Engineering Structures 42 (2012) 154–167

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Dynamic performance of RC beam-column substructures under the scenario of the

loss of a corner column—Experimental results
Qian Kai, Bing Li ⇑
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore 639798, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: Six one-third scale reinforced concrete (RC) beam-column substructures, each with different design
Received 20 June 2011 detail, span length and span aspect ratio, were designed and tested to investigate the dynamic load redis-
Revised 23 April 2012 tribution performance of RC substructures following predefined initial damage. The initial damage was
Accepted 24 April 2012
caused by the sudden removal of the corner support. The tested variables include: the longitudinal
Available online 31 May 2012
and transverse reinforcement ratio in the beams, columns and joints, design span length and span aspect
ratio. Extensive instrumentation was installed in predetermined locations on the tested specimens prior
Keywords:
to conducting the tests, recording data such as acceleration, velocity and displacement distribution. The
Dynamic performance
RC beam-column substructures
histories of the bending moments, as well as the vertical and horizontal reaction forces at the fixed sup-
Loss of a corner column ports were also noted. The test results indicated that the column removal apparatus was effective and
Progressive collapse confirmed that the design span length significantly affected the capability of the frames to resist progres-
sive collapse. Furthermore, it was demonstrated that seismically detailed specimens could mount a more
robust performance in resisting progressive collapse.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction A key issue in progressive collapse is to understand that it is a

Progressive collapse first attracted the attention of engineers
after the structural failure of a 22-storey building in the Ronan
Point apartment collapse in London in 1968. Progressive collapse
is defined as ‘‘the spread of an initial local failure from element
to element, eventually resulting in the collapse of an entire struc-
ture or a disproportionately large part of it’’ [1]. The structural
engineering community has long recognized the need to consider
means of mitigating progressive collapse. Several design guidelines
[2,3] have proposed design procedures to evaluate the likelihood of
progressive collapse of a structure following the notional removal
of vertical load bearing elements with the updated DoD [2] incor-
porating a number of improvements. The most significant
improvement is the application of the load increase factor (LIF)
and dynamic increase factor (DIF) respectively to relate the linear
static and nonlinear static procedures to nonlinear dynamic
procedure. However, this method requires the development of a
LIF and DIF data pool, which is reliant on numerical simulations
to provide the data upon which to develop the database. Thus, it
is imperative that more dynamic tests are needed to improve the
accuracy of the proposed DIF database, thereby allowing for a bet-
ter understanding of the dynamic performance of structures.
⇑ Corresponding author. Tel.: +65 67905090.
E-mail address: cbli@ntu.edu.sg (B. Li).
dynamic and nonlinear event. Many analytical studies have inves-
tigated the performance of structures under the scenarios of loss of
columns [4–11]. However, few studies have examined the dynamic
collapse performance of RC structures undergoing large deforma-
tion. Sasani et al. [12] conducted an in situ test of a reinforced con-
crete building with one-way floor slabs supported by transverse
frames. The dynamic performance of the building after the sudden
removal of an exterior ground bearing column was studied. Sasani
[13] and Sasani and Kropelnicki [14] conducted experimental and
analytical investigations of the dynamic response of a six-storey
reinforced concrete infilled frame structure following the simulta-
neous removal of two adjacent exterior columns (including one
corner column). For both transverse and longitudinal frames,
bi-directional Vierendeel action proved to be the main load redis-
tribution mechanism. Such in situ tests [12–15] provided urgently
needed information on the dynamic performance of the RC frame
after the removal of column elements. However, the peak vertical
displacement was relatively small in the tests (6.6 and 6.1 mm in
Sasani et al. [12] and Sasani [13] respectively), possibly due to
the absence of live or additional dead loads in both buildings prior
to the removal of the column due to safety considerations. This in
turn affected the applicability of the experimental results for vali-
dation of the finite element (FE) models. In addition, specially de-
signed full-scale in situ tests for progressive studies are extremely
costly and time-consuming, rendering systematic studies of the
in situ tests almost impossible. The experimental results of Sasani

0141-0296/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2012.04.016
 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 155

Nomenclature

D distance between the center of these two load cells _

uðtÞ velocity time series
(400 mm in this study) € ðtÞ
u recorded acceleration time series
DL dead load N the number of sampling points in the acceleration time
LL live load series
M calculated bending moment at the fixed support Ds the integral time step and is required to be sufficiently
R1 and R2 measured axial force histories of the two vertical load short to satisfy the condition of convergence
cells in the fixed support (as shown in Fig. 2)
u(t) displacement time series

[13] and Yi et al. [16] indicated that the upper and lower floors
operated in tandem as a unit, given that the dimensions and rein-
forcement details of both members were identical. Thus, the
behavior of multi-storey frames could be simplified to that of a sin-
gle-storey structure under the proper boundary conditions.
Six one-third scale substructures of varying detailing and
dimensions were designed and tested at NTU, Singapore to inves-
tigate the dynamic performance of three-dimensional (3D) beam-
column substructures following the removal of a ground corner
support. It should be emphasized that the results presented in this
paper are limited to experimental data based on an extensive
instrumentation plan. Moreover, the slab and in-fill wall effects
on the progressive collapse performance of the RC frame was not
included in this study although Kai and Li [17] has indicated that
RC slab could increase the resistant capacity up by 63% and Tsai
and Huang [18] has concluded that the in-fill walls could signifi-
cantly reduce the inelastic displacement.

2. Description of test program

2.1. Design of test setup

Fig. 1 illustrates the deformation of a nine-storey RC frame after

one of its corner columns was lost. The majority of the deformation
took place in the corner panels and the deformation of the remain-
Studied substructure ing panels was negligible. Therefore, one typical critical panel (the
corner panel on the second storey) was extracted and examined in
the current study. A schematic of the test setup is shown in Fig. 2.
Fig. 1. Deformation shape of prototype structures after one of corner column was The setup can be separated into three components. In Component
lost (no scale). 1, vertical, axial and rotational constraints were applied to the

1. Chain block

2. Upper steel column

1 1
3. Steel assembly

4. LVDT This jack was used in

5. Accelerometer another series of test
6. Axial force release mechanism
2
7. Concentrate weight assembly

8. Vertical double action load cell

3
9. Horizontal double action load cell Longitudinal direction Transverse direction
10. Vertical double action load cell
11 4 5
11. Specimen

12. Dynamic Strain Meter

9
13. Hi gh Speed Data Logger

7 R2 R1
10 7 8
6
D

12
13

Fig. 2. An overview of a specimen in position ready for testing.

156 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

corner column under extreme loading. In Component 3, a steel

assembly (Item 3 in Fig. 2) was designed to provide partial rota-
1: Hemisphere steel ball
tional and horizontal constraints in the corner joint, as found in
the in situ test of Sasani and Kropelnicki [14]. The detailing of
the steel assembly was shown in Fig. 4. One strong steel column
was connected to the corner stub of the RC specimen by merged
2: Special steel column
bolts in the corner stub. Four high strength and stiffness steel pins
in each direction were used to apply prescribed partial rotational
and horizontal constraints in the corner joint. Therefore, the steel
column could move freely in the vertical direction but the rota-
tional and horizontal freedoms were partially restrained. Further
details on the design of the steel assembly are provided in Kai
3: Pin support
and Li [17].

4: Load cell 2.2. Tested specimens

5: Nuts help to adjust The non-seismically detailed prototype RC moment resisting

frames were designed according to Singapore Standard CP 65
height of support
[19] while the seismically detailed prototype frame was designed
Fig. 3. The detailing of the designed axial force release mechanism.
according to ACI 318-08 [20]. It should be noted that the seismi-
cally designed prototype was assumed to be located on a D class
site (stiff soil profile) where the design spectral response accelera-
tion parameters, SDS and SD1, were 0.47 and 0.32, respectively. The
distributed dead load on the prototype structure due to a gravity
load of 210 mm thick slab was 5.1 kPa. The superimposed dead
Upper Steel Column load due to the ceiling weight was assumed to be 1.0 kPa. The
equivalent additional dead loads due to the weight of the in-fill
walls and beams were 2.25 kPa and 1.59 kPa, respectively. The live
Steel Pins load was assumed to be 2.0 kPa. The designed axial force in the cor-
ner column of each specimen was determined based on either the
full load combination as suggested in DoD [2] or the reduced load
combination, which was defined as the DoD [2]’s load combination
multiplied by a fraction factor. As shown in Table 1, Specimens
(V1)
(V2)

DF3, and DF4 were subjected to full service loads but wind and
snow loads were ignored. The load combination is 1.2DL + 0.5LL,
(V)

where DL is the dead load and LL is the live load. Specimens DF1
Steel Box and DF2 were subjected to a reduced load combination of
0.9  (1.2DL + 0.5LL). The calculation process of the design axial
Z
force in the corner support of each specimen is given in Appendix
A.
Corner Stub Fig. 5 illustrates the typical reinforcement layout of Specimens
0 DF1 and DF2. The concrete cover of the beam and column was
X 10 and 20 mm respectively. For DF2, the transverse reinforcements
were provided in the joint region using hoop stirrups with 135°
Fig. 4. The detailing of steel assembly.
bends. For the remaining specimens, non-seismic detailing was
provided while transverse reinforcements were provided using
adjacent columns to simulate the fixed boundary conditions as hoop stirrups with 90° bends. In addition, no transverse reinforce-
provided by the surrounding structural elements. In Component ment was installed in the joint region. It should be noted that dou-
2, a column removal apparatus comprising a specially designed bly continuous longitudinal rebar were installed in the beam as
steel column, a pin support and a load cell (Item 6 in Fig. 2, details scaled specimens were tested in the current study. The anchorage
shown in Fig. 3) was designed to simulate the sudden removal of a failure of the bottom longitudinal reinforcement at the beam end

Table 1
Specimen properties (unit: mm).

Specimen ID Elements Longitudinal rebar Transverse reinforcement Load case Measured axial force (kN)
Beam-T Beam-L Beam-T Beam-L Joint Beam-T Beam-L
Control specimen DF1 (F3) Type aa Type aa 4-T10 4-T10 None R6@180 R6@180 0.9 DoD[2] 16.9
Seismically detailed specimen DF2 (F2) Type aa Type aa 4-T13 4-T13 R6@55 R6@60 R6@60 0.9 DoD[2] 16.9
High service load specimen DF3 (F3) Type aa Type aa 4-T10 4-T10 None R6@180 R6@180 1.0 DoD[2] 18.7
Modified detailed specimen DF4 (F4) Type aa Type aa 4-T10 4-T10 None R6@80 R6@80 1.0 DoD[2] 18.8
Long span specimen DF5 (F5) Type bb Type bb 4-T10 4-T10 None R6@160 R6@160 0.8 DoD[2] 23.3
Unequal span specimen DF6 (F6) Type aa Type bb 4-T10 4-T10 None R6@180 R6@160 1.0 DoD[2] 23.2

Beam-L = Longitudinal beam Beam-T = Transverse beam; 0.9 DoD [2] = 0.9(1.2DL + 0.5LL)
a
Clear span = 2175 mm, cross-section = 180  100.
b
Clear span = 2775 mm, cross-section = 240  100.
 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 157

Table 2
Dynamic strain meter channel label for each specimen.
R6@180 R10@55 CH No. DF1 and DF2 DF3, DF4 and DF5 DF6
4T10 4T16 1 Reaction force D1 D1
Detail A-A Detail B-B 2 D1 BSTL1 BSTL1
3 D2 BSBL1 BSBL1
D4&A4

4 D3 BSTL8 BSTL8
5 D4 BSBL8 VR1-T
R6@60 6 BSTL1 VR1-T⁄ VR2-T
Bar anchorage 7 BSBL1 VR2-T⁄ HR1-T
D3&A3

4T13 detail 2 8 BSTL8 HR1-T⁄ VR1-L

Detail C-C 9 BSBL8 VR1-L VR2-L
10 BST4 VR2-L HR1-L
Note: R6=Plain round bar of 6 mm diameter
R10=Plain round bar of 10 mm diameter
D2&A2

T10=Deformed bar of 10 mm diameter Note: VR1-T⁄ and VR2-T⁄ were used to measure the vertical reaction of the trans-
T13=Deformed bar of 13 mm diameter verse support.
Bar anchorage T16=Deformed bar of 16 mm diameter HR1-T⁄ was utilized to measure the horizontal reaction in the transverse fixed
detail 1
support.
D1, D2, D3, and D4 were the LVDTs distributed along the transverse beam (see
D1&A1

R6@180 R6@250 R6@180

Fig. 5).
Center stub Fixed support BSTL1,BSBL1,BSTL8,BSBL8,and BST4 are strain gauges (see Figs. 16 and 17).
Plan View of Specimen DF1
R10@55

R6@180 R6@250 R6@180

A
Anchorage detail 2 channel distribution strategies were assigned for different speci-
B D
mens. Table 2 tabulates the channel distribution strategy of each
Anchorage detail 1 specimen. Finally, the corner support was suddenly knocked down
Elevation View of Specimen DF1 by a heavy hammer.
R10@55

R6@60 R6@125 R6@60

C
Anchorage detail 2
2.4. Material properties
B

Anchorage detail 1 The targeted compressive strength of concrete at 28 days was

R6@55 Elevation View of Specimen DF2 30 MPa. The average compressive strength of concrete fc0 obtained
Fig. 5. Dimensions and cross sectional details of DF1 and DF2 (in mm).
from the concrete cylinder samples was found to be 32.8, 31.4,
31.8, 33.1, 33.8, and 33.1 MPa for DF1, DF2, DF3, DF4, DF5, and
DF6, respectively. Grade 250 (R6) and Grade 460 (T16, T13 and
near to the corner joint (BENC), due to the changes in the directions T10) steel bars were used as the transverse and longitudinal
of bending moments, was mitigated in the current study by using reinforcements, respectively. Table 3 gives the measured tensile
hooked anchorages instead of the compression development properties of the bars used in the tests.
lengths. The development length of the hooked beam top longitu-
dinal reinforcement into the fixed support was also longer than the 2.5. Instrumentation
design development length according to the ACI 318-08 [20]. The
anchorage details are illustrated in Fig. 5. To monitor the response of the test specimens, extensive mea-
suring devices were installed both internally and externally. In the
2.3. Test procedures static load process, a total of 25 data channels were active in the
Static Data Logger. However, in the dynamic load process, only
Two chained blocks were hung the steel column (Item 2 in 10 data channels were active. Several accelerometers were con-
Fig. 2) to prevent the corner stub moving downward during the nected with the Digital High Speed Data Logger to monitor the
process of adding the simulated gravity weights. Six specific con- acceleration distribution along the beams following the initial
centrated weight assemblies (Item 7 in Fig. 2) were slowly applied damage. The sampling frequencies of the Dynamic Strain Meter
to the beams in a symmetrical sequence. It should be noted that and High Speed Data Logger were 2000 Hz and 1000 Hz, respec-
the weight of the concentrated weight assembly is designed tively. A load cell (Item 4 in Fig. 3) was positioned beneath the
according to the service load condition of each specimen (the de- pin support to monitor the increase in axial force in the corner col-
tailed design process can be found in Appendix A). The special steel umn during the static process and to record the axial force varia-
column (Item 2 in Fig. 3) was elevated and the height of the pin tion during the dynamic process. Four LVDTs were placed along
support (Item 3 in Fig. 3) was adjusted until the tip of the hemi- the beams to monitor the dynamic displacement distribution of
sphere steel ball made contact with the bottom surface of the cor-
ner stub. The chained blocks were then loosened to allow for slight
Table 3
adjustment of the height of the pin support until the reaction force Properties of reinforcing steel.
in the corner column reached the design axial force, as given in Ta-
Types Yield strength Yield strain ey Ultimate Ratio of
ble 1. A Static Data Logger was used to record the reinforcement
fy (MPa) (106) strength fu (MPa) elongation (%)
strain, axial load, and vertical reaction force during the static load
R6 530 2650 613 20.3
process. The crack development in the beams and corner joint was
R10 575 2895 695 21.7
also monitored during this static process. After all specific concen- R13 520 2595 637 22.6
trated weight assemblies have been applied, several selected R16 556 2897 635 21.1
cables initially connected with the Static Data Logger were
R6 = Plain round bar of 6 mm diameter.
switched onto the Dynamic Strain Meter by special bridge heads. T10 = Deformed bar of 10 mm diameter.
Only 10 cables could be connected with the Dynamic Strain Meter T13 = Deformed bar of 13 mm diameter.
due to channel limitations. In order to obtain more data, different T16 = Deformed of 16 mm diameter.
158 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

DF1 and DF2. For the remaining specimens, only one LVDT with asymmetrical failure modes in the longitudinal and transverse
300 mm travel was installed to measure the response of vertical beams of DF1 were observed. Severe diagonal shear cracks and
displacement in the corner column. Two compression/tension load slight concrete spalling occurred in the transverse beam end near
cells (Items 8 and 10 in Fig. 2) were installed vertically to measure to the fixed support (BENF) while narrow diagonal shear cracks oc-
the vertical reaction and to determine the moments on the fixed curred in the longitudinal BENF. Hairline flexural cracks were ob-
supports. One compression/tension load cell (Item 9 in Fig. 2) served in the BENC while symmetric hairline shear cracks were
was mounted horizontally to measure the horizontal reaction force observed in the corner joint. This indicated that the direction of
at the fixed support. In total, 12 strain gauges were mounted on the the bending moment in the BENC, which was initially negative
reinforcement at strategic locations before casting. However, only (tension at the top) under gravity loading, was changed after re-
the strain gauges listed in Table 2 were monitored during the dy- moval of the corner support.
namic test. The locations of the accelerometers, LVDTs and strain For DF2, only some flexural cracks were observed in the BENFs
gauges are illustrated in Fig. 5. (refer to Fig. 7b). Similar to DF1, the direction of the bending mo-
ment was changed in the BENC. However, no cracks occurred in
the corner joint of DF2. It should be emphasized that DF1 and
3. Test observations and results
DF2 were subjected to a reduced service load of 0.9(1.2DL + 0.5LL)
before removal of the corner support.
A total of six 3D beam-column substructures, each of different
DF3 had similar dimensions and reinforcement details as DF1.
design detailing and span length, were constructed and tested to
However, a full service load 1.0(1.2DL + 0.5LL) was applied. As pre-
evaluate the dynamic redistribution performance of the RC frame
sented in Fig. 7c, severe flexural and shear cracks were observed in
when subjected to the loss of a ground corner column. The test re-
the BENFs. Slight concrete crushing occurred in the bottom of the
sults of the six specimens are presented below.
BENF. More flexural cracks developed in the BENCs and more se-
vere diagonal shear cracks were observed in the corner joint of
3.1. Evaluation of the efficiency of the designed column removal DF3 compared to DF1. Although more severe damage took place
apparatus in DF3 due to the higher service load, both specimens managed
to survive the tests. DF4 had a higher transverse reinforcement ra-
The column removal apparatus designed in the current study tio in the potential plastic hinge zone of the beam compared to
involved impacting down by a heavy hammer to simulate the sud- DF3. In general, the crack patterns of DF4 were similar to those
den removal of the column by extreme loading. As the duration of of DF3. The crack width in the BENFs and corner joint was slightly
corner support removal will significantly affect the accuracy of the less than that in DF3. DF5 had a longer design span length than that
dynamic response of the substructures, the effectiveness of the de- of DF3. The dimensions and reinforcement details are given in
signed apparatus must be ensured. The history of the axial force in Table 1. DF5 was subjected to a reduced service load of
the corner column was monitored by a load cell (Item 4 in Fig. 3). 0.8(1.2DL + 0.5LL) before removal of the corner support. As dis-
As illustrated in Fig. 6, initially, the corner column of DF1 and DF2 played in Fig. 7e, DF5 suffered severe damage after removal of
was subjected to an axial force of 16.9 kN (negative represents the corner support. Extremely wide shear cracks and flexural
compressive force). The axial force started to release (compressive cracks were observed in the BENFs while significant spalling was
force begin to reduce) at 0.2 s. The column axial forces in DF1 and observed in the corner joint. The damage in the BENCs was negli-
DF2 were totally released by 0.2035 s and 0.2030 s, respectively. gible relative to that in the BENFs. It should be pointed out that
Thus, the duration of the force release was 0.0035 s and 0.0030 s the collapse of DF5 was stopped by the pin support of the column
for DF1 and DF2, respectively. This provided proof that the de- removal apparatus when the maximum displacement exceeded
signed apparatus satisfied the requirement of the guideline DoD the allowed displacement of about 360 mm. It can be predicted
[2], which requires the duration of the column removal to be less that in the absence of the pin support, DF5 would have collapsed.
than one-tenth of the natural period of the vertical motion of the DF6 had unequal spans in the longitudinal and transverse direc-
structure after losing the column (the measured nature period of tions. The dimensions and reinforcement details are given in Table
the vibration is about 0.15 s for DF1 and DF2). 1. DF6 was subjected to a full service load of 1.0(1.2DL + 0.5LL) be-
fore removal of the corner support. As shown in Fig. 7f, DF6 col-
3.2. Crack pattern and failure mode lapsed following the removal of the corner support. However,
asymmetrical damages were observed in the longitudinal and
The crack pattern and failure mode of each specimen after the transverse beams. Similar to DF5, extreme damage was concen-
dynamic tests are illustrated in Fig. 7. As presented in Fig. 7a, trated in the transverse BENF. However, the damages occurring
in the longitudinal BENF of DF6 were much milder. As illustrated
0
DF1 in Fig. 7f, severe cracks were observed in the corner joint; these
Axial force in the corner column (kN)

DF2 cracks were not symmetrical due to the beams being of unequal
-3
spans.

-6
3.3. Global displacements

-9
In order to estimate the global displacement responses of the
tested specimens, several LVDTs were placed on the specimens.
-12
Fig. 5 graphically depicts the locations of the LVDTs. One LVDT with
300 mm travel was placed in the center of the corner joint for all
-15
specimens. Three additional LVDTs were placed at the quarter,
-16.90 kN
middle and three-quarter spans of the transverse beams of DF1
-18
0.190 0.195 0.200 0.205 0.210 0.215 0.220 0.225 0.230
and DF2 to monitor the displacement distribution along the beams.
Fig. 8 shows the vertical displacement history of the corner joint of
Time (s)
each specimen. Negative values represent downward displace-
Fig. 6. Recorded reaction force history in the corner column during test. ments. The column was removed at time 0.2 s for all specimens.
 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 159

Specimen DF1 Specimen DF4

(a) (d)
Specimen DF2 Specimen DF5

(b) (e)
Specimen DF3 Specimen DF6

(c) (f)
Fig. 7. Cracking patterns and failure mode of each specimen.

For DF1, the first peak vertical displacement was 83.9 mm at
0.38 s. The downward displacement was reduced to 79.3 mm
by 0.48 s but subsequently rose back to 82.6 mm at 0.55 s. The
oscillation was almost eliminated by 1.5 s and the substructure
had a permanent downward displacement of 81.3 mm, which
was 96.9% of the first peak displacement. The first peak displace-
ments measured in DF2, DF3 and DF4 were 29.0, 146.6 and
99.4 mm, respectively. It should be noted that the peak displace-
ments of DF5 and DF6 exceeded 300 mm, which was the mea-
surement capacity of the LVDT placed in the corner joint. In
50
Vertical Displacement (mm)
addition, vertical motions of both specimens were stopped by the
pin support of the apparatus and the measured distance from the
bottom face of the corner column to the pin support was about
360 mm.
3.4. Displacement distribution
Fig. 9 illustrates the vertical displacement distribution along the
transverse beam of DF1. As can be seen from the figure, the maxi-
mum displacements at D1, D2, D3 and D4 of DF1 were 83.9,
73.4, 46.8 and 20.0 mm, respectively. This indicated that the
distribution of the deformation was non-linear and the deforma-
tion of the beam occurred in a manner different from that in a can-
tilever beam due to the partial rotational constraint applied to the

0
corner joint. As mentioned in the above section, the column was
-50
removed at a time of 0.2 s. However, the movement at the third-
quarter of the beam span only commenced at 0.235 s. From this
-100 it can be reasoned that the dynamic force took at least 0.035 s to
propagate to that point. DF2 had a similar displacement distribu-
-150 DF1
DF2 tion response as DF1; the maximum displacements at D1, D2, D3,
DF3 and D4 of DF2 were 29.0, 26.0, 17.0 and 4.9 mm, respec-
-200 DF4
DF5 tively. A similar initial time lag of 0.037 s at D4 was also observed.
DF6
-250
0.0 0.5 1.0 1.5
3.5. Acceleration response
Time (s)

Fig. 8. Recorded history of the vertical displacement of the corner joint of each In progressive collapse analysis, it is prudent to understand that
specimen. it is a dynamic event, and that the sudden removal of an element in
160 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

10 to be comparable to that of DF1, save for point A1. While vibrations

DF1-D1
0 DF1-D2 for other points occurred along the zero axes, A1 in DF2 displayed
Vertical Displacement (mm)

-10 DF1-D3 vibration along an inclined line. It indicated that large noise existed
DF1-D4
-20 in A1 of DF2. DF3 and DF4 displayed similar behavior to the first
-30 two specimens. However, no reliable data was recorded for DF5
-40 and DF6 as their collapse was stopped by the pin support of the
-50 apparatus, bringing about large random noises in the acceleration
-60 readings.
-70
-80 3.6. Integral velocity response
-90
0.0 0.5 1.0 1.5 No measured velocity response could be presented as no veloc-
Time (s) ity sensors were installed in the specimens. However, by assuming
a zero initial condition, it is possible to acquire the velocity re-
Fig. 9. Recorded vertical displacement distribution of DF1.
sponse by directly conducting an integral transform based on the
measured acceleration data:
a structure causes an immediate geometric change, resulting in the
Z t

release of potential energy and rapid alteration of internal static _

uðtÞ ¼ € ðsÞds
u ð1Þ
0
dynamic forces, including inertia forces. Hence, it is important to
record the acceleration history at varying locations on the speci- Z t

mens. As mentioned earlier, several accelerometers were placed uðtÞ ¼ _ sÞds

uð ð2Þ
0
in designated locations (refer to Fig. 5) prior to the removal of
the corner column. Considering the symmetry of DF1 to DF5, only Eqs. (1) and (2) can be conveniently expressed in the discrete form
the acceleration history in the transverse beam of these specimens as:
was monitored. However, for DF6, both beams were monitored.
X
N
1
Fig. 10 presents the recorded acceleration histories of DF1. The _
uðtÞ ¼ ðu € ðiÞÞDs
€ ði  1Þ þ u ð3Þ
maximum acceleration at A1, A2, A3 and A4 were 2.91 g, i¼0
2
1.89 g, 1.90 g and 0.43 g, respectively. Thus, the contribution
of the mass along the beams to inertia force was reduced signifi- X
N
1
cantly moving away from the corner stub. Therefore, it was diffi- uðtÞ ¼ _  1Þ þ uðiÞÞ
ðuði _ Ds ð4Þ
i¼0
2
cult to accurately evaluate the effects of the inertia force just
based on the existing acceleration recordings which were only ta- However, the direct integration of the acceleration data often
ken at key points along the beam. It is recommended that acceler- causes unrealistic drifts in the velocity and displacement, as pre-
ometers be installed more extensively along the specimens in sented in Fig. 12a, where the integral velocity response based on
future tests to better quantify the inertial force. All of the acceler- the recorded acceleration data of A1 in DF1 is given as an example.
ation diminished rapidly after the peak, and was negligible by The integral velocity without any baseline correction has significant
0.003 s. As shown in the history of A1, A1 started to go in the po- drifts with periodic oscillations. The drifts in velocities and
sitive phase at a time of 0.293 s. The vibration was along the zero displacements can be attributed to many potential factors. For
value after a time of 0.515 s. The remaining points (A2, A3 and A4) example, it is generally agreed that the mechanical or electrical
exhibited similar behavior as A1. For DF2, the observations proved hysteresis in the sensor can cause an offset occurring in the

2.0 2.0

1.0 (a) 1.0

(b)
Acceleration (g)
Acceleration (g)

0.0 0.0

-1.0 A1 -1.0 A2

-2.0 -2.0

-3.0 -3.0

-4.0 -4.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Time (s) Time (s)

2.0 2.0

1.0 (c) 1.0 (d)

Acceleration (g)
Acceleration (g)

0.0 0.0

-1.0 -1.0 A4
A3
-2.0 -2.0

-3.0 -3.0

-4.0 -4.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (s) Time (s)

Fig. 10. Recorded history of the acceleration at specified locations of DF1.

 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 161

10 10

Vertical Displacement (mm)

Vertical Displacement (mm)

0
-10
(a) 0 (b)
-10
-20 -20
-30 -30
-40 -40
-50 -50
DF1-Measured-D2
-60 DF1-Measured-D1 -60 DF1-Integral-D2
-70 DF1-Integral-D1 -70
-80 -80
-90 -90
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Time (s) Time (s)

10 10

Vertical Displacement (mm)

Vertical Displacement (mm)

0
-10
(c) 0 (d)
-10
-20 -20
-30 DF1-Measured-D3 -30 DF1-Measured-D4
DF1-Integral-D3 DF1-Integral-D4
-40 -40
-50 -50
-60 -60
-70 -70
-80 -80
-90 -90
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (s) Time (s)

Fig. 11. Comparison the measured displacement history with the corresponding integral displacement history of DF1.

acceleration records. Even a small offset in acceleration can produce and V4 in DF1 were 0.61, 0.60, 0.32 and 0.14 m/s, respec-
significant drifts in velocities and displacements. Another major tively. Similar to the displacement and acceleration response dis-
source of unrealistic drift in velocities and displacements may come tribution along the transverse beam, the velocity response
from the accumulation of the random noise in accelerations result- distribution was also non-linear. The velocities of V1, V2, V3, and
ing from the integrations. Thus, for the graph tail, polynomial curve V4 were reduced to zero at 0.425 s, 0.406 s, 0.437 s and 0.439 s,
fitting was performed and the graph was subsequently corrected to respectively. The velocity responses of the remaining specimens
the zero-velocity baseline as shown in Fig. 12a. were similar to DF1. It should be noted that the acceleration, veloc-
In order to evaluate the accuracy of this integral method to ob- ity and displacement response of the full-scale specimens may be
tain the velocity response, the displacement responses obtained by up to three times of the reported figures as only one-third scale
integral transform from the recorded acceleration data were com- specimens were tested.
pared with the measured displacement histories. As presented in
Fig. 11, the integral displacement matched well with the measured 3.7. Vertical reaction force response
displacement, although the integral displacement had a slight time
lag. Thus, the accuracy of the proposed integral method and the Fig. 13 illustrates the recorded history of the total vertical reac-
reliability of the velocity response obtained from integral trans- tion forces of DF3 and DF4. As displayed in Fig. 13, the initial total
form can be ascertained. Fig. 12 illustrates the velocity responses vertical reaction force measured from the longitudinal and trans-
at prescribed locations of DF1, the peak velocities of V1, V2, V3 verse fixed supports of DF3 was 20.4 kN. This vertical reaction

0.4 0.4
Intergal V1 after
(a) (b)
Vertical velocity (m/s)

baseline correction 0.2

Vertical velocity (m/s)

0.2
0.0 0.0

-0.2 -0.2 Integral V2

Intergal V1 before
-0.4 baseline correction -0.4

-0.6 -0.6

-0.8 -0.8

-1.0 -1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Time (s) Time (s)

0.4 0.4
(c) (d)
Vertical velocity (m/s)

0.2
Vertical velocity (m/s)

0.2
0.0 0.0
-0.2 Integral V3 -0.2 Integral V4
-0.4 -0.4
-0.6 -0.6

-0.8 -0.8
-1.0 -1.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Time (s) Time (s)

Fig. 12. Integral history of the velocity at specified locations of DF1.

162 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

0.0 enhancement, the dynamic increase factor for the released column
DF3
axial forces in DF3 and DF4 were 1.29 and 1.37 respectively. Fur-
DF4
-10.0
ther discussions for the dynamic effects are shown below.
Total Reaction Force (kN)

Initial Values
-20.4 kN for DF3
-20.0 -20.7 kN for DF4 3.8. Horizontal reaction force response

Fig. 14 presents the histories of the horizontal reaction forces

-30.0
-40.0
-50.0
0.0 0.5 1.0 1.5
Time (s)
Fig. 13. Recoded history of the total vertical reaction force of DF3 and DF4.
force suddenly increased to 44.5 kN at 0.29 s. After 0.70 s, the to-
tal reaction force vibrated along the line with the value of
39.1 kN. Similar behavior was observed for DF4. As listed in Table
1, the axial forces in the corner columns of DF3 and DF4 before
they were lost were 18.7 and 18.8 kN, respectively. However,
the maximum increased reaction forces in the fixed supports of
DF3 and DF4 were 24.1 and 25.8 kN respectively. In other
words, the maximum increased reaction force was significantly
larger than the initial axial force in the corner support. This can
be attributed to the inertia force which developed after the sudden
removal of the corner support. The measured inertia force for DF3
and DF4 was 5.4 kN and 7.0 kN, respectively. Moreover, if it is
focused on the dynamic effects for the total reaction force
measured in the fixed supports of the transverse beams of DF3
and DF4. For DF3, 4.1 kN tensile horizontal reaction force was mea-
sured before the removal of the corner support. After sudden re-
moval of the corner support, the horizontal reaction force
plunged to its peak value of 20.4 kN at 0.37 s. The permanent
2.0 2.5 3.0 compressive reaction force (14.7 kN) was measured after vibra-
tion. Similar behavior was recorded in DF4. Thus, it can be con-
cluded that no horizontal tensile force was measured during the
dynamic process and thus, no catenary action was developed to re-
sist progressive collapse for DF3 and DF4 during the test. It is pos-
sibly as the deformation of both specimens is insufficient to
develop catenary action in resisting the progressive collapse. Alter-
natively, the corner column could not provide enough horizontal
constraint to develop the catenary action in the beams.
3.9. Bending moment response
The histories of the bending moment at the longitudinal and
transverse fixed supports of DF3 and DF4 were presented in
Fig. 15. As mentioned in the instrumentation section, the bending
moment was calculated based on the axial force histories of the
two vertical tension/compression load cells in each fixed support
and is expressed as following equation:

M ¼ ðR1  R2 Þ  0:5D ð5Þ

10.0
Initial Values DF3
For DF3, as can be seen from the Fig. 15a, the initial bending mo-
4.1 kN for DF3 DF4 ments at the longitudinal and transverse supports were 2.66 and
Horizontal Reaction Force (kN)

4.6 kN for DF4 2.45 kN m, respectively. The bending moment in the longitudinal
0.0
and transverse fixed supports suddenly increased to 20.8 and
20.1 kN m, respectively after removing the corner support. After
free vibration, the permanent moment was 16.6 and 16.3 kN m in
-10.0
the longitudinal and transverse fixed supports, respectively. It can
be seen that the bending moment suddenly increased by 682.0%
and 720.5% respectively in the longitudinal and transverse fixed
-20.0
supports after the removal of the corner support. For DF4, the initial
bending moments at the longitudinal and transverse supports were
2.79 kN m and 2.51 kN m, respectively. The bending moments in
-30.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 the longitudinal and transverse fixed supports suddenly rose to
Time (s) 20.1 kN m and 19.5 kN m, respectively after removing the corner
support. After free vibration, the permanent moments recorded in
Fig. 14. Recoded history of the horizontal reaction force in the transverse beam of the longitudinal and transverse fixed supports were 14.0 kN m
DF3 and DF4. and 14.8 kN m, respectively. It can be seen that the bending

25 25
DF3 DF4
Bending Moment (kN.m)

Bending Moment (kN.m)

20 20

15 15

10 10
Initial Values Initial Values
2.45 kN.m in Trans. Beam 2.51 kN.m in Trans. Beam
5 2.66 kN.m in Long. Beam 5 2.79 kN.m in Long. Beam
Longitudinal Beam Longitudinal Beam
0 Transverse beam 0 Transverse Beam
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (s) Time (s)

(a) (b)
Fig. 15. Calculated histories of the bending moments in the fixed supports of DF3 and DF4.
 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 163

moment suddenly rose by 620.0% and 676.9% respectively in the 2000

longitudinal and transverse fixed supports, respectively. 1800
BST4 in DF1
1600
3.10. Local behavior-rebar strains 1400
BST4 in DF2

Strain (με)
1200
In general, the trends of the strain gauge results in each speci-
1000
men were similar and thus, only the results of DF1 are explained
800 Initial Values
in detail in this paper. As shown in Fig. 16, the strain in the top 430 με in BST4 of DF1
of the BENC (BSTL1) was 360le before the removal of the corner 600
390 με in BST4 of DF2
support. However, this tensile strain decreased and ultimately 400 BSTL1 BSTL8

reached the maximum compressive strain of 1001le at the time 200

BST1
BSBL1
BST4
BSBL8
of 0.41 s. On the other hand, the strain of BSBL1 initially had com- 0
pressive strain 163le. After removal of the corner support, the 0.0 0.5 1.0 1.5 2.0 2.5 3.0

compressive strain in BSBL1 suddenly decreased to zero and ulti- Time (s)
mately attained the maximum tensile strain of 2488le at a time
of 0.42 s. Both results verified that the direction of bending mo- Fig. 17. Strain gauge recording of the transverse reinforcement in DF1 and DF2.

ment in the BENC was changed after the removal of the corner sup-
port. It can also be seen that the strain reading significantly
of DF1. Comparing the peak vertical displacements of these two
decreased after 0.5 s. The permanent strain in BSTL1 and BSBL1
specimens (Fig. 8), it can be seen that the peak vertical displace-
at the end of vibration was 723 and 1731le, respectively. The ini-
ment of DF2 was only 34.6% that of DF1. Similar trends were ob-
tial strain readings of BSTL8 and BSBL8 were 578 and 205le,
served for the acceleration and velocity results.
respectively. After removing the corner support, the readings of
BSTL8 and BSBL8 increased to 4487 and 1817le at the times of
4.2. The effect of different service load conditions
0.41 and 0.42 s, respectively. This demonstrated that the bending
moment increased considerably in the BENF (consistent with the
As listed in Table 1, DF3 was subjected to a full service load
bending moment results). For other specimens, similar behavior
1.0(1.2DL + 0.5LL) while DF1 was subjected to a reduced service
was observed. Fig. 17 illustrates the strain gauge reading in the
load 0.9(1.2DL + 0.5 LL) before the removal of the corner support.
transverse reinforcements of DF1 and DF2. As presented in the fig-
Comparing the crack patterns of DF3 with DF1 (refer to Fig. 7),
ure, the strain in BST4 was initially 430le and it increased to
more severe cracks were observed in the corner joint and longitu-
1807le by 0.42 s. The permanent tensile strain in BST4 at the
dinal BENF in DF3. However, the shear cracks occurring in the
end of the vibration was 1425le. It should be emphasized that
transverse BENF of DF1 were more severe than those of DF3. This
the yield strain of steel rebar R6 was 2650le. Thus, the transverse
was an unexpected phenomenon and was possibly due to inherent
rebar did not yield during the test. However, as shown in the final
drawbacks existing in DF1, such like a honeycomb structure devel-
crack patterns of DF1 (refer to Fig. 7a), severe diagonal cracks oc-
oped in the BENF of DF1 due to insufficient vibration during cast-
curred in the BENF. Similar behavior was observed in DF2. How-
ing. It can be seen in Fig. 8 that DF3 increased the peak vertical
ever, the peak strain of BST4 was only 1451le, much less than
displacement by 74.7%.
the BST4 in DF1, possibly due to the higher transverse reinforce-
ment ratio in the beam potential plastic hinge zone.
4.3. The effect of modified detailing

4. Discussion of test results
4.1. The effect of seismic detailing
The non-seismically detailed specimen DF1 had severe flexural
and shear cracks in the BENFs while only limited flexural cracks in
the BENFs were observed in the seismically detailed specimen DF2.
Moreover, no cracks were observed in the corner joint of DF2 while
several hairline diagonal cracks were observed in the corner joint
DF4 had a much higher transverse reinforcement ratio in the
potential plastic hinge zone of beams compared to that of DF3.
Similar to DF3, DF4 was subjected to a full service load
1.0(1.2DL + 0.5LL) before removing the corner support. Comparing
the crack pattern of DF4 to that of DF3 (refer to Fig. 7), it can be
seen that milder cracks were observed in the BENFs and corner
joint in DF4 although, in general, similar crack patterns were ob-
served in both specimens. As seen in Fig. 8, DF4 sustained a de-
crease in the peak vertical displacement by 32.2%.

4.4. The effect of design span length

6000
Initial Values
360 με in BSTL1, -163 με in BSBL1 BSTL8
5000 BSTL1
578 με in BSTL8, -205 με in BSTL8 BST1 BST4
As illustrated in Table 1, DF5 had 2775 mm clear span and was
4000 BSBL1 BSBL8 subjected to a reduced service load of 0.8(1.2DL + 0.5LL). As pre-
3000 sented in Fig. 7e, DF5 failed with extensive damage in the BENFs
and corner joint. The peak vertical displacement exceeded
Strain (με)

BSTL8
2000
360 mm. In fact, without the pin support of the column removal
1000
BSBL1 apparatus or should there be sufficient space to allow the gravity
0 BSTL1 weights to drop freely, the specimen would be expected to fail
completely.
-1000

-2000
BSBL8 4.5. The effect of span aspect ratio
-3000
0.0 0.5 1.0 1.5 2.0 2.5 3.0
DF6 has different span lengths in the longitudinal and trans-
Time (s)
verse directions, and it was subjected to the reduced service load
Fig. 16. Strain gauge recording of the flexural reinforcement in DF1. of 0.9(1.2DL + 0.5LL). It can be seen from Fig. 7f that extensive
164 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

damage occurred in the transverse BENF. Although severe cracking 40

also occurred in the longitudinal BENF, it was much milder com- ML F MT
35
pared to the transverse beam. Compared to DF3, DF6 collapsed
D

Load on substructures (kN)

due to the longer span in the longitudinal beam, which resulted 30
in a larger axial force in the corner column of the substructure be- F2 (DF2)
F3 (DF1, DF3)
fore it was lost (refer to Table 1). The axial force, previously re- 25
F4 (DF4)
F5 (DF5)
sisted by the corner column, tried to redistribute into both fixed F6 (DF6)
20
supports after the sudden removal of the corner support. However,
almost half of the axial force will redistribute into the transverse 15
fixed support due to the similar stiffness in the longitudinal and
transverse beams for both DF3 and DF6. Therefore, more axial force 10

was distributed into the transverse fixed support in DF6 compared 5

to DF3. This resulted in the total collapse of the transverse beam of
DF6 while less damage was observed in DF3.
5. Discussion of dynamic effects
Two series (F and DF) of tests were conducted at NTU, Singa-
pore. F and DF represent RC frames were subjected to pushover
and dynamic load regimes, respectively. As shown in Table 1, the
dimension and reinforcement details of DF2, DF4, DF5, and DF6
are similar as that of F2, F4, F5, and F6, respectively. In addition,
the dimension and reinforcement details of DF1 and DF3 are sim-
ilar to F3. The details of the test results of the F-series specimens
are given in Kai [21] and Kai and Li [22].
5.1. Pushover performance of the test specimens
It should be noted that the pushover performances of Speci-
mens DF2, DF3, DF4, DF5 and DF6 were assumed same as Speci-
mens F2, F3, F4, F5, and F6, respectively, although they are cast
in two different batches. The setup and the load–displacement
curve of Specimens F2, F3, F4, F5, and F6 are shown in Figs. 18
and 19, respectively. As shown in Fig. 18, the setup for F-series
specimens is similar to the DF-series specimens. The adjacent col-
umns were assumed as an equivalent fixed-support. A steel assem-
bly was utilized to apply partial rotational constraint in the corner
joint. The only differences are no column removal apparatus was
designed and the load was applied on the corner stub by a hydrau-
lic jack with 600 mm stroke.
5.2. Dynamic load increase factor (DLIF) of each specimen
As shown in Fig. 19, the static load capacity of F2, F3, F4, F5, and
F6 are 36.5 kN, 25.8 kN, 27.5 kN, 26.8 kN, and 26.0 kN, respectively.
However, the axial force in the corner support of DF1, DF2, DF3,
DF4, DF5, and DF6 before removal of the corner column was
16.9 kN, 16.9 kN, 18.7 kN, 18.8 kN, 23.3 kN and 23.2 kN, respec-
tively. Based on the dynamic test results, it was found that DF1,
DF2, DF3, and DF4 are survived after removal of the corner support.
0
0 100 200 300 400 500
Vertical displacement (mm)
Fig. 19. Load–displacement curves of the F-series specimens.
Thus, it can be concluded that the DLIF of F2, F3, and F4 are less
than 2.16, 1.38, and 1.46, respectively. It should be noted that dy-
namic load increase factor was defined as the ratio of the static
load capacity to its dynamic load capacity. Conversely, DF5 and
DF6 are totally collapse after removal of the corner support. Thus,
it was found that the DLIF of Specimens F5 and F6 are larger than
1.15 and 1.12, respectively. Due to the exact value of the dynamic
load capacity for each specimen cannot be obtained by a single dy-
namic test, the exact value of the DLIF for each specimen could not
be specified. Therefore, the validated numerical and analytical
analyses were needed for accurate predicting the dynamic load
capacity of each specimen in the future.
5.3. Evaluate the DLIF suggested in DoD [2]
As illustrated in Fig. 19, the load–displacement behavior of each
specimen belongs to Type 2 (force controlled) in DoD [2]. It should
be noted that the DIF defined in DoD [2] and Marchand et al. [23] is
actually similar to the DLIF defined in this study. They recom-
mended the value of DIF equal to ‘2.0’ for the structural compo-
nents with force controlled behavior. Thus, the experimental
results indicated that the value of DIF suggested in DoD [2] for
structural components with force controlled behavior is over-con-
servative and further studies were needed to refine the design
guidelines.
6. Summary and conclusions
Considering the lack of related experimental tests for validation
of the finite element studies and the extremely costly and time-
consuming nature of in situ tests, an equivalent beam-column

1: Load cell measure applied load 1

2: Hydraulic jack with 600 mm stroke

3: Steel column
2
4: Comp/tension load cell

5: Steel assembly
3
6: LVDT with 300 mm travel

7: RC substructure 5 4
8 and 9: LVDTs 6
Longitudinal beam
9 8
10 and 11: Comp/tension load cell
7
11
Transverse beam 10

Fig. 18. An overview of a specimen in position ready for testing.

 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 165

substructure with suitable boundary conditions was tested. The Appendix A

following conclusions can be drawn.
1. The column removal apparatus proved to be effective in
dynamic tests as the measured release time did not exceed
0.0035 s, which was less than 10% of the natural period of the
substructures. Peak acceleration measured could reach up to
3.5 g in DF3 but the peak acceleration was decreasing with
moving away from the corner column.
2. The peak value of the total vertical reaction force measured in
the fixed supports was larger than the axial force of the corner
column before its subsequent removal. This was due to the iner-
tia force which was developed after the sudden removal of the
corner support.
3. No tensile force was observed in the horizontal reaction force in
the fixed supports of DF3 and DF4. This indicated that catenary
action did not develop to resist the progressive collapse during
the tests of DF3 and DF4.
4. The test results confirmed that seismically design and detailing
succeeded in increasing the resistant capacity of the structures
against progressive collapse. Structures with longer design span
lengths exhibited a higher vulnerability for progressive collapse
compared to structures with shorter span lengths. Furthermore,
more attention should be paid to structures subjected to a
higher service load.
5. Comparing the dynamic and pushover performance of the test
substructures, it was found that DLIFs in the specimens could
less than 1.38, which is significantly less than the value sug-
gested in the design guidelines for structures components with
force controlled behavior. Thus, it was proved that the DIF value
suggested in design guideline for structural components with
force controlled behavior is over-conservative.
6. The effects of the slab are not included in this current study.
Previous research studies have concluded that the presence of
a RC slab could significantly increase the capacity of the RC
frame in resisting progressive collapse and considerably reduce
the likelihood of progressive collapse. The slab effects on the
dynamic performance of the RC frame for progressive collapse
should be further evaluated in the future.
Acknowledgments
This research was supported by a research Grant provided by
the Defense Science & Technology Agency (DSTA), Singapore, under
the Protective Technology Research Center, Nanyang Technological
University, Singapore, Any opinions, findings and conclusions ex-
pressed in this paper are those of the writers and do not necessarily
reflect the view of DSTA, Singapore.
This appendix presents the calculation process to determine the
design axial force in the corner support under service condition. As
one-third scale test models were design and tested in this study,
the major steps of calculation process were shown in below:
Step 1: Determine the bending moment demand of the beam in
the prototype frame.
Step 2: Determine the ratio of the bending moment demand to
the yield capacity of the beam section in the prototype frame.
Step 3: Determine the bending moment demand in the model
frame based on the equivalent strain method.
Step 4: Determine the required pressure applied on the model
frame, which can produce required bending moment demand.
Step 5: Determine the axial force in the corner support of each
model frame.
The detailed configuration of the prototype corresponding mod-
el frames is given in Table A1.
Step 1: Determine the bending moment demand of the beam of
prototype frame.
The dead load due to slab self-weight is 0.21 m  2450 kg/
m3  9.81 N/kg = 5.05 kpa. The live load is taken to be 2.0 kPa at
each storey level while the dead load consists of the self-weight
of the building and 1.0 kPa additional dead load applied to each
floor. The 1.0 kPa additional dead load accounts for items such as
partitions, ceiling, mechanical ductwork, electrical items, plumb-
ing, and so forth.
The dead load due to in-fill walls (two exterior walls and two
half interior walls) is:
3ð0:3 m  5:5 kN=m3  7:2 m  3:3 m=ð7:2 mÞ2 Þ
¼ 2:25 kN=m2 ðA:1Þ
It should be clarified that the floor height is 3.3 m and the density of
hollow brick is assumed equal to 550 kg/m3.
The dead load due to RC beams (two exterior beams and two
half interior beams) is:
3ð0:3 m  23:5 kN=m3  7:2 m  0:54 m=ð7:2 mÞ2 Þ
¼ 1:59 kN=m2 ðA:2Þ
According to DoD [2], the design pressure at service condition is:

Table A1
Co-relationship between the prototype frames with corresponding test models in the current study (unit: mm).

Test Prototype Beam (mm) Prototype Beam Rebar Model Beam (mm) Model Beam Rebar
Trans. Beam Long. Beam Trans. Beam Long. Beam Beam-T Beam-L Trans. Beam Long. Beam
Top Bottom Top Bottom
DF1 (F3) 540  300 540  300 2T20 + T32 2T20 + T32 2T20 + T32 2T20 + T32 180  100 180  100 4T10 4T10
DF2 (F2) 540  300 540  300 3T32 3T32 3T32 3T32 180  100 180  100 4T13 4T13
DF3 (F3) 540  300 540  300 2T20 + T32 2T20 + T32 2T20 + T32 2T20 + T32 180  100 180  100 4T10 4T10
DF4 (F4) 540  300 540  300 2T20 + T32 2T20 + T32 2T20 + T32 2T20 + T32 180  100 180  100 4T10 4T10
DF5 (F5) 720  300 720  300 2T25 + T32 2T25 + T32 2T25 + T32 2T25 + T32 240  100 240  100 4T10 4T10
DF6 (F6) 540  300 720  300 2T20 + T32 2T20 + T32 2T25 + T32 2T25 + T32 180  100 240  100 4T10 4T10

Beam-L = Longitudinal beam.

Beam-T = Transverse beam.
T32 = Deformed bar of 32 mm diameter, T25 = Deformed bar of 25 mm diameter, T20 = Deformed bar of 20 mm diameter.
T16 = Deformed bar of 16 mm diameter, T13 = Deformed bar of 13 mm diameter, T10 = Deformed bar of 10 mm diameter.
166 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167

qp ¼ 1:2DL þ 0:5LL Myp ¼ 260 kN m ðA:7Þ

¼ 1:2ð5:05 þ 1:0 þ 2:25 þ 1:59Þ þ 0:5  2 ¼ 12:9 kPa ðA:3Þ
MRp 125:4
¼ ¼ 0:48 ðA:8Þ
As shown in Figs. A.1–A.3, the equivalent uniform line pressure Myp 260
could be determined.
Step 3: Determine the bending moment demand of the model
1 frame based on the same ratio of the bending moment demand
Pp ¼ qp l2 ¼ 0:5  12:9  7:2 m ¼ 46:44 kN=m ðA:4Þ
2 to the yield capacity as the prototype frame.
Mym ¼ 9:67 kN m ðA:9Þ
5
PEp ¼ Pp ¼ 29:02 kN=m ðA:5Þ
8 MRm ¼ 0:48  9:67 kN m ¼ 4:64 kN m ðA:10Þ

1 Step 4: Determine the line pressure of the model beam which

M Rp ¼ PEp  L2 ¼ 125:4 kN m ðA:6Þ can produce the required bending moment demand.
12
12  M Rm
where qp is the uniform pressure on the prototype slab; Pp is the PEm ¼ 2
¼ 9:67 kN=m ðA:11Þ
l
maximum value of triangle line pressure on the prototype beam;
PEp is the equivalent uniform line pressure on the prototype beam; 8
MRp is the bending moment demand on the prototype beam; Myp is Pm ¼ P Em ¼ 15:5 kN=m ðA:12Þ
5
the yield bending moment on the prototype beam.
Step 2: Determine the ratio of the bending moment demand to 2P m
the yield capacity of the beam section in the prototype frame.
qm ¼ ¼ 12:9 kPa ðA:13Þ
l
where Pm is the maximum value of the triangle pressure on the
model beam; PEm is the equivalent uniform line pressure on the
model beam; qm is the uniform pressure on the model slab; MRm
is the bending moment demand on the model beam; and Mym is
the yield bending moment on the model beam.
It can be found that the ratio of the beam bending moment
demand to its yield moment capacity of the model specimen is
same as that in the prototype frame as long as the total uniform
pressure applied on the slab of the model specimen is same as
its corresponding prototype slab.
Actually, this conclusion can also be drawn by stiffness equiva-
lent method.
Assuming one model beam is 1/3-scale, the length, depth and
the width of the beam are 1/3 of that in the prototype beam.
 3
1
Mym ¼ M yp ðA:14Þ
3

The yield moment of model beam is 1/27 of that of prototype beam.

 3
3 l
The demand of bending moment of model beam is kqm lm ¼ kqm 3p .
In order to obtain same ratio of the bending moment demand to its
yield capacity, only ensure qm = qp.
Fig. A.1. Slabload distribution to beam of prototype slab.
Step 5: Determine the design axial force in the corner support of
the tested specimens
A.1. For DF3 and DF4

The design pressure

qm ¼ 1:2DL þ 0:5L ¼ 12:9 kN=m2 ðA:15Þ

The design axial force is:
74:3 kN
¼ 18:6 kN ðA:16Þ
4
Fig. A.2. Triangle line pressure on beam of prototype slab.
A.2. For DF1 and DF2

The design pressure

qm ¼ 0:9  ð1:2DL þ 0:5LÞ ¼ 11:6 kN=m2 ðA:17Þ

The design axial force is:

66:8 kN
¼ 16:7 kN ðA:18Þ
Fig. A.3. Equivalent uniform line pressure on beam of prototype slab. 4
 Q. Kai, B. Li / Engineering Structures 42 (2012) 154–167 167

A.3. For DF5 [6] Tsai MH, Lin BH. Investigation of progressive collapse resistance and inelastic
response for an earthquake-resistant RC building subjected to column failure.
Eng Struct 2008;30(12):3619–28.
The design pressure [7] Kwasniewski L. Nonlinear dynamic simulations of progressive collapse for a
multistory building. Eng Struct 2010;32(5):1223–35.
qm ¼ 0:8  ð1:2DL þ 0:5LÞ ¼ 10:3 kN=m2 ðA:19Þ [8] Izzuddin BA, Vlassis AG, Elghazouli AY, Nethercot DA. Progressive collapse of
multi-storey buildings due to sudden column loss—Part I: simplified
assessment framework. Eng Struct 2008;30(5):1308–18.
The total gravity load of the slab: [9] Vlassis AG, Izzuddin BA, Elghazouli AY, Nethercot DA. Progressive collapse of
multi-storey buildings due to sudden column loss—Part II: application. Eng
10:3  3:0  3:0 ¼ 92:9 kN ðA:20Þ Struct 2008;30(5):1424–38.
[10] Mohamed OA. Assessment of progressive collapse potential in corner panels of
The design axial force is: reinforced concrete buildings. Eng Struct 2009;31(3):749–57.
[11] Fu F. 3-D nonlinear dynamic progressive collapse analysis of multi-storey steel
92:9 kN composite frame buildings—Parametric study. Eng Struct
¼ 23:2 kN ðA:21Þ 2010;32(12):3974–80.
4 [12] Sasani M, Bazan M, Sagirolu S. Experimental and analytical progressive
collapse evaluation of an actual reinforced concrete structure. ACI Struct J
2007;104(6):731–9.
A.4. For DF6 [13] Sasani M. Response of a reinforced concrete infilled-frame structure to
removal of two adjacent columns. Eng Struct 2008;30(9):2478–91.
[14] Sasani M, Kropelnicki J. Progressive collapse analysis of an RC structure. Struct
qm ¼ 1:2DL þ 0:5L ¼ 12:9 kN=m2 ðA:22Þ Des Tall Special Build 2008;17:757–71.
[15] Sasani M, Sagiroglu S. Gravity load redistribution and progressive collapse
The total gravity load of the slab: resistance of 20 story reinforced concrete structure following loss of interior
column. ACI Struct J 2010;107(6):636–44.
12:9  3:0  2:4 ¼ 92:9 kN ðA:23Þ [16] Yi WJ, He QF, Xiao Y, Kunnath SK. Experimental study on progressive collapse-
resistant behavior of reinforced concrete frame structures. ACI Struct J
The design axial force is: 2008;105(4):433–9.
[17] Kai Q, Li B. Slab effects on the response of reinforced concrete substructures
92:9 kN after the loss of a corner column. ACI Struct J, in press.
¼ 23:2 kN ðA:24Þ [18] Tsai MH, Huang TC. Numerical investigation on the progressive collapse
4
resistance of an RC building with brick infills under column loss. World Acad
Sci Eng Technol 2011;58:946–53.
[19] CP 65. Structural use of concrete, Part 1 Code of practice for design and
References construction. Singapore Standard; 1999.
[20] ACI Committee 318. Building code requirements for structural concrete (ACI
[1] ASCE/SEI 7. Minimum design loads for buildings and other structures. Reston 318-08) and commentary (318R-08). American Concrete Institute, Farmington
(VA): Structural Engineering Institute, American Society of Civil Engineers; Hills, MI; 2008. p. 456.
2010. [21] Kai Q. Experimental and analytical study of reinforced concrete substructures
[2] DOD. Design of building to resist progressive collapse. Unified Facility Criteria, subjected to a loss of ground corner column scenario. PhD dissertation,
UFC 4-023-03. Washington (DC): US Department of Defense; 2009. Nanyang Technological University, Singapore; 2011. p. 300.
[3] GSA. Progressive collapse analysis and design guidelines for new federal office [22] Kai Q, Li B. Performance of three-dimensional reinforced concrete beam-
buildings and major modernization projects. Washington (DC): US General column substructures under loss of a corner column scenario. J Struct Eng,
Service Administration; 2003. submitted for publication.
[4] Marjanishvili SM. Progressive analysis procedure for progressive collapse. J [23] Marchand K, Mckay A. Stevens DJ. Development and application of linear and
Perform Construct Facil ASCE 2004;18(2):79–85. non-linear static approaches in UFC 4-023-03 structures 2009: Don’t Mess
[5] Marjanishvili SM, Agnew E. Comparison of various procedures for progressive with Structural Engineers@2009. ASCE; 2009. p. 1729–38.
collapse analysis. J Perform Construct Facil 2006;20(4):365–74.