You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260908778

Collapse Assessment of Steel Moment Frames Based on E-Defense Full-Scale


Shake Table Collapse Tests

Article in Journal of Structural Engineering · January 2013


DOI: 10.1061/(ASCE)ST.1943-541X.0000608

CITATIONS READS
42 519

1 author:

Dimitrios G. Lignos
École Polytechnique Fédérale de Lausanne
130 PUBLICATIONS 1,554 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Deterioration Modeling and Collapse Risk Assessment of Steel Concentrically Braced Frames View project

SERA-TA / NSFuse: Ductile steel fuses for the protection of critical nonstructural components View project

All content following this page was uploaded by Dimitrios G. Lignos on 20 March 2014.

The user has requested enhancement of the downloaded file.


Collapse Assessment of Steel Moment Frames Based
on E-Defense Full-Scale Shake Table Collapse Tests
Dimitrios G. Lignos, A.M.ASCE1; Tsuyoshi Hikino2; Yuichi Matsuoka3; and
Masayoshi Nakashima, M.ASCE4

Abstract: This paper presents key parameters that affect numerical modeling of steel frame structures for reliable collapse simulations. The
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

collapse assessment is based on experimental data obtained from a full-scale shaking table collapse test of a 4-story steel moment frame and
a blind numerical analysis contest that was organized in parallel with the collapse test. It is shown that (1) there is no clear advantage between
three-dimensional (3D) and 2D analyses in the prediction of a sidesway collapse mechanism for buildings with a regular plan view as in the case
of study; (2) the assumption of Rayleigh damping leads to better predictions of structural response compared with stiffness proportional
damping; and (3) accurate prediction of collapse necessitates that P-D effects always be considered in the analysis. It is also proven that
accurate simulation of steel component deterioration is a key factor for reliable prediction of collapse behavior. On the basis of a synthesis of
experimental and analytical studies, a few collapse mitigation alternatives are investigated. In particular, the effects of the strong-column/
weak-beam ratio and exposed base plates on the collapse capacity are assessed. It is notable that a combination of bending strength
increase and delay of local buckling in first-story columns is most effective for the enhancement of seismic performance against collapse.
DOI: 10.1061/(ASCE)ST.1943-541X.0000608. © 2013 American Society of Civil Engineers.
CE Database subject headings: Structural failures; Deterioration; Steel structures; Buckling; Beam columns; Simulation; Tests.
Author keywords: Collapse; Component deterioration; Full-scale collapse tests; Steel structures; Local buckling; Strong-column/weak-beam ratio;
Exposed column bases; Collapse simulation; Sidesway collapse.

Introduction steel structures through collapse. These modeling and simulation


tools should be validated with realistic experimental data of steel
Steel building structures are very popular in highly seismic countries components and systems that have been tested to complete failure.
around the world such as the United States and Japan. Although the Until recently, there was no physical experiment that could
design and construction of such structures suffered a serious setback demonstrate the redundancy, connection performance, and collapse
after the disclosure of many fractures in welded beam-to-column behavior of steel structures under severe dynamic shaking. A few
connections in the 1994 Northridge and 1995 Hyogoken-Nanbu notable experimental studies in recent years aimed to evaluate (1) the
(Kobe) earthquakes, numerous studies have been conducted to design limits for minimizing P-D effects (Vian and Bruneau 2001);
overcome these problems, and improved alternatives have been (2) the effect of connection fractures causing substantial strength loss
offered and placed into practice. With these efforts, the concept of on the seismic capacity of structures (Rodgers and Mahin 2006); and
performance-based seismic design (Borzognia and Bertero 2006) (3) the combined effect of component deterioration and P-D on the
received attention, and a critical need was recognized for the collapse capacity of steel moment frames (Lignos et al. 2011).
quantification of collapse potential of steel structures under extreme However, because of the small scale and simplified assumptions of
earthquakes. To this end, reliable numerical models and simulation the test specimens used in those studies, the actual redundancy of real
techniques are needed that permit the performance evaluation of beam-to-column connections, including composite action, the con-
tributions of nonstructural components, and the effect of three-
1
Assistant Professor, Dept. of Civil Engineering and Applied Mechanics, dimensional (3D) dynamic loading on the collapse behavior of a
McGill Univ., Montreal, QC, Canada H3A 2K6 (corresponding author). steel structure, could not be assessed. Nakashima et al. (2006) con-
E-mail: dimitrios.lignos@mcgill.ca ducted a large-scale test to evaluate the seismic performance of a full-
2
Manager, Steel Structures Engineering Division, Nippon Steel Engi-
scale 3-story steel structure at large deformations. However, the effect
neering Co., Ltd., Osaka Center Building, 1-5-1 Osaki, Shinagawa-Ku,
Tokyo 141-8604, Japan; formerly, Researcher, E-Defense, National of dynamic loading on the overall performance of the structure could
Research Institute for Earth Science and Disaster Prevention (NIED). not be assessed because a predetermined quasi-static loading was
3
Manager, Steel Structures Engineering Division, Nippon Steel Engi- applied in these tests.
neering Co., Ltd., Tokyo, Japan; formerly, Researcher, E-Defense, National During September 2007, a full-scale shaking table collapse test
Research Institute for Earth Science and Disaster Prevention (NIED). of a 4-story steel structure was conducted at the E-Defense facility
4
Professor, Disaster Prevention Research Institute (DPRI), Kyoto Univ., in Japan (Suita et al. 2008). The structure was designed based on
Uji, Gokasho, Uji, Kyoto 611-0011, Japan. current Japanese seismic provisions [Architectural Institute of Japan
Note. This manuscript was submitted on March 22, 2011; approved on
(AIJ) 2006; Building Center of Japan (BCJ) 2008]. This test offered
March 7, 2012; published online on March 10, 2012. Discussion period
open until June 1, 2013; separate discussions must be submitted for an unprecedented test data set on the collapse with a sidesway first-story
individual papers. This paper is part of the Journal of Structural Engi- collapse mechanism. This failure mode occurred because the increased
neering, Vol. 139, No. 1, January 1, 2013. ©ASCE, ISSN 0733-9445/ strength of beams caused by strain hardening and deterioration in
2013/1-120e132/$25.00. column strength were not allowed for, even when the strong-column/

120 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


weak-beam (SCWB) criterion was used in accordance with AIJ These periods were obtained from 3D white noise tests conducted
(2006). In parallel with the testing program, an international blind prior to the main tests. The damping ratios for the first mode of vi-
analysis competition was carried out. The objective of the compe- bration were 2.1 (X) and 2.3% (Y) (Yamada et al. 2009).
tition was to assess efficient computational methods and techniques
for accurate numerical simulation of the seismic response of steel
structures subjected to design level and extreme earthquakes. Testing Protocol
The main objective of this paper is to address key factors that
are important for the reliable collapse assessment of steel frame The loading protocol for dynamic shaking of the test structure
structures subjected to strong earthquakes. This assessment is based consisted of progressively increased ground motion intensities of the
on (1) state-of-the-art component deterioration models that are JR Takatori motion recorded during the 1995 Kobe earthquake. All
validated by comparison with the experimental responses of the full- three components (two horizontal and one vertical) of this ground
scale collapse test of the 4-story steel structure through collapse and motion were applied simultaneously. The test specimen was sub-
(2) a statistical evaluation of the blind analysis competition with jected to a sequence of 20 (Level 1), 40 (Level 2), 60, and 100% of
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

emphasis on collapse. The second objective of this paper is to use the original JR Takatori record. The east-west (EW) component was
the validated deterioration model of the test structure to investigate assigned to the Y-direction of the frame, and the north-south (NS)
three collapse mitigation strategies: increase of the column strength, component was assigned to the X-direction. The two ground motion
increase of the column plastic deformation capacity, and column components have comparable spectral acceleration values at the
base plates with controlled anchor bolt failure. fundamental period of the structure in the Y-direction (T1 5 0:80 s).
However, the EW component of the ground motion was the
strongest because of its high frequency content (see range of periods
Outline of Full-Scale Collapse Test from 0 to 0.5 s); thus, it caused more damage in this direction than in
the X-direction. However, Nam and Kasai (2011) demonstrated that
A two-by-one bay full-scale 4-story steel structure was tested to the same structure would collapse at 100% of the JR Takatori motion
collapse at the E-Defense shaking table facility in September 2007. regardless of the orientation of the ground motion. For the Y-EW
E-Defense is an earthquake testing facility that includes the world’s component of the original record, peak ground acceleration
largest shaking table of 20 3 15 m (plan dimensions). A number of (PGAÞ 5 6:57 m=s2 and peak ground velocity ðPGVÞ 5 1:27 m=s.
large-scale structures have been tested in the past at this facility For 60% JR Takatori, the spectral acceleration Sa at the first mode
(Nakashima 2008; Ji et al. 2009; Chung et al. 2010; Sato et al. 2011). period of the test structure in the Y loading direction was
Fig. 1(a) shows the test structure after completion of its construction Sa ð0:8, 5%Þ 5 8:43 m=s2 for 5% damping, which is more than two
on the shaking table. The dimensions of the test structure were times higher that the PGA of the 60% record. The 5% damping
10 3 6 m in the longitudinal (Y) and transverse directions (X), acceleration spectrum is used so that it can directly be compared with
respectively, as shown in Figs. 1(a and b). The story height of the a design spectrum. This value of Sa corresponds approximately to
structure was 3.5 m excluding the first story, which was 3.875 m. An a maximum considered earthquake (MCE) in San Diego, California
elevation view of the steel moment resisting frame in the longitu- (32.80201N, 117.16765W), assuming Site Class D according to
dinal direction is shown in Fig. 1(c). The wide flange beams of the ASCE (2010). All three components of the unscaled JR Takatori
test structure ranged from 340 to 400 mm in depth (Table 1) and were record are shown in Fig. 3. The unscaled ground motion intensity
fabricated from Japanese SN400B structural steel with nominal yield (100%) was used to quantify the margin against collapse of the steel
strength of 235 MPa. The measured material properties (yield structure subjected to a very severe earthquake.
strength sy and ultimate strength su ) for both the flange and web of
the steel wide flange beams are summarized in Table 2. These
properties were based on the average of six coupon tests. The steel Seismic Performance of the 4-Story Steel Structure
columns were 300 3 9 hollow square sections (HSS) and were
fabricated with Japanese BCR295 structural steel with a nominal In the test, the two-bay moment frame arranged in the Y-direction
strength of 295 MPa. The measured material properties were sustained much larger deformations and damage than the one-bay
sy 5 330 MPa and su 5 426 MPa for the first-story column and moment frame in the X-direction. For this reason, the response of the
sy 5 332 MPa and su 5 419 MPa for the rest of the columns. These two-bay moment frame is examined in detail. Detailed information
properties were obtained based on the average of two coupon tests. regarding the response of the one-bay frame is discussed in Suita
Welding details with no weld access hole were adopted for all the et al. (2008) and Yamada et al. (2009). The test structure behaved
beam-to-column connections in accordance with the Japanese elastically during a Level I (20%) earthquake. Absolute peak story
fabrication standard JASS6 (AIJ 1996). An example of the second- drift ratios (SDRs) did not exceed 0.5% rad along the height of the
floor steel beam-to-column connection is shown in Fig. 2(a). Be- test structure (Fig. 4). During a Level 2 (40%) earthquake, yielding
cause of the 175-mm-thick concrete slab placed at each floor of the occurred at the base of the interior first-story column and the interior
structure, full composite action was expected between the steel panel zones of the second and third floor beam-to-column con-
beams and concrete slab as a result of adequate shear stud design. nections. However, all beams behaved elastically because of the
The test structure was located in the center of the shaking table [see composite action and their large measured yield strength (Table 2).
Fig. 1(a)], and each column base was connected to steel foundation From Fig. 4, the first and second stories of the test structure reached
beams, which were pretensioned to the shaking table. Exposed base about 1% rad in both loading directions.
plate connections with sufficient rigidity were designed to guarantee During the 60% earthquake, plastification occurred at the interior
fixed conditions at the base. A typical example of the base plate and right exterior second and third floor panel zones. The interior
connection is shown in Fig. 2(b). Around the perimeter of the test first-story columns also yielded at the top and bottom locations. This
structure, autoclaved lightweight concrete (ALC) panels were in- resulted in an overall multistory sideway yielding mechanism. Peak
stalled in both loading directions except for one side in the longi- SDRs along the height of the test structure are superimposed in Fig. 4
tudinal direction [see Fig. 1(a)]. The predominant periods of the test for both loading directions. The first-story drift ratio reached about
specimen were 0.80 and 0.76 s in the X- and Y-directions, respectively. 2% rad, and SDRs progressively decreased in the upper stories for

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 121

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Full-scale 4-story steel structure: (a) test setup after completion on the E-Defense shaking table; (b) plan view of the test structure; (c) steel
moment-resisting frame in the longitudinal (Y) direction of interest (units: millimeters)

the Y loading direction [see Fig. 4(b)]. The ALC panels and partition deteriorated in strength and stiffness during the 60% JR Takatori
walls of the building (nonstructural components) were severally earthquake.
damaged in the same direction (Matsuoka et al. 2008). After the The test structure collapsed with a first-story sideway mechanism
completion of the 60% JR Takatori earthquake, a residual SDR of during the 100% JR Takatori earthquake. The elapsed time until the
0.3% rad was notable in the first story, primarily caused by yielding complete collapse occurred was 6.57 s (counting from the beginning
of the first-story columns at the base of the test structure. However, of the 100% JR Takatori record until the time that the first-story drift
this residual drift had a negligible effect on the collapse capacity ratio of the test structure reached 15% rad). This was the time that the
of the test structure because none of its structural components test structure lost its lateral resistance (base shear equal to zero). The

122 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


Table 1. Beam and Column Steel Sections of the 4-Story Steel Structure
Beam sections
Column sections
Story C1, C2 (BCR295) Floor G1 (SN400B) G11 (SN400B) G12 (SN400B)
Roof H-346 3 174 3 6 3 9 H-346 3 174 3 6 3 9 H-346 3 174 3 6 3 9
4 HSS 300 3 9
4 H-350 3 175 3 7 3 11 H-350 3 175 3 7 3 11 H-340 3 175 3 9 3 14
3 HSS 300 3 9
3 H-396 3 199 3 7 3 11 H-400 3 200 3 8 3 13 H-400 3 200 3 8 3 13
2 HSS 300 3 9
2 H-400 3 200 3 8 3 13 H-400 3 200 3 8 3 13 H-390 3 200 3 10 3 16
1 HSS 300 3 9
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

first-story collapse mechanism of the test specimen is shown in Table 2. Measured Material Properties of Steel Beams of the 4-Story Steel
Fig. 5(a). Final resultant absolute SDRs in the first story reached 19% Structure
when the test structure rested on the safeguard system, which was Measured material properties
placed around the structure to protect the shaking table [Figs. 1(a)
and 5(a)]. In reference to the peak SDR profiles for all the scaled Wide flange section Location sy ðN/mm2 Þ su ðN/mm2 Þ
intensities, it is notable that there was a transition from the overall H-340 3 175 3 9 3 14 Flange 309 443
sideway yielding mechanism to a first-story collapse mechanism. Web 355 468
The primary reason for this transition was the severe strength H-346 3 174 3 6 3 9 Flange 333 461
deterioration of the first-story columns caused by local buckling at Web 382 483
the base [Fig. 5(b)] and top [Fig. 5(c)] locations. This caused re- H-350 3 175 3 7 3 11 Flange 302 441
distribution of moments in the steel components of the test spec- Web 357 466
imen. Strength deterioration occurred after the first-story columns H-390 3 200 3 10 3 16 Flange 297 451
exceeded 1% rad in rotation. This is confirmed in Fig. 6, which Web 317 458
shows the moment rotation diagram of the first-story interior column H-396 3 199 3 7 3 11 Flange 311 460
at the bottom and top in the Y loading direction. This agrees with the Web 369 486
information gathered from a recently developed steel HSS tubular H-400 3 200 3 8 3 13 Flange 326 454
database for deterioration modeling of steel columns (Lignos and Web 373 482
Krawinkler 2009, 2010). The implication is that even if the SCWB
criterion was satisfied during the design of the test structure, steel
columns were not designed for the increased forces caused by strain numerical model input parameters were fixed for all groups that
hardening after yielding of the panel zones at the second floor; thus, participated in the competition. The second step of the blind analysis
yielding and subsequently local buckling at the top of the first-story competition was needed because the actual table motion was known
columns occurred. Strength deterioration of the first-story columns only after the completion of the experiment. Furthermore, the mea-
resulted in unloading of the second and third floor panel zones, sured material properties of concrete slabs were obtained only a
which yielded during the 60% JR Takatori earthquake (see Fig. 7). few days prior to the test. The evaluation criteria of the competition
After the first-story drift ratio of the test specimen exceeded 15% rad mostly involved seismic responses of the 60% JR Takatori. For the
(see Fig. 8), i.e., collapse occurred, the test structure was rested on collapse level, part of the competition evaluation was the time at
the safeguard system. which the structure would reach 13% rad in any of its stories.
Damage progressed similarly in the one-bay frame arranged in In total, 47 groups of researchers and practicing engineers from
the X-direction, although peak SDRs were 25e60% smaller than seven different countries participated in the competition. Thirty of
those obtained in the two-bay frame [see Fig. 4(a)]. The principal those groups were researchers in various universities. The remaining
axis of JR Takatori was included by ∼45, which appeared to have 17 groups were engineers from different design firms. Among 47
resulted in some reduction in the end column bending capacity as groups, 30 groups used 3D analytical tools, and the remaining 17
a result of biaxial bending. However, throughout the shaking, tor- groups used 2D analysis tools. Seventeen groups used commercially
sional response remained minimal (Suita et al. 2008). The vertical available software, whereas the rest used research or personal soft-
component of JR Takatori was relatively small, and the variation of ware. The groups were classified into four ranks (Ranks 1e4) based
the axial load at the end columns of the test structure was primary on the total score S that they obtained (Ohsaki et al. 2008a, b). The
caused by the overturning moment. In summary, the effect of vertical k-means method (Hartigan and Wong 1979) was used to obtain
motion was secondary [Sa ðT1 5 0:80 sÞ for the vertical motion was the total score. A group that was classified as Rank 1 obtained the
0.26g for the unscaled motion as seen from Fig. 3]. highest total score S compared with the other groups that were
classified as Rank 2, 3, or 4.
Blind Analysis Competition Outline
Statistical Investigation of Blind Analysis Results
In conjunction with the E-Defense shaking table collapse test,
a blind analysis competition was carried out. The objective of this Table 3 compares the mean m, the SD s, the coefficient of var-
competition was to assess the capabilities of numerical models to iation (COV), and the maximum and minimum of the RMS errors
predict the seismic response of the steel moment frames during between experimental and simulated results of the absolute maxi-
design level earthquakes and near collapse. The blind analysis com- mum SDR and shear force in the first story of the test structure. From
petition was conducted in two steps. During the first step, the this table, it is notable that a 2D analysis is as good as a 3D analysis

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 123

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Typical design details for a steel beam-to-column connection and column base of the 4-story steel structure (units: millimeters)

for prediction of the responses of steel moment frames that undergo


minimal torsional effects (MacRae 1999; MacRae and Mattheis
2000). A 2D analysis can accurately capture the seismic behavior
of the moment frame in the Y loading direction in terms of the SDR
with a smaller s and COV compared with a 3D analysis. However,
the latter predicts the base shear force of the test structure more
accurately.
Fig. 9 summarizes major parameters that affected the numerical
simulations. Results are categorized with respect to the type of analysis
(2D versus 3D) and the type of investigators (researchers versus
engineers). In Fig. 9, N/A means that the participants did not nec-
essarily report details for individual parameters, because the compe-
tition initially emphasized the predicted response of various numerical
models rather than the parameters that could affect the numerical
analysis. From Figs. 9(aef), the following observations are notable: Fig. 3. Acceleration spectra of the Y-NS, X-EW, and Z-Up compo-
• Twenty-nine groups used the information from component tests nents of the unscaled JR Takatori motion
on beam-to-column connections including the floor slab, which
were released prior to the collapse tests, for calibration of the of the test specimen, Hikino et al. (2009) noted a significant
hysteretic response of structural component numerical models. influence of this factor on the accuracy of responses. Further in-
From those groups, 14 were classified as Rank 1. Only three of 12 vestigation on this effect is underway for the selection of an ap-
groups that did not use the component information were classi- propriate ratio (e.g., Chung et al. 2010; Tai et al. 2010), but more
fied as Rank 1 [see Fig. 9(a)]. This illustrates the importance of examinations are needed for the quantification of this effect.
comprehensive data for analytical model calibrations. Also 10 of • From Fig. 9(e), 10 of 18 groups that considered P-D effects in the
12 groups in third place of each category adopted component nonlinear response history analysis were classified as Rank 1.
test results. From the same figure, because 25 groups were ranked within
• Similar to the component tests, 10 of 29 groups that used the material Ranks 1e4 and did not specify whether geometric nonlinearities
test results ranked within the first three places in all categories. were included in their analysis models, we are not able to make
• Nine of 10 groups that assumed Rayleigh damping were clas- a conclusive summary regarding P-D effects from the blind
sified as Rank 1. None of the groups that performed a 2D analysis analysis contest results. However, the consideration of geometric
and assumed a stiffness proportional damping were classified nonlinearity and component deterioration in collapse numerical
within Ranks 1e4 [see Fig. 9(b)]. From the same figure, four simulations is a key factor for reliable prediction of the seismic
groups that performed a 3D analysis were classified as Rank 4. response near collapse, as will be shown later.
This implies that the assumption of stiffness proportional damp- • From Fig. 9(f), seven groups that considered an integration time
ing did not lead to reliable results, as shown in Fig. 9(b). Over- step dt smaller than 0.001 s were classified as Rank 1. Reliable
damping of higher-mode responses, characteristic of stiffness analytical predictions were found feasible with 0:001 , dt
proportional damping, might be responsible, as also noted by , 0:01, which was confirmed by 10 groups. The majority of
other researchers (e.g., Liang and Lee 1991; Chopra 2007). the analysis groups used the implicit Newmark integration
• On the basis of Fig. 9(c), 11 groups assigned a damping ratio method. Moreover, about 300 degrees of freedom (DOF) were
z . 2%. The justification was that all the groups acknowledged adequate for a group to be ranked among the first three for 3D
the contribution of nonstructural components to damping. Six of analysis. There were five groups that used more than 1,000 DOF.
these groups were classified as Rank 1. Thirteen of 25 groups that
assumed a damping ratio equal to 2% were classified as Rank 2.
• Fig. 9(d) summarizes how many groups considered the compos- Validation of Component Deterioration Models
ite slab effects in the analysis. This effect was assessed by an for Collapse Simulations
amplification factor that varied from 1.25 to 2.0. This large
variation implies that there is no well-established procedure Because the blind analysis competition emphasized the seismic re-
for the estimation of this factor. Using a fiber numerical model sponse of the test structure at 60% JR Takatori, only 20 groups were

124 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Maximum story drift ratios along the height of the test specimen in the X and Y loading directions for all testing phases

Fig. 5. First-story collapse mechanism of the 4-story steel structure and associated local buckling in the first-story columns after 100% JR Takatori
record: (a) first-story collapse mechanism of the test specimen; (b) local buckling at the first-story column (base location); (c) local buckling at the first-
story column (top location)

concerned with explicit consideration of component deterioration. the results of the competition and in particular to quantify the
Severe local buckling of steel columns occurred during the 100% JR combined effect of component deterioration and P-D on the collapse
Takatori record, and this did not affect the predictions of these capacity of the steel moment frame. The Open System for Earth-
groups up to the 60% JR Takatori. However, about 20 groups could quake Engineering Simulation (McKenna 1997) platform was used
not predict collapse during the blind analysis competition (Hikino to model the two-bay steel moment frame. Steel beams and columns
et al. 2009). The importance of component deterioration for accurate were modeled with elastic beam column elements and concentrated
collapse simulation is shown later. The groups that considered plasticity rotational springs at their ends. These springs followed
component deterioration in their simulations based it on a trilinear a bilinear hysteretic response and included deterioration based on the
skeleton curve that could simulate strength but not cyclic de- modified Ibarra-Medina-Krawinkler deterioration model (Ibarra et al.
terioration (Ohsaki et al. 2008b). The first award winner modeled 2005; Lignos and Krawinkler 2009, 2011). This model is able to
steel beams and columns with fiber elements. Because the fiber simulate cyclic deterioration both in strength and stiffness and has been
element is not able to capture local buckling of steel columns implemented in the OpenSees analysis platform (Version 2.2.2.e).
(i.e., deterioration), this was modeled indirectly with a parallel The deterioration parameters for the steel HSS columns and com-
material that encapsulated the Menegotto-Pinto model for cyclic posite beams were determined from the moment rotation diagram of
hardening and a hysteretic material that deteriorates in strength but the associated component cyclic tests that were conducted prior to
not cyclic hardening. Ohsaki et al. (2008b) presents more details the shaking table tests and were distributed to the blind analysis
about this model. competition participants for calibration of their numerical models.
After the blind analysis competition, a series of analysis using Fig. 10(a) shows an example of the HSS 300 3 9 column that was
a 2D numerical model was conducted to assist the interpretation of tested with a symmetric cyclic loading protocol at a 45 angle to

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 125

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Moment rotation diagrams of the interior column at bottom and top locations in the first story of the 4-story steel structure

Fig. 7. Panel zone hysteretic response at the second floor beam during Fig. 8. First-story drift ratio history up to collapse of the test specimen
100% JR Takatori in the Y loading direction during 100% JR Takatori

quantify the effect of coupled axial load and biaxial bending on the
cyclic behavior of the steel column. The deduced moment rotation Table 3. Statistics of RMS for Maximum First-Story Drift Ratio and
Normalized Base Shear Force in the Y Loading Direction for a 60% JR
diagram of a composite steel beam, which was nominally the same
Takatori Record Based on 2D and 3D Analysis Group Categories
with the one of the second floor beams of the test specimen, is shown
in Fig. 10(b). In both Figs. 10(a and b), the calibrated hysteretic 2D analysis 3D analysis
responses of the modified Ibarra-Medina-Krawinkler deterioration Normalized Normalized
model are superimposed, indicating a good match between the SDR1 base shear SDR1 base shear
simulated (noted as Simul. Data) and experimental results (noted as Statistics (rad) (V/W) (rad) (V/W)
Exper. Data).
Panel zones were explicitly modeled using a parallelogram Mean m 0.0104 0.62 0.0154 0.51
model that can deform in shear. The shear force-shear distortion (0.0191)a (0.54)a (0.0191)a (0.54)a
relationship was modeled with the Krawinkler model (Gupta and s 0.0067 0.39 0.0125 0.34
Krawinkler 1999). This model was calibrated from the deduced COV 0.6413 0.62 0.8136 0.68
shear force-shear distortion of the composite beam that was tested Maximum 0.0287 1.40 0.0439 1.86
prior to shake table tests. P-D effects were considered in the analysis Minimum 0.0019 0.18 0.0031 0.05
a
using the corotational transformation. The effect of slabs on the Experimental results for SDR1 and V/W are shown in parentheses.
flexural stiffness of the beams was considered by assuming an
equivalent section with a moment of inertia 1.8 times larger than the
bare steel section. This value was determined by using the method of the JR Takatori record were applied sequentially to the developed
effective equivalent steel areas discussed in Eurocodes 4 and 8 (Parts numerical model. Hence, the cumulative damage effect from phase
1e1 and 1, respectively) [European Committee for Standardization to phase was allowed for. This was deemed critical to accurately
(CEN) 2004a, b]. A cracked section was assumed for the concrete. capture the residual drift ratio prior to the 100% JR Takatori record.
The effective width B of the exterior beam slab was estimated as If the 100% JR Takatori motion was directly applied to the numerical
B 5 bf 1 0:1L, where bf is the flange width of the steel girder and L is model, dynamic collapse would still occur (Nam and Kasai 2011).
the span of the girder (Gupta and Krawinkler 1999). Ignoring the Fig. 8 illustrates the simulated response of the first-story drift ratio
composite effect on the flexural stiffness of the beams in general does history for the 100% JR Takatori record, together with the exper-
not lead to accurate simulations of the dynamic response of the test imental data (noted as Simul.Data, 2-D), indicating an excellent
specimen through collapse. match between the numerical simulation and experimental data of
To assess the seismic responses of the moment frame in the the first-story drift ratio history. Collapse is predicted slightly later
Y loading direction, the scaled intensities (20, 40, 60, and 100%) of (t 5 6:7 s) compared with the experimental data. This is likely to be

126 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Key analysis parameters adopted by different groups for 2D and 3D analysis categories

attributed to the fact that biaxial effects were not considered in the 2D structure. The base shear was normalized with respect to the total
analysis. For the given ground motion sequence (20, 40, and 60%), weight W of the structure. In the same figure, the numerical sim-
dynamic collapse of the test frame would occur at about 90% of the ulations from two different cases are superimposed. In the first case,
JR Takatori ground motion (Nam and Kasai 2011). Note that the both P-D effects and component deterioration were considered using
CPU time to run the numerical analysis up to collapse took about the modified Ibarra-Medina-Krawinkler model. The prediction
180 s. In Fig. 8, the same prediction by the first award winner matches the experimental data relatively well. The differences in the
group of the blind analysis competition is superimposed (noted as descending part of the response are in part attributed to the high
Simul.Data, 3-D). The CPU time to conduct the analysis was about frequencies affecting the acceleration measurements. In the second
3,700 s. case, only P-D effects are simulated (i.e., no component deteri-
Fig. 11 shows the base shear of the test structure versus the oration). In this simulation, the analytical model did not collapse
first-story drift ratio SDR1 as recorded during the test up to the instant during the 100% JR Takatori record and experienced only about 7%
that dynamic instability occurred. The total base shear V of the rad SDR. This simple comparison illustrates the importance of
building was deduced from the summation of the floor acceleration considering component deterioration for reliable collapse simulations,
recordings multiplied by the individual floor masses of the test as has also been confirmed by earlier studies (Lignos et al. 2011).

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 127

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Calibrated moment rotation diagrams from cyclic loading tests of steel components released prior to the E-Defense blind analysis competition
to the competition participants

Strategies to Increase Collapse Capacity

The validated numerical model discussed in the previous section


offers an opportunity to investigate numerically, efficient alter-
natives to reduce the collapse potential of steel moment resisting
frames. Two main strategies are investigated that aim at column
design: (1) enhancement of plastic deformation and (2) enhance-
ment of strength. For all the mitigation strategies presented in this
section, the same loading sequence adopted in the actual shaking
table test was repeated. In case collapse did not occur during the
simulation, the JR Takatori was successively scaled beyond 100%
until dynamic collapse occurred. Note that this discussion is pri-
marily concerned with the prediction of a specific collapse mode,
i.e., the sidesway instability.

Capacity Enhancement of Steel Columns Fig. 11. Base shear versus first-story drift ratio history during 100% JR
Because strength deterioration in the first-story columns of the test Takatori; comparison with deteriorating and nondeteriorating analytical
structure was the primary reason that the structure collapsed, the model
effectiveness of plastic deformation and strength enhancement of
these columns is investigated by using three different HSS column
sections. An HSS 300 3 12 section (noted as Case 1), an HSS Table 4. Comparison of Basic Parameters for Cases 1e3 Compared with
350 3 9 section (noted as Case 2), and an HSS 350 3 12 section Original Parameters
(noted as Case 3) were selected for the columns instead of an HSS HSS 300 3 9 HSS HSS HSS
300 3 9. These column sections result in a column weight increase (original 300 3 12 350 3 9 350 3 12
of about 1.33, 1.16, and 1.54 times, respectively, compared with the Parameter column) (Case 1) (Case 2) (Case 3)
original HSS 300 3 9 column. The corresponding depth D to
D/t 33.3 25.0 38.9 29.2
thickness t ratios for these columns is summarized in Table 4.
Mp /Mp, HSS 300 3 9 1.0 1.3 1.4 1.8
According to the compactness limits from the Japanese seismic
up /up, HSS 300 3 9 1.0 1.3 0.85 1.01
provisions (AIJ 2006), Cases 1 and 3 are ranked as high ductility
pffiffiffiffiffiffiffiffiffiffiffiffi Weight increase 1.00 1.33 1.17 1.54
members (D/t , 33 pffiffiffiffiffiffiffiffiffiffiffiffi y ). Case 2 is ranked as a low ductility
235/f
ratio
member (D/t . 33 235/fy ). The original column (HSS 300 3 9) of SCWB 1.51 1.62 1.73 2.25
the test structure is ranked as a moderate ductility member. In the
same table, the plastic bending strengths (noted as Mp ) and plastic
deformation capacities (noted as up ) for all cases are also summa- column section of the test structure, are ranked as low ductility
rized. These parameters have been normalized with respect to the members. It is also important to note that all the HSS columns that are
bending strength and plastic deformation capacity of the original summarized in Table 4 respect the slenderness ratio requirements for
section (HSS 300 3 9). The information on the plastic deformation both the AISC (2010) and AIJ (2006).
capacity is extracted from a recently developed database for de-
terioration modeling of tubular HSS columns (Lignos and Krawinkler
Exposed Column Bases with Controlled Anchor Bolt
2009, 2010). Case 1 reflects a plastic deformation enhancement (from
Failure
moderate to high ductility members) of the columns with a small
strength increase, whereas Cases 2 and 3 mostly reflect a column The original rigid plate column bases of the test specimen increased
strength enhancement with practically no increase of plastic de- the likelihood for local buckling and subsequently strength de-
formation. According to the AISC 341-10 (AISC 2010) seismic terioration to occur at the base locations. It has been shown exper-
provisions, all three cases discussed previously, including the original imentally that controlled yielding of anchor bolts in exposed column

128 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


bases substantially increases the plastic deformation capacity of the measured experimental data from the actual collapse test are
these columns (Cui et al. 2009; Cui and Nakashima 2011). An superimposed. As seen in this figure, collapse is prevented in all
example of the hysteretic response of a column with an exposed base cases. For Case 1 (high ductility member), absolute maximum SDR1
plate is shown in Fig. 12 for an axial load ratio N/Ny 5 0:20. To of about 6.5% develops, indicating that, despite the larger de-
examine whether such an increase in plastic deformation capacity is formation capacity of this story and the additional 25% of column
beneficial against collapse, an additional case (Case 4) was analyzed. bending strength increase compared with the original case, the
To put this case in perspective with Cases 1e3 discussed previously, maximum absolute SDR is still concentrated in the first story. For
Case 4 can be considered as a high ductility member. A self- Case 4 (yielding of anchor bolts in the exposed column bases), which
centering material (Christopoulos et al. 2008) available in Open- is also ranked as a high ductility member (see Fig. 12), the distri-
Sees (McKenna 1997) is used in the numerical model to simulate the bution of SDRs along the height of the steel moment frame is very
hysteretic response of the exposed column bases. This model was similar to Case 1. The reason is that these cases mostly enhance
calibrated with respect to the moment rotation diagrams for different plastic deformations of the first-story columns. However, in Case 1,
axial load ratios. An example of this calibration is shown Fig. 12. the column bending strength is much larger compared with the case
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Based on the available experimental data on exposed column bases with exposed column bases (0:60Mp in bending strength); thus,
(Cui et al. 2009; Cui and Nakashima 2011), 60% of the bending SDRs are slightly smaller than Case 4. Looking into Cases 2 and 3
strength Mp of the original column section HSS 300 3 9 was (column strength enhancement) it is notable that the larger the
assigned for the base plate yielding. bending strength of the column, the more uniform the distribution of
maximum SDRs along the height (see Fig. 13). This indicates that
plastification is uniformly distributed along the height of the steel
Collapse Potential of the 4-Story Steel Frame moment frame, and its yield mechanism shifts to a complete 4-story
mechanism.
Fig. 13 summarizes the peak SDRs along the height of the four-story To assess how beneficial each alternative would be in terms of the
steel moment frame for the 100% JR Takatori record. For reference, enhancement of capacity against collapse, incremental dynamic
analysis was carried out with JR Takatori. In Case 1 (use of HSS
300 3 12 columns), the frame collapsed at 140% JR Takatori with
a first-story collapse mechanism. Fig. 14 shows the SDR histories for
all the stories in Case 1 (high ductility member). Despite that local
buckling of columns is delayed in this case compared with the
original steel moment frame (moderate ductility member), its col-
lapse mechanism is still a local first story. The same observation
applies when the exposed column bases with controlled yielding of
anchor bolts are used. In this case, the frame collapses at 135% of
the JR Takatori record with a first-story collapse mechanism.
Figs. 15(a and b) show the base shear versus first-story drift ratios
for the two cases. The base shear in these figures is normalized with
respect to the seismic weight W of the test structure. It is observed

Fig. 12. Calibrated moment rotation diagram of cantilever steel column


with exposed base plates with anchor bolt failure for axial load ratio
N/Ny 5 0:20 (data from Cui et al. 2009; Cui and Nakashima 2011)

Fig. 13. Seismic response of the 4-story steel moment frame in the Y
direction at 100% JR Takatori based on the alternative collapse miti- Fig. 14. Seismic response of the 4-story structure with HSS 300 3 12
gation strategies (Case 1) at 140% JR Takatori

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 129

J. Struct. Eng. 2013.139:120-132.


Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Normalized base shear versus first-story drift histories: (a) Case 1 (HSS 300 3 12): 140% JR Takatori; (b) Case 4 (anchor bolt failure): 135%
JR Takatori

that by the time strength deterioration occurs in the first-story

SDR4 [rad]
0.15
Case 3
0.1
column bases, the frame collapses with the first-story mechanism. Story 1, Enh.
0.05
However, both strategies are effective in terms of increasing the 0
collapse capacity of the test structure. For Case 2 (use of HSS 0.05
300 3 9 columns, i.e., low ductility members), the frame collapses at 0 5 10 15
Time [sec]
160% of the JR Takatori motion. Note that this case clearly shows

SDR3 [rad]
0.15
a column strength enhancement compared with the previous two 0.1 Case 3
cases (Table 4). This strength enhancement is also more economic Story 1, Enh.
0.05
compared with Case 1 (high ductility member), because the increase 0
in column weight is negligible (Table 4). 0.05
0 5 10 15
Fig. 16 shows the SDR histories along the height for Case 3 (use Time [sec]
of HSS 350 3 12 columns, mostly strength enhancement). This 0.15
SDR [rad]

frame has been subjected to a 200% JR Takatori record. The Case 3


0.1
Story 1, Enh.
0.05
2

maximum SDRs along the height are almost uniform even at


this extreme shaking, indicating a complete story collapse mecha- 0
0.05
nism that does not involve plastic hinging in the columns except for 0 5 10 15
the column bases. Note that the SCWB ratio in this case is 2.1 for the Time [sec]
interior second floor joint. This indicates that many present seismic
SDR1 [rad]

0.15
Case 3
provisions around the world [AISC 2010; International Code 0.1
Story 1, Enh.
0.05
Council (ICC) 2009; AIJ 2006; CEN 2004a] underestimate the effect
0
of dynamic loading on moment redistribution in columns, partic- 0.05
ularly for the first story of steel moment frames. To avoid plastic 0 5 10 15
Time [sec]
hinges in columns during severe ground motions, a SCWB ratio of at
least 2.0 appears to be more effective. This confirms earlier ana-
Fig. 16. Comparison of seismic responses of 4-story structures with
lytical studies on moment-resisting frames with different heights
HSS 350 3 12 (noted as Case 3) and enhancement only in the first story
[Nakashima and Sawaizumi 2000; Ibarra and Krawinkler 2005;
at 200% JR Takatori
National Institute of Standards and Technology (NIST) 2010;
Lignos et al. 2011]. An additional analysis was conducted in which
only the first-story columns of the steel frame were assumed to be
HSS 350 3 12, and the columns in the rest of the stories were kept as numerical simulation of the seismic response of the 4-story structure
the original HSS 300 3 9. The story drift histories of this moment are identified through (1) a blind analysis completion that was
frame are superimposed with a dashed line in Fig. 16 for 200% JR conducted in parallel with the test series and (2) a state-of-the-art
Takatori. These histories indicate that the associated collapse numerical model that is able to simulate strength and stiffness de-
mechanism in this case is also a complete 4-story collapse mech- terioration of steel components under cyclic loading. A few collapse
anism, suggesting that with just the strength enhancement of the mitigation alternatives are investigated for plastic deformation and
first-story columns, the collapse capacity of the steel frame sub- strength enhancement of the frame so that its first-story collapse
stantially increases. mechanism would be delayed or shifted to a complete-story mech-
anism. These strategies involve the use of stronger steel columns and
exposed column bases with controlled anchor bolt failures, which
Summary and Conclusions takes advantage of the large deformation capacity that these column
bases can undergo under cyclic loading. The main findings of this
This paper summarizes the collapse assessment of a full-scale paper are summarized as follows:
4-story structure that was recently tested to collapse at the E-Defense • There is no clear advantage between 3D and 2D analyses and vice
shaking table facility. The key parameters that would affect the versa for reliable seismic evaluation of steel moment frames that

130 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

J. Struct. Eng. 2013.139:120-132.


develop sideway-yielding mechanisms, provided that these Architecture Institute of Japan (AIJ). (1996). Japanese architectural stan-
frames are part of steel buildings with a regular plan view as in dard specification JASS 6 steel work, 7th Ed., AIJ, Tokyo. (in Japanese)
this case. This conclusion also holds true for cases that the effect of Architectural Institute of Japan (AIJ). (2006). “Report of seismic perfor-
the vertical component of the ground motion is minimal on the mance improvement of civil, architectural structures subjected to long-
period ground motions generated by subduction zone.” Rep., Japan
seismic performance of steel moment frames through collapse.
Society of Civil Engineering, Tokyo (in Japanese).
• The assumption of Rayleigh damping typically leads to better ASCE. (2010). “Minimum design loads of buildings and other structures.”
predictions of seismic response compared with other methods. ASCE/SEI 7-10, Reston, VA.
To simply consider the effect of nonstructural components on the Borzognia, Y., and Bertero, V. (2006). Earthquake engineering: From
seismic response of steel moment frame structure, a damping engineering seismology to performance-based engineering, 1st Ed., Y.
ratio z that is larger than 2% is likely to be more adequate. Borzognia and V. Bertero, eds., CRC Press, Boca Raton, FL, 9-1e9-59.
• As expected, P-D effects should be explicitly considered in the Building Center of Japan (BCJ). (2008). Building standard law of Japan,
nonlinear response history analysis of steel frame structures, BCJ, Tokyo.
particularly when these structures undergo large deformations Chopra, A. (2007). Dynamics of structures. Theory and applications to
earthquake engineering, 3rd Ed., Prentice Hall, Upper Saddle River, NJ.
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

and eventually collapse.


• Strength and stiffness deterioration of steel components should Christopoulos, C., Tremblay, R., Kim, H.-J., and Lacerte, M. (2008). “Self-
centering energy dissipative bracing system for the seismic resistance of
always be considered in the analytical model to accurately simulate
structures: Development and validation.” J. Struct. Eng., 134(1), 96e107.
collapse. Relatively simple phenomenological models are able to Chung, Y. L., Nagae, T., Hitaka, T., and Nakashima, M. (2010). “Seismic
simulate sideway collapse of steel moment frames, provided that resistance capacity of high-rise buildings subjected to long-period
there is sufficient information related to deterioration character- ground motions: E-Defense shaking table test.” J. Struct. Eng., 136(6),
istics of steel components as part of the steel structure. 637e644.
• Increasing the plastic deformation capacity of columns (moder- Cui, Y., Nagae, T., and Nakashima, M. (2009). “Hysteretic behavior and
ate or low versus high ductility members) that are part of strength capacity of shallowly embedded steel column bases.” J. Struct.
a yielding mechanism of a steel frame structure typically results Eng., 135(10), 1231e1238.
in an increased collapse capacity of this structure. However, this Cui, Y., and Nakashima, M. (2011). “Hysteretic behavior and strength
rarely can change the collapse mechanism from a single-story to capacity of shallowly embedded steel column bases with SFRCC Slab.”
J. Earthq. Eng. Struct. Dyn., 40(13), 1495e1513.
a complete-collapse mechanism (involving all the floor beams
European Committee for Standardization (CEN). (2004a). “Design of
and columns at the base location) when column plastification is
composite structures—Part 1.1: General rules and rules for buildings.”
concentrated in this story. Eurocode 4, Brussels, Belgium.
• The effect of exposed column bases that allow for elongation of European Committee for Standardization (CEN). (2004b). “Design of
the anchor bolts on the collapse capacity of a steel moment frame structures for earthquake resistance—Part 1: General rules, seismic
is clearly beneficial, primarily because of the large deformation actions and rules for buildings.” Eurocode 8, Brussels, Belgium.
capacity that these column bases can undergo prior to anchor bolt Gupta, A., and Krawinkler, H. (1999). “Prediction of seismic demands for
failures. However, once the base plates deteriorate in strength, the SMRFs with ductile connections and elements.” Rep. No. SAC/BD-99/
collapse mechanism cannot change to a complete mechanism. 06, SAC Joint Venture, Sacramento, CA.
This is worth investigating further. Hartigan, J. A., and Wong, M. A. (1979). “A K-Means clustering algorithm.”
• A SCWB ratio of ∼2.0 seems to be adequate to avoid column J. R. Stat. Soc. Ser. C Appl. Stat., 28(1), 100e108.
Hikino, T., Ohsaki, M., Kasai, K., and Nakashima, M. (2009). “Simulation
plastic hinges and consequently the development of individual
of E-Defense full-scale shake-table test results of moment-resisting steel
story mechanisms in a steel structure. In that respect, presently frame.” Proc., 6th Int. Conf. on Behavior of Steel Structures in Seismic
available seismic provisions around the world are not necessarily Areas, Taylor & Francis Group, London.
adequate to prevent plastification of columns caused by dynamic Ibarra, L. F., and Krawinkler, H. (2005). “Global collapse of frame structures
redistribution of moments that typically occur under severe under seismic excitations." Rep. No. TB 152, The John A. Blume
ground shaking. Earthquake Engineering Center, Stanford Univ., Stanford, CA.
Ibarra, L. F., Medina, R. A., and Krawinkler, H. (2005). “Hysteretic models
that incorporate strength and stiffness deterioration.” J. Earthq. Eng.
Acknowledgments Struct. Dyn., 34(12), 1489e1511.
International Code Council (ICC). (2009). International building code
D. G. Lignos was supported by an overseas fellowship from the Ja- IBC 2009,Birmingham, AL.
pan Society for the Promotion of Science (JSPS, Award Number Ji, X., Kajiwara, K., Nagae, T., Enokida, R., and Nakashima, M. (2009).
P09291). This support is acknowledged and is greatly appreciated. “A substructure shaking table test for reproduction of earthquake
The test presented in this study was conducted as part of a compre- responses of high-rise buildings.” J. Earthq. Eng. Struct. Dyn., 38(12),
hensive research project on quantification of the safety margin of 1381e1399.
structures under earthquake load administered by the National Liang, Z., and Lee, G. C. (1991). “Damping of structures. Part 1—Theory of
complex damping.” NCEER Rep. No. 91-0004, State Univ. of New York
Research Institute for Earth Science and Earthquake Mitigation
at Buffalo, Buffalo, NY.
(NIED). The writers are grateful to Drs. Y. Okada and K. Abe of Lignos, D. G., and Krawinkler, H. (2009). “Sidesway collapse of de-
NIED, Prof. M. Ohsaki of Hiroshima University, and Prof. T. Hitaka teriorating structural systems under seismic excitations.” Rep. No. TR
of Kyoto University for their continuous support and valuable com- 172, John A. Blume Earthquake Engineering Center, Dept. of Civil
ments. The writers also thank Prof. Y. S. Yang of National Taipei Engineering, Stanford Univ., Stanford, CA.
University of Technology for his willingness to share the simulated Lignos, D. G., and Krawinkler, H. (2010). “A steel database for component
response data of his 3D numerical model. deterioration of tubular hollow square steel columns under varying axial
load for collapse assessment of steel structures under earthquakes.”
Proc., 7th International Conf. on Urban Earthquake Engineering
References (7CUEE), CUEE, Tokyo, Japan.
Lignos, D. G., and Krawinkler, H. (2011). “Deterioration modeling of steel
AISC. (2010). Seismic provisions for structural steel buildings, including components in support of collapse prediction of steel moment frames
supplement No. 1, AISC, Chicago, IL. under earthquake loading.” J. Struct. Eng., 137(11), 1291e1302.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013 / 131

J. Struct. Eng. 2013.139:120-132.


Lignos, D. G., Krawinkler, H., and Whittaker, A. (2011). “Prediction and Reduction Program (NEHRP) Consultants Joint Venture for the National
validation of sidesway collapse of two scale models of a 4-story steel Institute of Standards and Technology, Gaithersburg, MD.
moment frame.” J. Earthq. Eng. Struct. Dyn., 40(7), 807e825. Ohsaki, M., Kasai, K., Hikino, T., and Matsuoka, Y. (2008a). “Overview of
MacRae, G. A. (1999). “Parametric study on the effect of ground motion 2007 E-Defense blind analysis contest results.” Proc., 14th World Conf.
intensity and dynamic characteristics on seismic demands of steel mo- on Earthquake Engineering, China Earthquake Administration, Ministry
ment resisting frames.” Rep. 99/01A, SAC Steel Project, Richmond, CA. of Housing and Urban-Rural Development, Beijing, China.
MacRae, G. A., and Mattheis, J. (2000). “Three-dimensional steel building Ohsaki, M., Kasai, K., Thiagarahan, G., Yang, Y. S., and Komiya, Y.
response to near-fault motions.” J. Struct. Eng., 126(1), 117e126. (2008b). “3-D analysis methods for 2007 blind analysis contest.” Proc.,
Matsuoka, Y., Suita, K., Yamada, S., Shimada, Y., and Akazawa, M. (2008). 14th World Conf. on Earthquake Engineering, China Earthquake
“Non-structural component performance in 4-story frame tested to Administration, Ministry of Housing and Urban-Rural Development,
collapse.” Proc., 14th World Conf. on Earthquake Engineering, China Beijing, China.
Earthquake Administration, Ministry of Housing and Urban-Rural Rodgers, J., and Mahin, S. (2006). “Effects of connection fractures on global
Development, Beijing, China. behavior of steel moment frames subjected to earthquakes.” J. Struct.
McKenna, F. (1997). “Object oriented finite element programming Eng., 132(1), 78e88.
frameworks for analysis, algorithms and parallel computing,” Ph.D. Sato, E., Furukawa, S., Kakehi, A., and Nakashima, M. (2011). “Full-scale
Downloaded from ascelibrary.org by MCGILL UNIVERSITY on 01/23/13. Copyright ASCE. For personal use only; all rights reserved.

Dissertation, Univ. of CaliforniaeBerkeley, Berkeley, CA. shaking table test for examination of safety and functionality of base-isolated
Nakashima, M. (2008). “Roles of large structural testing for the advance of medical facilities.” J. Earthq. Eng. Struct. Dyn., 40(13), 1435e1453.
earthquake engineering.” Proc., 14th World Conf. Earthq. Eng., Beijing, Suita, K., Yamada, S., Tada, M., Kasai, K., Matsuoka, Y., and Shimada, Y.
China, WCEE. (2008). “Collapse experiment on 4-story steel moment frame: Part 2
Nakashima, M., Matsumiya, T., Suita, K., and Liu, D. (2006). “Test on full- detail of collapse behavior.” Proc., 14th World Conf. on Earthquake
scale three-storey steel moment frame and assessment of ability of Engineering, China Earthquake Administration, Ministry of Housing
numerical simulation to trace cyclic inelastic behavior.” J. Earthq. Eng. and Urban-Rural Development, Beijing, China.
Struct. Dyn., 35(3), 3e19. Tai, T., Hitaka, T., Cui, Y., Sou, S., and Nakashima, M. (2010). “De-
Nakashima, M., and Sawaizumi, S. (2000). “Column-to-beam strength ratio velopment of steel beam-to-column connections using SFRCC floor
required for ensuring beam-collapse mechanisms in earthquake responses of slabs as exterior diaphragms.” J. Struct. Constr. Eng., 653, 1369e1376
steel moment frames.” Proc., 12th World Conf. on Earthquake Engineering, (in Japanese).
New Zealand Society for Earthquake Engineering, Silverstream, New Vian, D., and Bruneau, M. (2001). “Experimental investigation of P-Delta
Zealand. effects to collapse during earthquakes.” Rep. MCEER 01-0001, Multi-
Nam, T. T., and Kasai, K. (2011). “Dynamic analysis of a full-scale four- disciplinary Center For Earthquake Engineering Research, Univ. at
story steel building experimented to collapse by strong ground motions.” Buffalo, Buffalo, NY.
Proc., 8th Int. Conf. on Urban Earthquake Engineering (8CUEE), Yamada, S., Kasai, K., Shimada, Y., Suita, K., Tada, M., and Matsuoka, Y.
CUEE, Tokyo, Japan. (2009). “Full scale shaking table collapse experiment on 4-story steel
National Institute of Standards and Technology (NIST). (2010). “Evaluation moment frame: Part 1 outline of the experiment.” Proc., 6th Int. Conf. on
of the FEMA P-695 methodology for quantification of building seismic Behavior of Steel Structures in Seismic Areas, Taylor & Francis Group,
performance factors.” GCR 10-917-8, National Earthquake Hazards London.

132 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JANUARY 2013

View publication stats J. Struct. Eng. 2013.139:120-132.

You might also like