Beruflich Dokumente
Kultur Dokumente
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a r t i c l e
i n f o
Article history:
Received 19 August 2011
Revised 19 June 2012
Accepted 31 May 2013
Available online 2 July 2013
Keywords:
Progressive collapse
Steel buildings
Column failure
Load redistribution
Collapse experiment
a b s t r a c t
A eld experiment and numerical simulations were performed to investigate the progressive collapse
potential of an existing steel frame building. Four rst-story columns were physically removed from
the building to understand the subsequent load redistribution within the building. Experimental data
from the eld tests were used to compare and verify the computational models and simulations. Due
to the scarcity of data from full-scale tests, the experimental data produced during this research is a valuable addition to the state of knowledge on progressive collapse of buildings. The progressive collapse
design guidelines typically recommend simplied analysis procedures involving instantaneous removal
of specied critical columns in a building. This paper investigates the effectiveness of such commonly
used progressive collapse evaluation and design methodologies through numerical simulation and experimental data.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
Progressive collapse is generally dened as small or local structural failure resulting in damage and failure of the adjoining members and, in turn, causing total collapse of the building or a
disproportionately large part of it. Progressive collapse of building
structures is initiated by loss of one or more vertical load carrying
members, usually columns. After one or more columns fail, an
alternative load path is needed to transfer the load to other structural elements. If the neighboring elements are not designed to resist the redistributed loads, failure will happen with further load
redistribution until equilibrium is reached, resulting in partial or
total collapse of the structure.
Progressive collapse is triggered by abnormal loading that
causes local failure of one or more columns if the building lacks
sufcient ductility, continuity and/or redundancy. The local or
complete collapse may cause signicant casualties and damage
disproportionate to the initial failure. A notable example is partial
collapse of the Ronan Point apartment building in London. An accidental gas explosion in a corner kitchen on the 18th oor initiated
progressive collapse of the 24-story building in 1968. This event
triggered extensive progressive collapse research and led to development of design guidelines for the prevention of progressive
collapse [13].
The World Trade Center 7 (WTC 7) in New York City was a 47story ofce building adjacent to the WTC towers (WTC 1 and 2)
that collapsed following the terrorist attacks of September 11,
2001. WTC 7 collapsed several hours after the collapse of twin
WTC towers. The NIST report [11] concluded that: An initial local
failure occurred at the lower oors (below oor 13) of the building
due to re and/or debris induced structural damage of a critical
column (the initiating event) which supported a large span oor
bay with an area of about 2000 square feet. Vertical progression
of the initial local failure occurred up to the east penthouse, as
the large oor bays were unable to redistribute the loads, bringing
down the interior structure below the east penthouse. Horizontal
progression of the failure across the lower oors triggered by
damage due to the vertical failure, resulting in a disproportionate
collapse of the entire structure. The FEMA 403 [6] study emphasized the signicance of res on the collapse. This is a good example of disproportionate collapse caused by debris and/or re
induced failure of a column or columns in a tall steel building. In
this research, several columns were sequentially removed from a
building, which can resemble the initial debris damage and gradual
and intensifying re damage or a various other loads.
Failure of one or more columns in a building and the resulting
progressive collapse may be a result of a variety of events with different loading rates, pressures or magnitudes. The magnitude and
probability of natural and man-made hazards are usually difcult
to predict. Therefore, most of the current progressive collapse
design guidelines are threat-independent and do not intend to
prevent such local damage, e.g., ACI 318 [1]. Rather, their purpose
665
666
Fig. 1. (a) Building before demolition, (b) four rst-story columns exposed, (c) columns removed, and (d) building during the gradual demolition process.
Table 1
Column and beam sections of the Ohio Union building.
Column section
Beam section
Column number
Column type
Beam number
Beam type
C1
C2
C3
C4
C5
C6
C7
C8
C9
10
12
12
10
10
10
10
10
10
B1
B2
B3
B4
B5
B6
B7
B8
B9
24
21
16
21
18
14
14
24
18
WF
WF
WF
WF
WF
WF
WF
WF
WF
72
133
120
100
89
54
112
60
33
B 76
B 68
B 58
WF 62
WF 50
B 17.2
B 22
WF 76
WF 45
columns were sequentially removed in the following order: columns 27, 22, 2, and 7.
Linear static, nonlinear static, linear dynamic, and nonlinear
dynamic analysis methods, in order of increasing complexity, can
be used to analyze a structure to investigate its structural behavior.
Researchers investigated the advantage and disadvantage of each
of these procedures for progressive collapse analysis [14]. A complex analysis is desired to obtain more realistic results representing the actual nonlinear and dynamic response of the structure
during the progressive collapse. However, both GSA and DOD
guidelines recommend the simplest method, linear static, for the
progressive collapse analysis since this method is cost-effective
and easy to perform. One of the objectives of this paper is to
Fig. 2. Longitudinal frame elevation including beam and columns sections (see Table 1).
667
Fig. 4. Plan view of strain gauge placement in Ohio Union building with columns and beam labeled (15 strain gauges are shown in the circles).
comparing the calculated DCR values based on the recommendations of GSA guidelines. DCR for moment is dened as the ratio
of the maximum moment demand Mmax of the beam or column
calculated from linear elastic analysis to its expected ultimate
moment capacity Mp, which is calculated as the product of plastic
section modulus and yield strength. In Mp calculations for columns,
the effect of the axial load is neglected because the column axial
loads were relatively small and did not signicantly affect the
moment capacity of the cross section.
DCR
M max
Mp
668
Fig. 5. (a) Two-dimensional SAP2000 model with frame member numbers and (b) three-dimensional SAP2000 model of the Ohio Union building (circled columns are
removed in the order shown).
Fig. 6. Moment diagrams: (a) after one column was removed, (b) after two columns were removed, (c) after three columns were removed, and (d) after four columns were
removed.
Fig. 8 shows DCR values for each frame member for all column
removal cases. Frame member numbers up to 45 are columns, and
beams are numbered from 46 to 85 (Fig. 5a). After the rst column
was removed, DCR values for all columns and beams were below
0.5. The DCR values after the loss of second column was similar
to those of third column loss, all of which were less than 1.5. The
669
2.83
0.34
2.25
0.32
0.21
1.56
0.16
0.22
0.09
0.44
0.17
0.63
1.54
0.91
0.24
0.07
0.12
0.34
0.45
0.14
0.18
0.41
0.40
0.17
0.26
0.56
0.48
0.07
0.86
0.55
0.35
0.41
1.44
1.21
0.45
0.51
0.37
0.42
0.00
0.94
0.58
0.91
0.47
0.45
2.24
0.83
0.56
0.89
2.14
0.38
2.38
1.10
0.82
0.40
0.27
0.42
0.32
0.15
0.80
0.49
0.77
0.39
0.11
0.03
Fig. 7. Moment diagram and corresponding DCR values after the loss of four columns in the Ohio Union building.
Fig. 8. Change in DCR values of each frame member for all cases.
DCR values for columns were remarkably increased after the fourth
column was lost. Columns were impacted more than beams when
all four columns were removed from the frame. As acceptance criteria, the maximum DCR limits specied in GSA [8] are 2.0 and 3.0
for columns and beams for the test building, respectively. After all
four columns were removed, no beams and ve columns (i.e., columns 8, 9, 10, 20 and 25) exceeded the DCR criteria. The change in
DCR values for beams was not signicant compared with that of
columns. The DCR values of beams were always less than 1.0. This
is probably due to potential redistribution of loads to the adjacent
beams in the analyzed frame.
After four columns were removed, the building was more
susceptible to progressive collapse. This was also reected in the
maximum displacements calculated from linear static analysis.
As columns were sequentially removed, the maximum vertical displacements were calculated as 11.40, 11.54, 30.73, and 17.93 cm at
the joints immediate above the rst (column 27), second (column
22), third (column 2) and forth (column 7) removed columns,
respectively.
670
Fig. 11. Displacement of joints above each removed column after all columns were
removed.
Fig. 12. Comparison of DCR values determined from 2-D and 3-D linear static
analysis after removal of four columns.
signicantly deformed in the transverse direction after each column removal. 2-D linear static analysis may lead to limited and
underestimated demands for beams.
More interestingly, it was observed that DCR values calculated
from the 3-D linear static analysis were smaller than those from
2-D linear static analysis for columns and most beams. As shown
in Fig. 12, all members had DCR values of less than 1.5, and satised GSA acceptance criteria of 2.0 for columns and 3.0 for beams.
This could be mainly due to contribution of transverse beams. The
transverse beams can distribute loads to the connected columns
and beams in the transverse direction, leading to a decrease of
force demands in structural members.
Table 2 shows the comparison of maximum vertical displacements calculated from 2-D and 3-D analyses. 3-D models showed
lower maximum displacements than 2-D models for both linear
static and nonlinear dynamic analysis. Similar to the DCR results,
the transverse beams connected to the interior columns and the
beams increased the overall resistance of structure, leading to
smaller deformations in the 3-D model. As shown in Table 2, Linear
static analysis resulted in higher maximum vertical displacements
than nonlinear dynamic analysis in both 2-D and 3-D models. For
example, the maximum vertical displacement calculated from
2-D linear static analysis was 30.73 cm at Joint 3 while that from
671
2-D model
3-D model
Joint
Joint
Joint
Joint
1
2
3
4
11.40
11.53
30.73
17.93
Permanent
7.11
7.24
20.47
11.33
6.05
6.12
17.93
9.98
7.85
7.95
10.03
7.37
Permanent
4.22
4.27
5.64
3.68
3.66
3.71
5.08
3.38
Table 3
Plastic hinge rotations (h, degree) at the location where each column was removed after all columns removal.
Removed columns
Joint 1
Joint 2
Joint 3
0.53
0.54
1.80
0.31
0.32
0.50
the 2-D nonlinear dynamic analysis was 20.47 cm. It seems that
the impact factor of 2 (i.e., dead loads multiplied by 2) in linear static analysis led to very conservative results. Marjanishvili reported
that a more complicated analysis method such as nonlinear dynamic analysis may result in less severe structural response, due
to more accurate estimates of load distribution and less stringent
evaluation criteria.
Table 3 shows plastic hinge rotations at the location where
columns were removed after four columns removal. Plastic hinge
rotation was chosen as acceptance criteria for the nonlinear
dynamic analysis [8], which also possibly evaluates whether the
moment connection of the frame is strong enough to survive the
excessive moment rotation of the joints. Plastic hinge rotation
angle for beam members on each side of the removed column
can be measured between horizontal line and tangent to maximum
deected shape, which is dened by Eq. (2).
h tan1
dmax
L
where h is the maximum hinge rotation, dn the maximum displacement of columns at the location where the column is removed, and
L is the beam length or column spacing in the longitudinal direction.
As shown in Table 3, hinge rotations calculated from the 3-D
model were smaller than those from the 2-D model, because of
lower maximum displacement values in 3-D nonlinear dynamic
analysis. The maximum plastic hinge rotation was only 1.80 at
the hinge above third removed column (Column 2) in linear static
analysis. For both 2-D and 3-D nonlinear dynamic procedure, the
values of plastic hinge rotation were much smaller than 12 of
GSA [8] criteria, indicating that the Ohio Union building was not
susceptible to progressive collapse. Considering that no signicant
deformations were observed during eld testing, GSA criteria for
Table 4
Comparison of change in strain (De) obtained from the eld test after last column torching with that calculated from 2-D and 3-D analyses after all columns removal (% difference
was indicated in parentheses).
Strain gauge
Field test
2 (Column)
4 (Column)
8 (Column)
9 (Column)
11 (Column)
15 (Beam)
55 106
37 106
29 106
28 106
33 106
37 106
2-D model
3-D model
32 106
17 106
4 106
20 106
7 106
46 106
(42%)
(54%)
(86%)
(29%)
(79%)
(24%)
672
were reached during this study based on the evaluation of experimental data and structural analysis of the test building.
The measured strain data compared relatively well with the
analysis results. In particular, 3-D model was more accurate than
the 2-D model, because 3-D models can avoid overly conservative
solutions as well as account for 3-D effects such as contribution of
transverse beams to overall resistance of the frame. The 3-D model
had lower DCR values and vertical displacements than 2-D model,
which was possibly due to inclusion of transverse beams in the 3-D
model. The 3-D model is believed to be more realistic than 2-D
model for the progressive collapse analysis.
The strain values calculated from the nonlinear dynamic analysis were smaller than those from the linear static analysis, and
were closer to the measured strains. Also, linear static analysis
showed higher DCR values and vertical displacements than nonlinear dynamic analysis for both 2-D and 3-D models. The amplication factor of 2 required for the dead load in linear static analysis
may lead to very conservative analysis results.
For future research, it would be better to consider the actual
material properties and connections of the building in the analytical models in order to obtain more reliable results.
Acknowledgements
This research was partially funded by the National Science
Foundation (CMMI 0745140), American Institute of Steel Construction, and URS Corporation; this is gratefully acknowledged. The
authors would like to thank SMOOT Construction, Loewendick
Demolishing Contractors, and the Ohio State University for providing access to the test building and help with the experiment.
References
Fig. 13. Comparison of calculated strains and strain measured by (a) Strain Gauge 4
and (b) Strain Gauge 15 after all columns were removed.
[1] ACI 318-11. Building code requirements for structural concrete and
commentary. Farmington Hills, MI: American Concrete Institute (ACI); 2011.
[2] AISC. Manual of steel construction. 6th ed. American Institute of Steel
Construction (AISC); 1969.
[3] ASCE. Minimum design loads for buildings and other structures. Reston, VA:
American Society of Civil Engineers (ASCE); 2005.
[4] DOD. Design of buildings to resist progressive collapse. Unied Facilities
Criteria (UFC) 4-023-03. Department of Defense (DOD); 2005.
[6] FEMA 403. World trade center building performance study: data collection,
preliminary observations and recommendation. Report: FEMA 403.
Washington, DC: Federal Emergency Management Agency (FEMA); 2002.
[8] GSA. Progressive collapse analysis and design guidelines for new federal ofce
buildings and major modernization projects. Washington, DC: General
Services Administration (GSA); 2003.
[11] NIST SP 1000-5. Progress report on the federal building and re safety
investigation of the world trade center. Gaithersburg, MD: National Institute
of Standards and Technology; 2004.
[12] NIST. The collapse of the world trade center towers. Final Report.
Gaithersburg, MD: National Institute of Standards and Technology (NIST);
2005.
[13] NIST. Best practices for reducing the potential for progressive collapse in
buildings. Gaithersburg, MD: National Institute of Standards and Technology
(NIST); 2006.
[14] Powell G. Progressive collapse: case study using nonlinear analysis. ASCE
structures congress and forensic engineering symposium. New York, NY; April
2024, 2005.
[15] SAP2000. SAP 2000 advanced structural analysis program. Version 12.
Berkeley, CA, USA: Computers and Structures, Inc. (CSI); 2009.
[16] Song BI. Experimental and analytical assessment on the progressive potential
of existing buildings. Masters thesis. The Ohio State University; 2010. p. 125.
[17] Song B, Sezen H. Evaluation of an existing steel frame building against
progressive collapse. In: ASCE structures conference, Austin, Texas; April 30
May 2, 2009.
[18] Song B, Sezen H, Giriunas K. Experimental and analytical assessment on
progressive collapse potential of actual steel frame buildings. In: ASCE
structures conference and North American steel construction conference,
Orlando, Florida; May 1215, 2010.