ASCE-41 and FEMA-351 Evaluation: of E-Defense Collapse Test

EARTHQUAKE ENGINEERING PRACTICE
ASCE-41 and FEMA-351 Evaluation

of E-Defense Collapse Test
Bruce F. Maison,a) M.EERI, Kazuhiko Kasai,b) M.EERI,
and Gregory Deierlein,c) M.EERI
A welded steel moment-frame building is used to assess performance-

based engineering guidelines. The full-scale four-story building was shaken to
collapse on the E-Defense shake table in Japan. The collapse mode was a
side-sway mechanism in the first story, which occurred in spite of a strong-
column and weak-beam design. Computer analyses were conducted to
simulate the building response during the experiment. The building was then
evaluated using the Seismic Rehabilitation of Existing Buildings (ASCE-41)
and Seismic Evaluation and Upgrade Criteria for Existing Welded Steel
Moment-Frame Buildings (FEMA-351) for the collapse prevention
performance level via linear and nonlinear procedures. The guidelines
had mixed results regarding the characterization of collapse, and no
single approach was superior. They mostly erred on the safe side by
predicting collapse at shaking intensities less than that in the experi-
ment. Recommendations are made for guideline improvements.
关DOI: 10.1193/1.3224159兴
INTRODUCTION
Performance-based engineering (PBE) is evolving as the preferred approach to
earthquake design, especially for building rehabilitation. By assessing specific perfor-
mance objectives, PBE provides an important link between design, risk management,
regulation, and policy. Implicit is the notion that probable seismic response can be
predicted.
Guidelines such as the ASCE standard Seismic Rehabilitation of Existing Buildings
(ASCE-41 2006) and the FEMA report Seismic Evaluation and Upgrade Criteria for Ex-
isting Welded Steel Moment-Frame Buildings (FEMA-351 2000) are now available and
accepted by some jurisdictions. These represent the state-of-the-art in PBE, but incor-
porate expert opinion to bridge gaps where definitive research is lacking. Benchmarking
PBE against actual building response is essential to assess its efficacy and to help guide
its future development.
a)
Structural Engineer, 7309 Lynn Avenue, El Cerrito, CA 94530
b)
Professor, Tokyo Institute of Technology, Tokyo, Japan
c)
Professor, Stanford University, Stanford, California
927
Earthquake Spectra, Volume 25, No. 4, pages 927–953, November 2009; © 2009, Earthquake Engineering Research Institute
928 MAISON, KASAI, AND DEIERLEIN
Figure 1. Elements of case study building: Steel moment-frames on each column row and au-
toclaved lightweight concrete panels for nonstructural exterior wall cladding system.
The welded steel moment-frame (WSMF) is a relatively young structural system,

and much is yet to be revealed about its behavior during actual earthquakes—recall the
surprise connection fractures caused by the 1994 Northridge earthquake. PBE guidelines
cover WSMFs, but their effectiveness for collapse prevention evaluation cannot be
judged via post-quake damage surveys because collapse of modern steel buildings has
yet to be documented. That leaves comparison to shake table tests as one of the best
ways currently available to assess PBE for this class of buildings.
This paper presents a case study of a full-scale, four-story WSMF office building
shaken to collapse as part of an E-Defense steel building research project (Kasai et al.
2007, Yamada et al. 2008, Suita et al. 2008, Matsuoka et al. 2008). An objective is to
assess how well current United States PBE guidelines characterize an actual collapse.
CASE STUDY BUILDING

The building (Figure 1) was tested in September 2007 on the E-Defense shake table
located in Miki, Japan. Story heights and bay widths were typical full-scale dimensions,
but the numbers of bays were fewer than usual in order to optimize the specimen in re-
lation to the shake table capabilities. Typical Japanese design and construction practice
was followed, namely, the use of tubular columns and welded steel moment-frames on
each column row (Figure 2). Enhanced welded connection details were used to reduce
the possibility of brittle fracture (Figure 3). The exterior-wall cladding consisted of au-
toclaved lightweight concrete (ALC) panels designed to be nonstructural relative to the
moment-frames (Figure 4). Table 1 contains pertinent information, and additional de-
scriptions can be found elsewhere (Yamada et al. 2008).
E-DEFENSE EXPERIMENT
SHAKE TABLE EXCITATIONS

The experiment had repeated applications (tests) of the same wave forms with in-
creasing amplitude scale factors: SF= 0.05, 0.2, 0.4, 0.6, and 1.0. The motions were those
recorded at the JR Takatori train station in Kobe, Japan during the 1995 Hyogo-Ken Nambu
earthquake (Mw 6.9). This is a severe excitation recorded in close proximity to the faulting.
ASCE-41 AND FEMA-351 EVALUATION OF E-DEFENSE COLLAPSE TEST 929
Figure 2. Exploded view of typical girder-to-column framing with shop-welded components

that were erected via site-bolting (Table 1 contains additional member descriptions).
Figure 3. Typical joint detail at bottom flange with enhanced details reflecting post-Kobe earth-
quake experience, notably having no through-web access hole.
Figure 4. Nonstructural exterior wall autoclaved lightweight concrete (ALC) panel attachment
to moment-frames (for angle L-65⫻ 65⫻ 6: leg length= 2.6 in and thickness= 0.24 in).
Table 1. Building properties
Attribute Comment
Design Basis Japan Standard Building Law. Designed for two earthquake levels. Level 1 had
a base shear coefficient= 0.2 and required a peak story drift ratio less than 0.005.
Level 2 required demonstration of lateral stability with a base shear
coefficient= 0.3.
SCWB Strong-column and weak-beam (SCWB) design at floors 2 through 4. The SWCB
ratio was ⬎1 for each floor and in each building principal direction according to:
兺Mpc
, where Mpc, Mpb and Mpp are the plastic moment
兺 min共1.5Mpb , 1.3Mpp兲
strengths of the column, beam and panel zone, respectively. Panel zones governed
over beams in the calculation (weaker versus beams).
Girders Hot-rolled wide-flange girders varying from H-346⫻ 174⫻ 6 ⫻ 9 (depth d = 13.6 in,
width bf = 6.9 in, web tw = 0.24 in, flange tf = 0.35 in) at roof level to H-400⫻ 200
⫻ 8 ⫻ 13 (d = 15.7 in, bf = 7.9 in, tw = 0.32 in, tf = 0.51 in) at second floor. SN400B
steel. Nominal and expected (as tested) Fy = 34 ksi and Fye = 45 ksi (235 MPa and
313 MPa), respectively.
Girder ratios bf
2tf
⬇ 6 → 10, compact per Japan code and AISC (2005): 0.38 冑 E
Fy
= 11
Columns Cold-formed steel hollow structural section (HSS), SHS-300⫻ 300⫻ 9 (h = 11.8 in,
t = 0.35 in). BCR295 steel. Nominal and expected (as-tested 0.2% offset rule) Fy
= 43 ksi and Fye = 48 ksi (295 MPa and 331 MPa), respectively.
Column ratios b h − 3t 300− 27
= = = 30.3⬇ 30, borderline compact by Japan code and AISC:
冑
t t 9
E 190
1.12 = 29. Noncompact by ASCE-41 and AISC Seismic (2005):
Fy 冑Fye = 27 and
0.64冑 E
Fy
= 17, respectively
Column bases Columns welded to baseplates bolted to fixtures attached to shake table. Connection
strength was greater than plastic moment of column. Baseplate provided moderate
base fixity having rotational stiffness ⬃6EI / L of first-story column.
Roof and floors Composite steel deck with normal weight concrete. Thickness= 6.9 inch 共175 mm兲.
Exterior walls Autoclaved lightweight concrete (ALC) panels. Typical panel dimensions: 11.5 feet
共3,500 mm兲 tall by 2 feet 共600 mm兲 wide by 5 inches 共125 mm兲 thick.
Interior walls Partitions located on the second to fourth floors consisted of drywall gypsum boards
attached to metal-stud framing. Ceiling consisted of gypsum board attached to metal
framing. Both seismic and non-seismic connection details were used to compare
performance. Seismic details had clearances to mitigate the effect of structure drifts on
partitions.
Weights Actual weights having equivalent lumped floor weights of 186 psf 共8.92 kN/ m2兲 at
roof and 140 psf 共6.7 kN/ m2兲 at floors. Included were collapse safety-catch steel
frame systems located on each floor.
As shown in Figure 5, the spectra mostly envelopes a Maximum Considered Earthquake

(MCE) that is the usual basis of collapse safety checks in the western United States.
Figure 5. Takatori earthquake spectra at 5% damping.
The building period of ⬃0.9 sec indicated in Figure 5 is the apparent fundamental
value in both principal directions during the SF= 0.6 test. Ambient vibration measurements
of the pre-experiment pristine building indicate periods of ⬃0.7 sec. The building may be
especially vulnerable to the Takatori motions because the period lies in the
spectral “valley”—the shaking intensity could increase as the period lengthens from
yielding.
EXPERIMENT SUMMARY
The building survived shaking having SF艋 0.6 and collapsed at SF= 1.0 via a mecha-
nism in the first story (Table 2). The behavior may be explained by the following sequence of
events:
1. Yielding initially occurred in panel zones and columns at the baseplates.
2. As the strength of the panel zones increased via strain hardening, the strength of
the columns effectively decreased due to biaxial bending and local buckling.
The girders remained elastic due to their overstrength caused by actual yield
stress being much greater than the nominal design value and slab-girder com-
posite action.
3. Deformations then coalesced in the columns in response to increasing story
drifts, and yielding and local buckling occurred at both the top and bottom of
the columns in the first story. A story mechanism was created, and the building
collapsed.
The building may be considered to have met its performance goal because it with-
stood shaking much greater than the Level 2 design basis. See Suita et al. (2008) for
additional description of the experiment.
DISPLACEMENT RESULTS
The side-sway mechanism had the primary direction of collapse in the Y-direction
(Figures 6 and 7), and severe local buckling occurred at top and bottom of the columns
in the first story (Figure 8). The exterior wall cladding was omitted on one side (Figure
Table 2. Experiment summary
Scale Factor
(SF) Building Response
0.05 No yielding. Small amplitude linear-elastic behavior.

0.2 No yielding. Peak drift ratio of ⬃0.005. Equivalent to a Level 1 loading condition
customarily taken as motions with a peak ground velocity of 10 in/ sec 共25 cm/ sec兲.
0.4 Slight yielding. Peak drift ratio of ⬃0.01 and peak plastic rotations of ⬃0.005 rad.
Equivalent to a Level 2 loading condition (20 in/ sec 共50 cm/ sec兲).
• Yielding in columns at all baseplates
• Yielding in panel zones at second floor
0.6 Yielding. Peak drift ratio of ⬃0.02 and peak plastic rotations of ⬃0.01 rad.
• Yielding in columns at all baseplates
• Slight local buckling of mid-side columns at baseplates
• Yielding in panel zones at second floor and in middle columns at third floor
• Peak base shear attained ⬃0.7 times building weight
• Some damage to ALC panels (corner cracking) and interior partitions
1.0 Collapse. Peak drift ratio of 0.19 as building was restrained by safety-catch systems.
• Mechanism in first story with severe local buckling in columns
• No fracture of moment-frame connection welds
• Damage to ALC panels in first story, including 3 falling off and numerous
fractured connections resulting with hanging panels
7), but this did not cause a lateral twisting collapse. The floor loads were relatively heavy
so that the mass eccentricity from the cladding was minimal, being only ⬃3.5% of the
X-direction plan dimension. Good simulation of the experiment was achieved using a planar
computer model (having no twisting degree-of-freedom), thus somewhat confirming that tor-
sion did not play a major role in the collapse.
Figure 9 shows the first story displacement trajectories, and Figure 10 shows the dis-
placement time histories. Large oscillations were oriented diagonally in plan. The dis-
placement patterns for SF艋 0.6 were similar indicating a quasi-linear-elastic behavior, but
Initial Final
8 in (200 mm)
30 (770) –
Safety
Y
Catch
Mechanism B
View A
X Plan
Elevation
Figure 6. Building collapse mechanism (no scale).

Figure 7. Building post-collapse configuration (from View A depicted in Figure 6). Safety-
catch braced-frame systems were located at the corners of building. Note tilting of wall panels
in first story.
the pattern started to change late in the SF= 0.6 test. For SF= 1.0, the response was markedly
different leading up to collapse. The motions were still diagonal, but the oscillatory signature
and peak amplitude trends were very different, indicating a softening of the structure caused
by only a few large displacement excursions.
Figures 11 and 12 illustrate collapse during the SF= 1.0 test. Deformations progres-
sively concentrated in the first story as it softened, and as they increased unbounded in the
first story, they decreased in the other stories. Collapse was sudden, occurring within a four-
second duration of intense shaking. Some progressive oscillatory one-direction offsets (ratch-
eting) was evident.
EFFECT OF NONSTRUCTUAL COMPONENTS

Figure 13 shows peak story shears from the SF= 0.6 test (Suita et al. 2008). The dif-
ference between the total building and moment-frame shears may be attributed mostly to the
nonstructural elements. The figure indicates they resisted ⬃9% of the shear at the base and
⬃15% in the third story. Their role in the collapse behavior is difficult to ascertain because
of their complex interaction with the moment-frames (Figure 4) and the fact that such com-
ponents have deteriorating pinched hysteretic loop behavior and brittle failure modes when
their connections to the frame fracture. For the SF= 1.0 test, most of the ALC panels in the
first story suffered major damage, with three falling off and many hanging (Matsuoka et al.
2008). It is noteworthy that good simulation of the experiment was achieved using a com-
puter model based only on the moment-frames, suggesting that nonstructural elements had
negligible influence on the collapse.
Figure 8. Severe local buckling of HSS column at baseplate, with drift ratio of 0.19 (from View
B depicted in Figure 6).
Figure 9. Plan view of the first-story motion (SF= 0.2, 0.4, 0.6, and 1.0). Circles at 0.01 indicate
the drift customarily associated with the initiation of yielding in steel moment-frame buildings.
Figure 10. First-story drift-ratio time histories (SF= 0.2, 0.4, 0.6, and 1.0 results are superim-
posed to have the same starting time).
Figure 11. Displacement time history leading to collapse during SF= 1.0.
STRONG-COLUMN AND WEAK-BEAM DESIGN

The building was designed to have strong-column and weak-beam (SCWB) framing,
yet a story-mechanism collapse occurred. The SCWB check (Table 1) differs from
customary United States methods, but the idea of balancing member moment capacities
is similar. The check did not preclude a story-mechanism because of the following:
• Column Biaxial Bending. The dominant motions were oriented along the plan di-
agonal resulting with large concurrent bending moments in both column princi-
pal directions. This has the effect of reducing the strength of the columns, which
was not considered, since the check was applied separately in each building prin-
cipal direction.
• Column Local Buckling. The columns suffered strength deterioration from local
buckling starting in the SF= 0.6 test. The check used plastic flexural strengths.
• Girder Overstrength. The girders’ as-tested yield stress was much larger than
nominal value (by 32%), and slab-girder composite action was not included in
the check. Hence, the girders were much stronger than the column and panel
zone strengths used in the check.
Figure 12. “Snapshots” of displacement and drift patterns at times a to d, per Figure 11. Nor-
malized displacements are set with roof displacement equal to 1.0.
Figure 13. Envelopes of Y-direction peak story shears for total building and moment-frames.
SEAOC (1999) recognized deficiencies of customary SCWB checks and has sug-
gested an alternative method in their Blue Book commentary. Column moments above
the floor level under consideration are omitted from the proposed calculation.
COMPUTER SIMULATION OF EXPERIMENT

The minimum intensity causing collapse under a single application of the scaled
Takatori record was needed to benchmark the PBE evaluations. The authors formulated
a two-dimensional planar model having the best predictions in the category of 2-D
Analysis by Practicing Engineers for an E-Defense blind analysis contest (EERI News-
letter 2008). This model was used to estimate the smallest earthquake scale factor 共SF兲
causing collapse.
The model was essentially the same as the Y-direction model used in the PBE evalu-
ation below, except for the nonlinear rotational springs representing the column flexure
behaviors. They were based on component test data rather than ASCE-41 guidelines
(Figure 14). A key assumption was the reduction of column flexural strength to account
for biaxial bending due to multidirectional earthquake excitation (Figure 14c).
The blind analysis had excellent agreement with the test oscillatory signature, instant
in time of collapse, and pattern of yielding (Figures 15 and 16). However, the analysis
had the exterior girders yielding at the second floor, whereas the test had panel zone
yielding. This may be due to girder overstrength from composite action not accounted
for in the analysis.
Figure 17 shows the incremental dynamic analysis (IDA) graph. The graph closely
agrees with the test results (solid diamonds) confirming the adequacy of the computer
model. It is also close to analyses having pre-existing damage from prior runs with
SF= 0.4 and 0.6 (hollow diamonds) indicating prior yielding had negligible effect on col-
lapse response. The blind analysis model suggested that a single application of the Takatori
record scaled to ⬃0.8 is close to the lower bound intensity just sufficient enough to cause
collapse.
Figure 14. Column component flexure tests and backbone curves: (a) and (b) component test
results for cantilever HSS column specimens (length= 56.8 in 关1468 mm兴, and axial load ratio,
P / Py = 0.15), and (c) derivation of backbone model based on test data.
PBE GUIDELINES
4
The building was evaluated by ASCE-41 and FEMA-351, and they are quite differ-
ent (Table 3). ASCE-41 is a deterministic approach that computes individual member
demand-to-capacity ratios. For linear analysis, the checks use ductility m-factors in-
tended to relate member elastic forces to inelastic deformations. For nonlinear analysis,
ASCE-41 provides criteria for member and connection inelastic deformations. FEMA-
351 computes confidence estimates for particular building failure modes.
The guidelines are very detailed (ASCE-41 is 400 pages and FEMA-351 is 200
pages) with prescriptive code-type command language (e.g., shall), which some engi-
neers contend can inhibit good practice (Searer et al. 2008). This study strictly followed
the guidelines as close as possible, but some compromises had to be made in order to
apply them to the experiment. The main ones follow:
Single Set of Earthquake Records. The evaluation was based on the single set of
records used in the experiment, whereas the guidelines stipulate use of multiple records.
It would be ideal to have multiple collapse tests using the same building specimen type,
but it is clear that such an ambitious experiment is cost prohibitive and probably will
never be performed. Therefore, this exception was unavoidable, and it should be recog-
nized that if other earthquake records were used, then different outcomes could result.
Regarding FEMA-351, the evaluation was based on a 50% confidence level to rep-
resent a median value for comparison to the experiment. FEMA-351 Appendix A was
used to account for the deterministic earthquake input in the experiment. Reducing the
demand uncertainty has the effect of increasing the capacity at a given confidence level.
4
Using ASCE-41 for seismic evaluation is a Tier 3 approach per ASCE standard Seismic Evaluation of Existing
Buildings (ASCE-31 2003). However, the demand reduction factor of 0.75 was not used here because it appears
to be intended to reduce conservatism in design versus evaluation earthquake motions (“mean-plus-one” stan-
dard deviation aspects) that are not relevant to the experiment having a given deterministic input.
Figure 15. Time histories of first-story drift ratios comparing test and blind analysis.
Moment Connection Type. The connections were typical for Japan but are not com-
mon in the United States, so the guidelines did not specifically address them. They were
assumed to be equivalent to an improved WUF-welded web type, i.e., a fully welded con-
nection. This is a rugged connection having a large collapse prevention m-factor (5.3)
and drift ratio capacity 共␪U = 0.064兲. It turns out that connections did not govern in
ASCE-41 but did control in FEMA-351 (floor vertical collapse check).
COMPUTER MODELS
Practical application of PBE requires use of computer-based analysis. Both linear
and nonlinear computer models were created using guideline criteria. Modeling and
scaling of results were based on ASCE-41 (FEMA-351 refers to FEMA-273 for model-
ing, which can be considered as superseded by ASCE-41). For linear analysis, a center-
line model was used with girder properties based on the bare steel sections. For nonlin-
ear analysis, a more detailed model having panel zones and slab-girder composite action
was used. Nonstructural components were not modeled.
Separate planar models representing the moment-frames in each building principal
direction were created (Figure 18). Results from the two directions were combined via
the 30%-combination rule per ASCE-41. Member properties were well known so the
knowledge factor 共␬兲 used to scale ASCE-41 capacities was set to unity. Three different
procedures were applied: linear dynamic (response spectrum), nonlinear dynamic (time
history), and nonlinear static (push over).
Figure 16. Distribution of yielding in test and blind analysis 共SF= 0.6兲.
Figure 17. Blind analysis model incremental dynamic analysis (IDA) graph. Each symbol rep-
resents the peak-story drift ratio from shaking by the Takatori record, scaled by the particular
scale factor.
LINEAR DYNAMIC PROCEDURE (LDP) MODELS

Girders and column members were modeled with standard linear-elastic beam-
column line elements. Nodes were located at the intersection of the column and girder
centerlines (Figure 19b). The baseplates were idealized as rotational springs having stiff-
ness as estimated by Tada et al. (2007). The effects of slab-girder composite action were
ignored, unlike the nonlinear models below. The fundamental periods of 1.1 sec and
1.0 sec in the X- and Y-directions, respectively, were greater than the measured periods from
the test.
Dynamic analysis was conducted by the response spectrum method using 5% damp-
ing. ASCE-41 displacement modification factors, relating linear-elastic to inelastic re-
sponse, were equal to unity: inelastic displacement 共C1兲 and hysteretic behavior 共C2兲.
NONLINEAR DYNAMIC (NDP) AND STATIC (NSP) PROCEDURES MODELS

The models were like the linear versions, but with the inclusion of panel zones (Fig-
ure 19c). Inelastic actions were accounted for by nonlinear rotational springs connecting
the members to the panel zones and baseplates. The spring properties were based on
ASCE-41 parameters (Figure 20a). Degrading strength and stiffness hysteretic behavior
Figure 18. Computer models of case study building.

Figure 19. Computer modeling of girder-to-column framing.
(Figure 20b) was used for connections, girders, and columns, and a full-loop trilinear
hysteretic behavior (Figure 20c) was used for panel zones, since they do not deteriorate
per ASCE-41. P-delta effects were accounted for on the element level by geometric stiff-
ness based on a truss bar analogy using axial force.
Composite action was assumed to increase girder stiffness, but not strength (girder
moment of inertias were scaled by 1.5). The girders were relatively shallow versus the
slab thickness. The periods in the X- and Y-directions were 0.98 sec and 0.91 sec, re-
spectively. These are in better agreement with the test values versus those from the linear
models because of the larger stiffness resulting from member clear spans and composite ac-
tion.
NDP analysis was carried out using the time-history method. Raleigh damping of
3% was used in addition to the damping from inelastic member actions. NSP analysis
used a lateral load pattern based on the fundamental mode shape.
For column flexure, the parameters (strength, a-value, b-value, CP acceptance crite-
ria) depend on the axial force in the member. This was accounted for by first performing
an analysis using assumed axial forces and then performing another analysis using ad-
justed parameters based on the peak axial forces from the first analysis.
It was found that the columns in the first story controlled the collapse behavior, in
part because they were considered noncompact (Table 1). Per ASCE-41, the column
plastic rotation at the onset of strength deterioration was very small (a-value
⬃0.002 rad), and once yielding occurred in the first story, a collapse mechanism formed
almost immediately as the moment capacity dropped off quite rapidly (b-value ⬃0.003 rad).
This is discussed below.
COMPUTER SOFTWARE
It was intended that several analysis programs would be used to contrast the use of
different software packages: OpenSees (2007), Perform-3D (Powell 2007), and PC-
ANSR (Maison 1992). Unfortunately, project schedule and budget constraints precluded
this exercise. However, quality assurance analyses were run to test features relevant to
modeling the building. It was found that all three packages were capable of producing
similar results, and any one of them could have been employed. OpenSees was used for
Table 3. Overview of ASCE-41 and FEMA-351 procedures for steel moment-frame buildings
ASCE-41 FEMA-351
Originated from: FEMA-356 (2000) and FEMA- Originated from: SAC Joint Venture Project pur-
273 (1997). suant to 1994 Northridge earthquake.
Performance levels: Criteria for immediate occu- Performance levels: Criteria for IO and CP.
pancy (IO), life safety (LS), and collapse preven-
tion (CP).
Checks members: Connections, girders, columns, Checks failure modes: Side-sway collapse (drift
panel zones. check), floor vertical collapse (drift), column
buckling (force), column splice fracture (force).
Collapse-safe when: Member demands are less Collapse-safe when: Acceptable confidence
than capacities via member-by-member achieved for precluding the failure modes.
demand-to-capacity ratio pass-fail type checks.
Analysis procedures: Choice of linear static Analysis procedures: Choice of LSP, LDP, NSP,
(LSP), linear dynamic (LDP), nonlinear static and NDP.
(NSP), and nonlinear dynamic (NDP).
Acceptance criteria Acceptance criteria:

(for deformation-controlled actions): 1. Compute ␭ = ␥␥aD / ␾C, where D = demand
For linear analysis: D / mC ⱕ 1, where D = demand (force or drift) from linear or nonlinear analysis, C
force from analysis factored to account for inelastic = capacity per guideline rules, and ␥ , ␥a , ␾
actions, C = capacity force per guideline rules, and = parameters to account for bias and uncertainties in
m = ductility, m-factor depending on particular computation of D and C.
member properties and performance level (force is 2. Given ␭ computed above and another uncertainty
used as a proxy for deformations). For nonlinear parameter ␤UT, the confidence level is read from
analysis: D / C ⱕ 1, where D = demand deformation tables.
from analysis, and C = capacity deformation If the computed confidence levels for precluding all
depending on particular member properties and failure modes exceed recommended values, then the
performance level. probable performance is deemed acceptable.
If demand-to-capacity ratios for all members⬍ 1,
then the probable performance is deemed
acceptable.
No checking of drifts. No checking of member ductilities.
all nonlinear analyses primarily because it was used first. For linear analysis, the
SUPER-ETABS program (Maison and Neuss 1983) was used. It must be stressed that the
quality of structural analysis depends more on the skill of the engineer and less on the
particular software package.
EVALUATION BY ASCE-41 AND FEMA-351

The building was evaluated for the collapse prevention (CP) performance level. The
smallest scale factors 共SF兲 causing the building to fail the CP criteria were taken as the
Figure 20. Computer modeling of nonlinear flexure behaviors.
“collapse” intensities for comparison with the experiment. How well the guidelines estimated
the earthquake intensity causing collapse was quantified by computing a “safety” margin
共M兲 defined as
M = SFtest/SF 共1兲
where SFtest = earthquake scale factor causing collapse in the experiment, taken as 0.8, and
SF= scale factor at which the building failed the CP criteria. When margins were greater
than unity, this implied the guidelines were conservative because they predicted collapse at
earthquake intensities smaller than the actual value.
LINEAR DYNAMIC PROCEDURE (LDP)

Figure 21 shows envelopes of peak displacements and drifts at two response levels:
slight yielding 共SF= 0.4兲 and during collapse in the test 共SF= 1.0兲. Analysis overestimated
the tests because the models were more flexible than the actual building. At
SF= 0.4, analysis had peak drifts in the middle stories whereas those from the test dimin-
ished with height. At SF= 1.0, the collapse mechanism in the first story was clearly evident
Figure 21. Comparison of peak displacements and drifts from LDP analyses and tests.
Figure 22. ASCE-41 and FEMA-351 evaluation by linear dynamic procedure (LDP).
in the test, and linear analysis could not capture such a localized nonlinear displacement re-
sponse. The largest drift from analysis would incorrectly suggest the second or third story as
key whereas the actual collapse mechanism was in the first story.
Figure 22 summarizes the PBE guideline evaluations. The LDP analysis plots as a
straight line because the computer model is linear, and thus peak drifts are proportional
to scale factor 共SF兲. The peak drifts from the experiment (solid dots) and the blind analysis
model (hollow dots) are also shown.
For ASCE-41, the columns in the first story failed the CP acceptance criteria at SF
= 0.4, translating to a conservative safety margin of two (Figure 22a). The columns failed due
to excessive flexure, as in the test. Table 4 contains the governing CP criteria. Connections,
girders, and panel zones were within their acceptance criteria at SF= 0.4 meaning they did
not govern.
For FEMA-351, the local drift check controlled at SF= 1.1 resulting with an uncon-
servative margin of 0.7 (Figure 22b, and Table 4). This is for the failure mode where the mo-
ment connections lose their ability to resist gravity loads leading to vertical collapse of the
floor system. This did not occur in the test. The global drift check corresponds to a lateral
sway-type collapse like that in the test, and it had an unconservative margin of 0.4. Col-
umn compressive buckling checks failed at larger scale factors indicating this failure
mode did not govern.
NONLINEAR DYNAMIC PROCEDURE (NDP)

Figure 23 shows envelopes of peak deformations at slight yielding and during col-
lapse in the test. Analysis displacement patterns were similar to those from the test. Note
the change when going from slight yielding to collapse; displacements increased in the
first story as the collapse mechanism forms. Because nonlinear analysis captured this
effect, its displacements were in much better agreement to the test versus those from
linear analysis.
Figure 24a summarizes the ASCE-41 evaluation depicted in an incremental dynamic
Table 4. Governing conditions in PBE evaluations
Method Governing Condition
LDP At SF = 0.4, column demand-to-capacity ratios in first story attained limiting
冋册
ASCE-41 criterion:
PUF PUF 8 Mx My
0.2ⱕ ⱕ 0.5 and + + ⬇ 1.0
PCL PCL 9 mxMCEx myMCEy
where PUF = axial force based on limit-state analysis, Mx, My = bending moments
from analysis, PCL = lower bound compressive strength, MCEx, MCEy = expected
bending strengths, and mx, my = m-factors= 1.5 for CP evaluation.
LDP At SF = 1.1, drift ratio in second story attained limiting criterion for floor vertical
FEMA-351 collapse at 50% confidence.
␾C␭
D= where, D = drift ratio, C = 0.064 drift capacity for the connections,
␥␥a
␾, ␥, ␥a = parameters to account for bias and uncertainties in computation of D and
C, and ␭ = index parameter for 50% confidence.
NDP At SF = 0.4, columns in the first story attained limiting criterion: column plastic
ASCE-41 rotations= CP acceptance criterion= 0.8␪y ⬃ 0.002 rad.
NDP For SF ⬎ 0.5, drift ratios in first story increased abruptly and exceeded floor vertical
FEMA-351 collapse and side-sway collapse criteria.
analysis (IDA) format. The IDA curve flattened out for SF⬎ 0.5 reflecting collapse of the
analysis model at a smaller scale factor than the actual collapse intensity. For SF艋 0.4, the
IDA curve agreed favorably with the peak drifts from the test indicating that the elastic prop-
erties of the building were reasonably captured. Columns in the first story were correctly
identified as being the weak link in the building (Table 4). They failed the CP acceptance
criterion at a SF= 0.4, resulting in a conservative margin of two. Some connections, girders,
Figure 23. Comparison of peak displacements and drifts from NDP analyses and tests (analysis
at SF= 0.5 was incipient to collapse).
Figure 24. ASCE-41 and FEMA-351 evaluation by nonlinear dynamic procedure (NDP).
and panel zones experienced yielding, but their plastic rotations were within their acceptance
criteria at SF= 0.4, thus indicating they did not govern.
Figure 24b summarizes the FEMA-351 evaluation. The building became unstable
(collapsing) at drifts well below the capacities defined by the local and global CP crite-
ria (Table 4). Although the acceptance criterion for the local drift check was less than
that for the global check, both of these failure modes had the same conservative margin
of 1.6 because the CP drifts were on the IDA plateau. The column compressive forces
were within the acceptance criteria indicating that column buckling did not govern.
NONLINEAR STATIC PROCEDURE (NSP)

The pushover graphs exhibited sudden strength degradation, with collapse occurring
when the first story columns lost their flexural strength (plastic rotations exceeding
shoulder point a-values). ASCE-41 does not allow NSP when the strength ratio R
⬎ Rmax, as was the case here. Nevertheless, evaluation was continued to see the outcome.
Taking the target displacement 共␦t兲 as the instant when the first story columns exceeded their
acceptance criteria, a spectral acceleration was back-calculated using ASCE-41 equations.
This acceleration corresponded to the NS Takatori spectrum when scaled by SF⬃ 0.5.
This was similar to the scale factor causing collapse via nonlinear time history analysis and
hence the conclusions are the same as those for the NDP above.
DISCUSSION
The exercise revealed several points that warrant comment as follows:
Panel Zone and Composite Action Modeling. ASCE-41 has general discussion about
panel zones but is silent on slab-girder composite action, and hence there is latitude re-
garding modeling these aspects. This study used a simple linear model (no panel zones
or composite action) and a more detailed nonlinear model (explicit panel zones and
composite action increasing girder stiffness). The linear model was too flexible and the
Table 5. Comparison of nonlinear modeling parameters for columns
Shoulder Point Failure Point Residual Strength

Situation a-value (rad) b-value (rad) c-factor
Columns in lower stories of case study building:
Noncompact and Moderate Force, 0.2⬍ P / PCL ⬃0.002 ⬃0.003 0.2
⬍ 0.5
Columns in upper stories of case study building:
Noncompact and Low Force, P / PCL ⬍ 0.2 ⬃0.01 ⬃0.015 0.2
If columns were:
Compact and Moderate Force ⬃0.01 ⬃0.02 0.2
If columns were:
Compact and Low Force ⬃0.02 ⬃0.024 0.6
Successful Blind Analysis Model 0.01 0.06 0.4
P = axial force from analysis, and PCL = lower bound compressive strength.
When Noncompact and Moderate Force: a = 1␪y, b = 1.5␪y, where ␪y =
6EI共
ZFyeL
1−
P
Pye 兲
nonlinear model had a better fit to the test periods and peak displacements. Accounting
for these stiffness qualities therefore was appropriate.
The nonlinear model did not account for increased girder strength from composite
action and this may be the reason for girder yielding that was not observed in the test
(Figure 16). The effects were ignored because it was arguably considered conservative,
which may not always be the case in view of above.
Another aspect with panel zone and composite action modeling is the distribution of
yielding among the components. Gupta and Krawinkler (1999) found that such com-
puter modeling features (that can be difficult to define accurately) influence computed
member plastic rotations. Concentration of plastic deformations can occur in either the
girders or panel zones or shared between these elements and possibly also with the col-
umns, depending on the model formulation. ASCE-41 uses plastic rotations for accep-
tance criteria when using nonlinear analysis and this is a potential drawback.
Column Modeling and Evaluation. The first story columns controlled the building
collapse behavior, and how they were handled by the guidelines was crucial. The column
section was borderline compact per Japan code and AISC rules, but noncompact per
ASCE-41 and AISC Seismic (Table 1). The compactness aspect had a large impact.
Table 5 illustrates the effect of column section compactness and axial force on the
flexure backbone curve. There are huge reductions when going from compact to non-
compact and low to moderate force. Per ASCE-41, the lower story columns (noncompact
and moderate force) had very little ductility, essentially being type 3 force-controlled
members. However, ASCE-41 stipulates that column flexural behaviors are deformation-
controlled when P / PCL ⬍ 0.5, and hence they were evaluated as such.
Figure 25. Incremental dynamic analysis (IDA) showing influence of different shoulder point
a-values on collapse ruggedness.
Also listed are the parameters based on component tests used in the blind analysis
model that successfully simulated the building response, and these are much larger than
the ASCE-41 values. While it is understandable that ASCE-41 values may have been set
low in the absence of test data, unfortunately the guideline offers no comment to alert
the engineer on whether particular provisions were set conservative as was apparently
the case here.
The magnitude of the backbone curve a-value has a large effect on the building col-
lapse behavior. Figure 25 shows the IDA graphs from models that vary the shoulder
point a-value of the columns (the rate at which the columns loose strength after a-value
exceedance was held constant). With no strength deterioration, the building is robust,
having no collapse at SF= 1.5. With a-values set at 10-times the ASCE-41 values, analysis
agrees with the actual collapse intensity. While it is clear this is an important parameter,
other factors are likely to be influential as well, such as member strength and the rate at
which strength deteriorates.
Floor Vertical Collapse. The floor vertical collapse failure mode governed in the
FEMA-351 evaluations (local drift check). However, no distress in the girder-to-column
connections was observed in the experiment. FEMA-351 uses drift as a proxy for con-
nection demand that was not the case in the experiment; distortions occurred mostly in
the panel zones and columns. Using drift to check for connection damage was inappro-
priate for this case.
Moreover, floor vertical collapse, resulting from moment connection fracture, is a

failure mode that has yet to be documented in lab tests, computer simulations, or post-
earthquake surveys. It is relevant to note that this check did not have a strong consensus
among the SAC Joint Venture researchers (Lee and Foutch 2004); some wanted a high
confidence against its occurrence, while others wanted to omit the check altogether.
Guideline Interpretations. There were situations where judgments had to be made
about the intent of various provisions and how to comply with them. While example ap-
plications were undoubtedly part of the guideline development, they were not available
(referenced) to clarify guideline use. Example cases are essential to promote uniformity
in guideline application. In addition, the guidelines seemed to imply that linear proce-
dures were intentionally biased to be conservative versus nonlinear procedures. Al-
though this may have been intended, it was not specifically stated in the guidelines nor
was this apparent in the case study.
FINDINGS
This study is the first of its kind and it must be emphasized that it dealt with only one
collapse experiment, and it is therefore prudent not to over-generalize the findings.
Strictly speaking, the results are for the particular experiment, and caution must be ex-
ercised when applying them elsewhere. A summary of key points follow.
Shake Table Test Results. The building achieved its design objective by withstanding
shaking intensities much greater than that required by the Japan building code (design
basis corresponded to SF= 0.4). For SF艋 0.6, the behavior was fairly linear-elastic with the
building responding in a desirable side-sway manner having yielding in panel zones and at
the column bases. At SF= 1.0, the dynamic response changed markedly due to inelastic ac-
tions and the building then entered into a sequence of lengthened period oscillations having
progressively increasing amplitudes leading to collapse via a mechanism in the first story.
The story-mechanism occurred in spite of a strong-column and weak-beam design
because of factors not accounted for in customary calculations: large concurrent bending
moments in both column principal directions (biaxial bending), column local buckling,
and beam overstrength.
Test Simulation by Computer Analysis. State-of-practice nonlinear analysis was able
to simulate the experiment in terms of yielding pattern, peak displacements and instant
in time of collapse (blind analysis modeling). The computer model did not include tor-
sion or nonstructural components. Successful analysis used specific data about hysteretic
behavior from component tests that are not often available when creating computer mod-
els in engineering practice. The ASCE-41 modeling parameters, which would otherwise
be used, were much smaller (more conservative) than those from component tests and
using these led to less accurate results.
The deformation at which strength deterioration occurred in the columns was an im-
portant factor in the building collapse behavior (a-values). This and member strength
were probably the two most important parameters governing the building collapse rug-
gedness. Analysis also suggested that yielding in shaking runs prior to the final SF=1.0
collapse run had a negligible effect on the collapse behavior (pre-existing damage did
not appreciably weaken the building).
PBE Performance Assessment. ASCE-41 and FEMA-351 had mixed results regarding
characterization of collapse and neither approach was clearly superior. Collapse evalua-
tion using the guidelines was mostly conservative—erring on the safe-side. In terms of
safety margins, FEMA-351 predicted collapse more accurately than ASCE-41, but it in-
correctly identified floor vertical collapse as controlling.
ASCE-41 linear (LDP) and nonlinear (NDP and NSP) procedures had the same mar-
gins because the CP acceptance criteria was exceeded when the building was essentially
linear-elastic. This had a relatively large safety margin of about two, highlighting the fact
that the linear procedure m-factors, and the nonlinear deformation parameters (a-values
and plastic deformation acceptance criteria) were too small (over-conservative) for this
case.
The FEMA-351 linear procedure (LDP) produced unconservative margins, in part
because linear analysis could not capture the inelastic concentration of drift in the first
story occurring incipient to collapse. Hence, use of linear procedures for CP evaluation
was not well suited for this type of behavior. For the nonlinear procedure (NDP), the
building was unstable (collapsing) at drifts well below the local and global CP criteria at
50% confidence.
The highly detailed ASCE-41 and FEMA-351 provisions gave the impression that the
results should be quite accurate. However, the case study revealed that the predicted col-
lapse intensity was as much as a factor of two from the actual value.
ASCE-41 was suitable as a design tool for building rehabilitation because although it
was not especially accurate in predicting the intensity causing collapse, it did identify
the weak link in the building thereby targeting the right members for upgrading. Its pro-
cess of member-by-member checking is pragmatic within a design context. FEMA-351
is appealing as a building performance predictor because it provides probability esti-
mates for specific failure modes, but it does not have member-acceptance criteria.
RECOMMENDATIONS
The following needs became apparent during the course of the exercise. It must be
recognized they are from the study of a single collapse experiment, and hence may not
be universal. With this caveat, they are offered here for consideration by practicing en-
gineers and researchers:
1. Some ASCE-41 component-modeling parameters (a, b, and c) and acceptance
criteria (m-factors and plastic deformations) appear to be too small (conserva-
tive), and hence require improved calibration to better simulate actual behavior.
A recent supplement to ASCE-41 recognizes this for concrete structures (El-
wood et al. 2007), and a similar update ought to be created for steel moment-
frames.
2. ASCE-41 should adopt a quality ranking system to indicate how well
component-modeling parameters values (a, b, and c) and acceptance criteria
(m-factors and plastic deformations) are defined. It would be ideal if the existing
tables containing the numerical values had a column with High and Low desig-
nations for each component: High, meaning values are supported by consensus
research, and Low, meaning values are based mostly on expert opinion and thus
may be conservative. Engineers could then make better-informed decisions
when applying the guidelines.
3. The check for vertical collapse of floor systems (local drift check) in FEMA-351
needs to be improved. The current criterion, based on peak drift, does not dis-
tinguish between cases where drifts are excessive due to a column hinging
story-mechanism versus excessive deformations in the girder-to-column con-
nections, potentially leading to failure of the floor system.
4. Use of linear procedures for collapse prevention performance evaluation is dis-
couraged. Inelastic deformations dominate behavior incipient to collapse, and
they coalesce in certain stories to form the failure mechanism. It is unlikely that
any set of linear analysis modification factors can reliably capture these effects.
5. Better collapse prevention performance measures should be sought. ASCE-41
and FEMA-351 have a fundamental weakness when evaluating for collapse pre-
vention. Both use finite peak deformation as basis of judgment: component duc-
tility in ASCE-41 and story drift in FEMA-351. The case study demonstrated
when the shaking intensity is close to that causing collapse, the key peak defor-
mations can be sensitive to small changes in intensity. Reliably estimating peak
deformations in this intensity range is difficult. In addition, subtle differences in
the way certain aspects—e.g., panel zones or composite action—are modeled
can influence the distribution of plastic deformations among the components,
thereby injecting unpredictability in local ductility estimates. Thus, basing col-
lapse prevention acceptance criteria on finite peak deformations is open to ques-
tion. Global behavior parameters are better, and a performance measure based
on ground shaking intensity is described in the next point.
6. For collapse prevention evaluation, future generations of PBE ought to consider
the use of a safety margin informed by incremental dynamic analysis (IDA) as
the measure (Figure 26). The acceptance criterion is that the safety margin must
be greater than unity 共SIc / SId ⬎ 1兲. The shaking intensity causing collapse (in-
stability) is more meaningful than various finite deformations just on the thresh-
old of collapse and is thus preferred. It also avoids use of local member ductility
as acceptance criteria since these can be difficult to accurately estimate due to
computer modeling limitations. Such performance measures are the subject of
ongoing research (Deierlein et al. 2008).
7. Finally, guideline writers are encouraged to publish more example applications
illustrating intended provision use, and such examples should be referenced in
guideline commentary. This would go a long way toward making guidelines bet-
ter understood and embraced by practicing engineers.
Figure 26. A collapse prevention evaluation approach based on shaking intensity (shaking in-
tensity could be the scale factor applied to a suite of earthquake records, and deformation could
be roof displacement or peak story drift).
CLOSING THOUGHTS
Accurate characterization of collapse within the context of PBE is akin to the search
for the Holy Grail—perhaps a noble, but highly elusive, goal. The case study provides a
sobering illustration on some of the challenges. A relatively simple lab specimen build-
ing was evaluated by current PBE guidelines, yet the results had considerable variation
with some procedures:
• Being unconservative in terms of safety margin
• Identifying incorrect failure mode
• Having complicated methods, such that engineers could miss the forest for the
trees by being consumed with calculation minutia
• Producing an illusion of accuracy, given that PBE calculations for shaking inten-
sities causing rejection of collapse prevention acceptance criteria varied by a fac-
tor of two, versus the actual intensity causing collapse
Despite these shortcomings, the bottom-line conclusions from using the guidelines
were generally conservative, so “appropriate” outcomes—in terms of providing for pub-
lic safety—was mostly achieved. PBE is an evolving science, and engineers should rec-
ognize that current guidelines are probably more conservative than accurate, and they
should not yet be treated as building-code dogma. Case studies, like that presented in
this paper, are essential for gauging the guidelines against reality, even if they can only
provide fleeting glimpses, rather than definitive judgments.
ACKNOWLEDGMENTS
This study is part of the 2007 NEHRP Professional Fellowship in Earthquake Hazard
Reduction, administered by the Earthquake Engineering Research Institute and funded
by the Federal Emergency Management Agency. The financial support is greatly appre-
ciated. The shake-table experiment was part of the NEES/E-Defense collaborative re-
search program on steel structures, and their outstanding assistance is gratefully noted.
The authors are most appreciative of the contributions of many individuals who provided
valuable discussions on performance-based engineering, including: John Eidinger (G&E
Engineering), Tom Hale (California OSHPD), Ronald Hamburger (Simpson, Gumpertz
& Heger), Douglas Hohbach (Hohbach-Lewin), Abbie Liel (University of Colorado), R.
Jay Love (Degenkolb Engineers), Joseph Maffei (Rutherford and Chekene), Yuichi Mat-
suoka (E-Defense), Yoji Ooki (Tokyo Institute of Technology), Yuko Shimada (Tokyo In-
stitute of Technology), Mason Walters (Forell/Elsesser), and members of the Existing
Buildings Committee of the Structural Engineers Association of Northern California
(SEAONC), chaired by Colin Blaney (Crosby Group) and Russell Berkowitz (Forell/
Elsesser). The authors also appreciate the useful comments provided by anonymous re-
viewers. However, all opinions and conclusions expressed herein are solely those of the
authors.
REFERENCES
AISC, 2005. Specification for Structural Steel Buildings, American Institute of Steel Construc-
tion (AISC).
AISC Seismic, 2005. Seismic Provisions for Structural Steel Buildings, American Institute of
Steel Construction.
ASCE-41, 2006. Seismic Rehabilitation of Existing Buildings, American Society of Civil Engi-
neers, ASCE Standard ASCE/SEI 41–06, prepublication edition.
ASCE-31, 2003. Seismic Evaluation of Existing Buildings, American Society of Civil Engi-
neers, ASCE Standard ASCE/SEI 31-03.
Deierlein, G., Liel, A., Haselton, C., and Kircher, C., 2008. ATC 63 methodology for evaluating
seismic collapse safety of archetype buildings, Proceeding of ASCE Structures Congress,
Vancouver, British Columbia, Canada, April 24–26.
EERI Newsletter, 2008. Earthquake Engineering Research Institute, February, also see May
2007 and November 2007 Newsletters.
Elwood, K. J., Matamoros, A. B., Wallace, J. W., Lehman, D. E., Heintz, J. A., Mitchell, A. D.,
Moore, M. A., Valley, M. T., Lowes, L. N., Comartin, C. D., and Moehle, J. P., 2007. Update
to ASCE/SEI 41 concrete provisions, Earthquake Spectra 23, 493–523.
FEMA-351, 2000. Recommended Seismic Evaluation and Upgrade Criteria for Existing Welded
Steel Moment-Frame Buildings, Federal Emergency Management Agency, FEMA Publica-
tion No. 351.
FEMA-356, 2000. Prestandard and Commentary for the Seismic Rehabilitation of Buildings,
Federal Emergency Management Agency, FEMA Publication No. 356.
FEMA-273, 1997. NEHRP Guidelines for the Seismic Rehabilitation of Buildings, Federal
Emergency Management Agency, FEMA Publication No. 273.
Gupta, A., and Krawinkler, H., 1999. Seismic Demands for Performance Evaluation of Steel
Moment Resisting Frame Structures, John A. Blume Earthquake Engineering Center Report
No. 132, Stanford University.
Kasai, K., Ooki, Y., Motoyui, S., Takeuchi, T., and Sato, E., 2007. E-Defense tests on full-scale
steel buildings: part 2—experiments using dampers and isolators, Proceeding of ASCE
Structures Congress, Long Beach, California.
Lee, K., and Foutch, D. A., 2004. Closure to “Seismic performance of pre-Northridge steel
frame buildings with brittle connections” by Kihak Lee and Douglas A. Foutch, J. Struct.
Eng. 130, 691–692.
Maison, B. F., 1992. PC-ANSR: A Computer Program for Nonlinear Structural Analysis, avail-
able from National Information Service for Earthquake Engineering, NISEE/Computer Ap-
plications, University of California, Berkeley.
Maison, B. F., and Neuss, C., 1983. SUPER-ETABS: An Enhanced Version of the ETABS Pro-
gram, available from National Information Service for Earthquake Engineering, NISEE/
Computer Applications, University of California, Berkeley.
Matsuoka, Y., Suita, K., Yamada, S., Shimada, Y., and Akazawa, M., 2008. Nonstructural com-
ponent performance in four-story frame tested to collapse, Proceeding of the 14th World
Conference on Earthquake Engineering, Beijing, China.
OpenSees 2007. An Open Source Structural Analysis Program for Earthquake Engineering,
available from the Pacific Earthquake Engineering Research Center, University of Califor-
nia, Berkeley.
Powell, G. H., 2007. PERFORM-3D Nonlinear Analysis and Performance Assessment for 3-D
Structures, available from Computer & Structures International, Inc., Berkeley, California.
SEAOC, 1999. Recommended Lateral Force Requirements and Commentary, Seismology Com-
mittee of Structural Engineers Association of California, Section C402.5, September.
Searer, G., Paret, T., and Freeman, S., 2008. ASCE-31 and ASCE-41: what good are they? Pro-
ceedings of the ASCE Structures Congress, Vancouver, British Columbia, Canada.
Suita, K., Yamada, S., Tada, M., Kasai, K., Matsuoka, Y., and Shimada, Y., 2008. Collapse ex-
periment on four-story steel moment frame: part 2, Proceeding of the 14th World Conference
on Earthquake Engineering, Beijing, China.
Tada, M., Ohsaki, M., Yamata, S., Motoyui, S., and Kasai, K., 2007. E-Defense tests on full-
scale steel buildings, part 3—analytical simulation of collapse, Proceeding of the ASCE
Structures Congress, Long Beach, California.
Yamada, S., Suita, K., Tada, M., Kasai, K., Matsuoka, Y., and Shimada, Y., 2008. Collapse ex-
periment on four-story steel moment frame: part 1, Proceeding of the 14th World Conference
on Earthquake Engineering, Beijing, China.
(Received 3 March 2008; accepted 26 March 2009兲

ASCE-41 and FEMA-351 Evaluation: of E-Defense Collapse Test

Uploaded by

Copyright:

Available Formats

ASCE-41 and FEMA-351 Evaluation: of E-Defense Collapse Test

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ASCE-41 and FEMA-351 Evaluation: of E-Defense Collapse Test

Uploaded by

Copyright:

Available Formats

EARTHQUAKE ENGINEERING PRACTICE

ASCE-41 and FEMA-351 Evaluation

A welded steel moment-frame building is used to assess performance-

The welded steel moment-frame (WSMF) is a relatively young structural system,

CASE STUDY BUILDING

SHAKE TABLE EXCITATIONS

Figure 2. Exploded view of typical girder-to-column framing with shop-welded components

Table 1. Building properties

As shown in Figure 5, the spectra mostly envelopes a Maximum Considered Earthquake

Figure 5. Takatori earthquake spectra at 5% damping.

Table 2. Experiment summary

0.05 No yielding. Small amplitude linear-elastic behavior.

Figure 6. Building collapse mechanism (no scale).

EFFECT OF NONSTRUCTUAL COMPONENTS

STRONG-COLUMN AND WEAK-BEAM DESIGN

COMPUTER SIMULATION OF EXPERIMENT

LINEAR DYNAMIC PROCEDURE (LDP) MODELS

NONLINEAR DYNAMIC (NDP) AND STATIC (NSP) PROCEDURES MODELS

Figure 18. Computer models of case study building.

Figure 19. Computer modeling of girder-to-column framing.

Acceptance criteria Acceptance criteria:

No checking of drifts. No checking of member ductilities.

EVALUATION BY ASCE-41 AND FEMA-351

Figure 20. Computer modeling of nonlinear flexure behaviors.

LINEAR DYNAMIC PROCEDURE (LDP)

NONLINEAR DYNAMIC PROCEDURE (NDP)

Table 4. Governing conditions in PBE evaluations

Method Governing Condition

LDP At SF = 0.4, column demand-to-capacity ratios in first story attained limiting

NONLINEAR STATIC PROCEDURE (NSP)

Table 5. Comparison of nonlinear modeling parameters for columns

Shoulder Point Failure Point Residual Strength

Moreover, floor vertical collapse, resulting from moment connection fracture, is a

You might also like