Professional Documents
Culture Documents
Asce-Collapse Performance of Existing Buildings After Loss of Load-Bearing Walls
Asce-Collapse Performance of Existing Buildings After Loss of Load-Bearing Walls
Asce-Collapse Performance of Existing Buildings After Loss of Load-Bearing Walls
H. Sezen1
1
Dept. of Civil, Environmental and Geodetic Engineering, Ohio State Univ., 470
Downloaded from ascelibrary.org by University Of British Columbia on 07/14/16. Copyright ASCE. For personal use only; all rights reserved.
Abstract
© ASCE
INTRODUCTION
The ongoing research project involves testing of actual buildings to monitor
their collapse behavior while the columns and load bearing walls are physically
removed from the buildings. The main research goal is to investigate progressive
collapse response of these existing buildings thorough experimental testing and
numerical simulations. The research aims to investigate the effectiveness of
commonly used progressive collapse evaluation and design methodologies and
propose efficient methods and recommend improvements for the structural models
Downloaded from ascelibrary.org by University Of British Columbia on 07/14/16. Copyright ASCE. For personal use only; all rights reserved.
and existing analysis methods. Prior to demolition of the test buildings, one to four
columns or load bearing masonry walls were physically removed first story of the
buildings and structural response was recorded during the experiments. The
computational phase involves: 1) development of detailed and simplified building
models for static and dynamic analysis, 2) implementation of current code
standards and guidelines for assessing progressive collapse potential of the test
buildings, and 3) comparison of simulated and measured building response data.
The current experimental and computational research enhances the understanding
of building progressive collapse mechanism.
Vertical or progressive collapse can be defined as total collapse of a
building or a disproportionately large part of it, which may be a result of small or
local structural failure leading to propagation of failure and collapse of the
adjoining members in the building. Progressive collapse of building structures is
typically initiated by loss of one or more vertical load carrying members such as
columns or structural load carrying walls. After one or more columns or walls fail,
an alternative load path is needed to transfer the load to other structural
components. If the neighboring components are not designed to resist and
redistribute the additional loads, failure will happen with further load redistribution
until equilibrium is reached, sometimes after a sizeable part of the structure
collapses.
American Society of Civil Engineers (ASCE 7, 2010), General Services
Administration (GSA, 2003), Department of Defense (Unified Facilities Criteria,
UFC, 2009), and American Concrete Institute (ACI 318, 2011) have developed
criteria and guidelines to evaluate, design and improve structural integrity and
progressive collapse resistance of existing and new buildings. Indirect methods
involve prescriptive requirements for minimum strength, ductility and element
continuity to increase the overall robustness and integrity of the structure. The
direct design approach considers resistance to progressive collapse explicitly
during the design process through the alternate path method or specific local
resistance method. The alternate path method requires redistribution of the loads
within the structure after the loss of a primary structural element. The specific local
resistance method requires design of key primary members against a specific
extreme load or event.
Generally, design code requirements prescribe simplified analysis
procedures requiring instantaneous removal of certain vertical load carrying
members. Collapse potential can be investigated using the demand-to-capacity ratio
(DCR) calculated for each structural element. DCR is defined as the ratio of the
force (moment, shear, or axial force) calculated after the instantaneous loss of a
column and the corresponding capacity of the member. In this study, each test
© ASCE
building are being analyzed using the load combinations specified by the GSA and
the corresponding DCRs are being calculated. The acceptance criteria provided by
the GSA is then be used to assess the potential for progressive collapse.
BUILDING EXPERIMENTS
Structural response of nine existing buildings was recorded during and after
the removal of one or more perimeter columns or walls. The goal was to determine
whether actual building performance could be predicted using the available
Downloaded from ascelibrary.org by University Of British Columbia on 07/14/16. Copyright ASCE. For personal use only; all rights reserved.
analysis techniques. Basic modeling and analysis of the first few buildings have
been completed (Table 1). All but two buildings had four stories. Seven of the test
buildings were located on The Ohio State University campus in Columbus, Ohio.
Four of the nine exterior columns were removed from The Ohio Student
Union building, as shown in Figure 1. The steel columns were first torched near
their top and bottom. Only a small portion of the flange was left intact when the
cross sections were cut. The middle column segment between the torched sections
Figure 1 (a) Ohio Union building, (b) four columns removed, and (c) middle of
column removed
© ASCE
was then pulled out using a steel cable (Figure 1.c). The columns were removed
within a very short period of time representing an instantaneous column removal as
recommended in the design guidelines. Columns and beams neighboring the
removed columns were instrumented to monitor the redistribution of loads during
the process of column removals. Detailed discussion of experimental tests and
numerical simulations are provided in Song (2010), and Song and Sezen (2013).
columns from the two- and three-dimensional frame models in SAP2000 (2009).
Displacements and internal force demands calculated from linear static analysis
showed that the building became most susceptible to progressive collapse after the
last column was removed.
Figure 2 Column removal procedure for dynamic analysis, and time history
function used to model sudden column loss in dynamic analysis
A 3-D model of the Ohio Union building was also analyzed. Beam DCR values
calculated from the 3-D linear static analysis showed an opposite trend compared
to 2-D results. Beams were more influenced by the column loss while the column
DCR values were reduced significantly. The maximum DCR value for beams was
1.49 while that of columns was 0.96. The reason that beams had higher DCR
values than columns in the 3-D model was possibly due to the larger deformation
and participation of beams in the transverse direction. It was found that beams,
especially in the top story, were significantly deformed in the transverse direction
© ASCE
after each column removal. 2-D model may not capture this and lead to limited and
underestimated demands for beams. All members had DCR values of less than 1.5.
This is possibly 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.
The steel frame building, Bankers Life and Casualty Company (BLLC) building
was located in a suburb Chicago (Naperville, Illinois) (Figure 3). The building was
modeled and analyzed using linear static and nonlinear dynamic analysis
procedures. The calculated DCR values, after all four columns were removed, are
shown in Figure 6. DCR values for many structural members were larger than the
specified GSA limits. Experimental details and analysis results are provided in
Song (2010).
Figure 3 Bankers Life and Casualty Company building before the experiment
(left), and deflected 3-D SAP2000 model with corresponding DCR values after
loss of four columns (right)
Strains calculated from static analysis of the 2-D and 3-D models are
compared with the average strain measured by a selected strain gauge attached on a
column during the removal of each of the four columns. Strains were calculated by
considering the combination of axial load and a bending moments from the 2-D
and 3-D SAP2000 analyses. Details of strain calculations from SAP2000 models
are provided in Song (2010) and Sezen et al. (2014). Figure 4 compares the
calculated and measured strains. 2-D model significantly overestimated the
measured response. The results from the 3-D model were in closer agreement with
the experimental results than the 2-D model.
© ASCE
Three first story columns, including one corner column, were removed from
the Swallen’s building (Figure 5). The RC flat slab structure with drop panels had
no beams in the lower floors and included a grid of 130 circular columns with a
uniform spacing of 6 m. The demolition contractor used a processor to crush and
remove the columns from the building. Redistribution of loads from the removed
columns to the neighbouring columns was successfully monitored using 14 strain
gauges attached on the longitudinal steel bars of selected columns around the
removed columns. A series of models, ranging from very detailed to very simple,
are developed and validated for various modeling simplifications (Morone and
Sezen, 2014). Ultimately, a procedure is developed for creating simplified models
and a spring model consisting of few structural elements (Figure 6).
Figure 5 Swallen’s Department Store building and its detailed 3-D computer
model
Blackburn House
Four wall piers were removed from the first story of the four-story
BlackburnHouse in November 2014. Only masonry block walls were used as the
main vertical load carrying system in this building. Construction equipment was
used to remove the exterior walls as shown in Figure 7. Unmanned aerial vehicle
© ASCE
(UAV) was used to capture and construct 3-D images of the building. Similarly,
several high resolution cameras were used to monitor the deformation of the walls
above the removed wall piers using the 3-D markers shown in the figure on the
right in Figure 7. LVDT’s were also used to measure the vertical displacements in
the tested region. Currently the finite element and simplified models of the building
is being developed and analyzed, and laboratory experiments are being conducted
on masonry materials retrieved from the building.
Figure 7 Blackburn House (photo taken by drone - left), and four wall sections
are removed (right)
© ASCE
Boyd Laboratory was a three story reinforced concrete frame building. One
of the first-story perimeter columns was removed using a 2-ton wrecking ball
attached to a track hoe. Eighteen strain gauges and two LVDTs were used to record
strains and displacements during testing of the Boyd Laboratory. Although the
wrecking ball was used successfully to remove the column, it was not as effective
as the processor or crusher, which was used to remove columns from the Swallen’s
building (Figure 5) and Johnston Hall (Figure 8).
Downloaded from ascelibrary.org by University Of British Columbia on 07/14/16. Copyright ASCE. For personal use only; all rights reserved.
Johnston Hall was a four story reinforced concrete frame structure with an
approximately 40 m by 13.5 m rectangular floor plan including twelve bays in the
longitudinal direction and three bays in transverse direction. Typical story height of
the building was approximately 4 m. Vertical displacement of the joint above the
removed column and strains at 18 locations on the beams and columns neighboring
the removed column were recorded during the experiment. The column shown in
Figure 8 was cut at mid-height with the help of concrete crushing/pinching
demolition equipment in a short period of time. Vertical displacements, and
changes in axial loads and moments in the neighboring beams and columns were
recorded using displacement sensors (LVDT’s) and strain gauges attached on the
steel bars of the columns and beams around the removed column.
Figure 8 Column being removed (left), and column removed (middle) from
Johnston Hall; and measured and calculated strains in the column on the
right side of the removed column (right)
Linear static (LS) and nonlinear dynamic (NLD) analyses of 3-D model of
the building were performed by removing the column from the model as described
in Figure 2. Figure 8 compares the calculated strains from static and dynamic
analyses with strains measured by three different strain gauges (S.G. 10, 11 and 12)
attached on the column next to the removed column in Johnston Hall. Nonlinear
dynamic analysis is proven to be more realistic and accurate than linear static
© ASCE
analysis, which is simpler but usually more conservative. In almost all test
buildings, internal forces and deformations calculated from linear static analysis
were typically larger than those from nonlinear dynamic analysis. In Figure 8, the
average measured axial strain is very close to the calculated dynamic strains.
removed using the same demolition equipment used for Johnston Hall (Figure 8).
Columns and beams near the removed corner column were instrumented using a
total of 18 strain gauges. Computational simulations show that the ribbed floor
slabs contributed significantly to distribute the gravity loads within the structure.
As a result, 2-D frame model simulations were less reliable, and therefore a 3-D
model had to be used to better understand how the loads were redistributed within
the frame after the column was removed.
Haskett Hall was a steel framed structure built in 1925 and was remodeled
in 1970s in which additional secondary steel columns and beams were added. The
structure had rectangular floor plan with an area of approximately 29 m by 41 m
having six bays along the longer side and five bays in the shorter side. Typical
story height of the building was 3.67 m. One column along the longitudinal side of
the building near the corner was removed using the same demolition equipment
used earlier for Johnston Hall and Aviation Building experiments (Figure 8). Three
LVDTs and 16 strain gauges were installed to record strain and displacement data
on beams and columns surrounding the removed column. Two of the LVDTs
recorded vertical displacement and one was used to measure horizontal
displacement or slip at the joint where column was removed. A residual vertical
displacement of 18 mm was recorded at the joint above the removed column.
© ASCE
lost. The remaining loads (approximately 20-30%) on the removed column are
carried by the interior columns.
ACKNOWLEDGMENTS
This research was partially funded by the National Science Foundation (CMMI
0745140, 1130397 and 1435446), and American Institute of Steel Construction;
this is gratefully acknowledged. The author would like to thank undergraduate and
graduate students involved in this research since 2007 including Brian Song, Kevin
Giriunas, Gregory Ullom, Justin Morone, Shadab Lodhi, Curtis Wood, Ebiji Akah,
John Wade, Hongsen Shi, Kai Li, and Michael Savage. The author also would like
to thank SMOOT Construction, Loewendick Demolishing Contractors, Messer
Construction Co, and The Ohio State University for providing access to the test
buildings and help with the experiments.
REFERENCES
© ASCE