
and plastic hinges were formed at the beam end and column upper end of the first
story. However, at second excitation, the structure suffered large residual SDA, as
shown in Figure 20b. Finally, at third excitation, the maximum SDA reached over
10%, causing the story to completely lose its stability and collapse. Figure 20c shows
the story shear versus story drift relationship of the first story at the three stages of excita-
tion; the black circle indicates the peak story shear, which is located in the second excita-
tion. This figure shows that somewhere during the second excitation, the first story reaches
maximum capacity (indicated by the black circle in the graph) and starts to suffer a large
deformation until it finishes at approximately 0.05 rad of residual SDA at the end of second
excitation. Because the deterioration effect caused by the local buckling of members is not
considered in this case, the story shear capacity is almost constant, and only a small
decrease occurs via the P  Δ effect. Figure 20d shows the cumulative damage of the
critical beam, which exhibits a trend similar to that of the cumulative damage of the
beam-fractured case (Figure 19b).

Figure 20. PGV 0.75 m/s Hachinohe EW excitation of the 3_07 model: (a) weak story collapse
mechanism; (b) first SDA; (c) first story shear versus story drift; and (d) cumulative damage.
308 TENDERAN ET AL.

CONCLUSIONS
The seismic performance of ductile SMRFs under multiple earthquakes was evaluated in
the present study by considering the cumulative damage to fracture of the beam as the
main DI.
First, a steel beam-to-column connection test considering ductile fracture failure mode
under various cyclic loading cases (i.e., constant, variable, and random) was conducted to
establish the beam cumulative damage evaluation method. By using the test results of
constant amplitude loading cases, the number of cycle to failure–peak-to-peak amplitude
rotation angle of the beam (N f  Δθb ) equation was generated, and a method to evaluate
the beam cumulative damage to fracture was established by combining the equation with
Miner’s rule. This method was verified using the test results of variable and random
cases, and a good agreement was obtained that shows the reliability of the method to evaluate
the beam cumulative damage under the excitation of a real earthquake record.
Then an inelastic time history response analysis simulating the occurrence of multiple
earthquakes was conducted with the six SMRF models considering three different input
ground motion intensities (PGV of 0.5 m/s, 0.75 m/s, and 1.0 m/s), and the seismic
performance of those SMRF models was evaluated using the previously established
beam cumulative damage evaluation method. The findings of the analysis are summarized
as follows:
• In most cases, the structures were found to be able to dissipate the energy with stable
behavior under multiple earthquakes, which is indicated by the linear increment in
cumulative damage, particularly in the case in which the ground motion intensity is
equal to the design level (PGV of 0.5 m/s). However, when the ground motion
intensities are larger than the design level (PGV of 0.75 m/s and 1.0 m/s), collapse
might occur in several cases.
• The collapse could be a local collapse caused by the fracture of a critical beam or
a weak story collapse, both of which are related to the occurrence of large
residual SDA.
• In addition, the influence of the two model parameters (the number of stories and the
column base type) is identified. The three-story building is found to experience the
largest cumulative damage at the critical beam, followed by the six-story and then
the nine-story buildings because of the concentration of dissipated energy in a
certain story. Moreover, the influence of using either a weak or a strong column
base on the cumulative damage is not significant; as long as the upper structure
is ductile, similar stable behavior can be achieved.
However, the deterioration effects of the structural member from local buckling are not
considered in the present study. Overall, the results of this study could be a good benchmark
for further studies as representative of the upper limit of the seismic performance of SMRFs
under multiple earthquakes.

ACKNOWLEDGMENTS
This work is supported by Japan Society for the Promotion of Science KAKENHI Grant
Number JP17H01302. The experiments in this research are financially supported by
SEISMIC PERFORMANCE OF DUCTILE SMRFS SUBJECTED TO MULTIPLE STRONG GROUND MOTIONS 309

The Japan Iron and Steel Federation. The first author would also like to acknowledge the
Indonesia Endowment Fund for Education, Ministry of Finance of Republic of Indonesia,
which financially supports his study at Tokyo Tech, Japan.

REFERENCES
Akiyama, H., and Takahashi, M., 1990. Influence of Bauschinger effect on seismic resistance of
steel structures (in Japanese), Journal of Structural and Construction Engineering (Transactions
of AIJ) 418, 49–57.
Architectural Institute of Japan (AIJ), 1996. Japan Architectural Standard Specification JASS 6
Steel Work (in Japanese), Tokyo, Japan.
Building Center of Japan, 2013. Stipulation of horizontal load-carrying capacity calculation and
allowable stress and supplemental analysis calculation, in The Building Standard Law of
Japan: Notifications of Ministry of Land, Infrastructure, Transport, and Tourism
(MLIT) No. 594, CD-ROM, Tokyo, Japan, available at https://www.bcj.or.jp/en/services/
publication.html.
Endo, T., Matsuishi, M., Mitsunaga, K., Kobayashi, K., and Takahashi, K., 1974. Rainflow
method, the proposal and the applications (in Japanese), Bulletin of Kyushu Institute of
Technology 28, 33–26.
Fragiacomo, M., Amadio, C., and Macorini, L., 2004. Seismic response of steel frames under
repeated earthquake ground motions, Engineering Structures 26, 2021–2035.
Hatzigeorgiou, G. D., 2010. Behavior factors for nonlinear structures subjected to multiple
near-fault earthquakes, Computers and Structures 88, 309–321.
Hatzigeorgiou, G. D., and Beskos, D. E., 2009. Inelastic displacement ratios for SDOF structures
subjected to repeated earthquakes, Engineering Structures 31, 2744–2755.
Iancovici, M., and Ionică, G., 2007. Evaluation of the inelastic demand of structures
subjected to multiple ground motions, Journal of INTERSECTII/INTERSECTIONS
4(2), 143–154.
Japanese Industrial Standard Committee (JISC), 2011. Japanese Industrial Standard (JIS) Z 2241,
Metallic Materials—Tensile Testing—Method of Test at Room Temperature, Tokyo, Japan.
Kato, B., Akiyama, H., and Yamanouchi, H., 1973. Predictable properties of materials under
incremental cyclic loading, IABSE reports of the working commissions 13, 119–124.
Kojima, K., and Takewaki, I., 2016. A simple evaluation method of seismic resistance of
residential house under two consecutive severe ground motions with intensity 7, Frontiers
in Built Environment 2, 15.
Krawinkler, H., and Zohrei, M., 1983. Cumulative damage in steel structures subjected to
earthquake ground motions, Computers and Structures 16, 531–541.
Lee, K., and Foutch, D. A., 2004. Performance evaluation of damaged steel frame buildings
subjected to seismic loads, Journal of Structural Engineering 130, 588–599.
Li, Q., and Ellingwood, B. R., 2007. Performance evaluation and damage assessment of steel
frame buildings under main shock–aftershock earthquake sequences, Earthquake Engineering
and Structural Dynamics 36, 405–427.
Loulelis, D., Hatzigeorgiou, G. D., and Beskos, D. E., 2012. Moment resisting steel frames under
repeated earthquakes, Earthquakes and Structures 3, 231–248.
Park, Y. -J., and Ang, A. H. -S., 1985. Mechanistic seismic damage model for reinforced
concrete, Journal of Structural Engineering 111, 722–739.
310 TENDERAN ET AL.

Ruiz-García, J., and Negrete-Manriquez, J. C., 2011. Evaluation of drift demands in existing steel
frames under as-recorded far-field and near-fault mainshock–aftershock seismic sequences,
Engineering Structures 33, 621–634.
Sato, R., Kishiki, S., and Yamada, S., 2017. Cyclic rotation capacity of beam-to-column
connections using various steel grades, Paper No. 3644, in Proceedings of the 16th World
Conference on Earthquake Engineering, 9–13 January, 2017, Santiago, Chile.
Yamada, S., Akiyama, H., and Sadamoto, M., 1997. Influence of the elasto-plastic behavior of
column bases with slip-type hysteresis characteristics on the ultimate earthquake resistance of
multi-story steel moment frames (in Japanese), Journal of Structural and Construction
Engineering (Transactions of AIJ) 62, 141–147.
Zhai, C., Ji, D., Wen, W., Lei, W., Xie, L., and Gong, M., 2016. The inelastic input energy spectra
for main shock-aftershock sequences, Earthquake Spectra 32, 2149–2166.
Zhai, C. -H., Wen, W. -P., Chen, Z., Li, S., and Xie, L. -L., 2013. Damage spectra for the
mainshock–aftershock sequence-type ground motions, Soil Dynamics and Earthquake
Engineering 45, 1–12.
Zhai, C. -H., Wen, W. -P., Li, S., Chen, Z., Chang, Z., and Xie, L. -L., 2014. The damage
investigation of inelastic SDOF structure under the mainshock–aftershock sequence-type
ground motions, Soil Dynamics and Earthquake Engineering 59, 30–41.
(Received 12 November 2017; Accepted 25 August 2018)

You might also like