Static and Dynamic Responses of Reinforced Concrete

Structures under Sudden Column Removal Scenario

Subjected to Distributed Loading
Anh Tuan Pham 1 and Kang Hai Tan 2

Abstract: In this paper, static and dynamic experiments on reinforced concrete beam-column frames under single-column-removal scenario
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

applying a multipoint loading method were conducted. One of the objectives was to investigate structural behavior compared with the single-
point loading method, which has been popularly used in previous studies. By equally applying point loads at four locations of the double-span
beam structure, the test setup successfully simulated the uniformly distributed loading condition, which is generally applicable for gravity
loads in practice. Compared with the previously conducted static tests based on concentrated loading method, structural response from the
static tests under multipoint loading condition differed not only on load-bearing capacity but also on failure sequence and displacement
profile. On the basis of the test results, the analytical relationship between behaviors of the two loading methods was developed and verified.
Compared with the static tests, the dynamic tests highlighted dynamic effects created by the column loss event and confirmed the structural
behavior and failure modes observed in the static environment. The dynamic tests also verified the correctness and conservatism of the
dynamic assessment framework using a previously proposed energy-based approach. Most important, both the static and the dynamic tests
of structures under distributed loads showed less development of catenary action against progressive collapse compared with concentrated
loading tests. DOI: 10.1061/(ASCE)ST.1943-541X.0002214. © 2018 American Society of Civil Engineers.
Author keywords: Reinforced concrete; Progressive collapse; Catenary action; Compressive arch action; Concentrated loading; Uniformly
distributed loading.

Introduction blast-induced tests on existing multistory RC buildings (Sasani

In a missing-column scenario, two different conditions of gravity
loading can be considered for the affected structure in a typical
floor above the removed column: (1) loads from above floors would
transfer to below floors through the column (above the missing col-
umn) in the form of concentrated loading (CL); and (2) every story
above the missing column behaves identically under gravity action,
and the loads transferred through the column are negligible. As a
result, the main action is uniformly distributed loading (UDL) in
the single story. Concerning structural response of planar RC beam-
column structures against progressive collapse under quasi-static
conditions, most of the experimental and analytical studies con-
ducted (Bao et al. 2008; Sasani and Kropelnicki 2008; Yi et al.
2008; Su et al. 2009; Choi and Kim 2011; Sadek et al. 2011;
Stinger and Orton 2013; Valipour et al. 2013; Yu and Tan
2013a, b; Bao et al. 2014; Lew et al. 2014; Ren et al. 2016; Kang
and Tan 2016; Lu et al. 2017; Yu and Tan 2017) applied CL con-
figuration in which the middle joint of the beam-column structure
received a push-down displacement. In these studies, the point load
from the column above was considered the main action causing
the collapse. However, Sasani and colleagues conducted three
1
Researcher, Vietnam Institute for Building Science and Technology,
81 Tran Cung St., Cau Giay District, Hanoi, Vietnam. Email: anhtuanpham
.vn@gmail.com
2
Professor, School of Civil and Environmental Engineering, Nanyang
Technological Univ., 50 Nanyang Ave., Singapore 639798, Singapore
(corresponding author). Email: ckhtan@ntu.edu.sg
Note. This manuscript was submitted on November 21, 2016; approved
on June 5, 2018; published online on November 9, 2018. Discussion period
open until April 9, 2019; separate discussions must be submitted for indi-
vidual papers. This paper is part of the Journal of Structural Engineering,
© ASCE, ISSN 0733-9445.
et al. 2007; Sasani 2008; Sasani and Sagiroglu 2010) and observed
the opposite trend of mechanism. In those tests, concrete strain
gauges were installed on the columns above the removed ground
column. All three studies showed a consistent phenomenon of rapid
increase of tensile strains in the measured columns within a very
short time, much faster than the flexure response (moving down-
ward) of the structure. That is, shortly after the column loss event,
axial compressive forces in the above columns dropped signifi-
cantly. In reality, depending on structural configuration, gravity ac-
tion causing the double-span structure to move downwards can be a
combination of both CL (from the columns in the above floors) and
UDL (from the floor of the double-span structure that is being con-
sidered). Nonetheless, there is no study investigating the effect of
UDL condition on two-dimensional (2D) beam-column structures
so far, although UDL has already been considered when studying
three-dimensional (3D) beam-slab systems under progressive col-
lapse (Pham and Tan 2013, 2015; Qian et al. 2016).
The important question is whether catenary action (CA) in 2D
frames can be fully mobilized under the UDL case. In terms of fail-
ure modes, experimental studies on RC subassemblages under the
CL case showed fracture of bottom rebars at the middle joint re-
gion, followed by a sudden drop of structural capacity (Yu and Tan
2013a, b). Subsequently, as the deformation was kept increasing,
structural resistance increased again under CA mobilization, and a
complete collapse occurred when beam top rebars fractured at the
supports. Nevertheless, the change of loading method from CL to
UDL may change this sequence of failure and the deformation
shape of the double-span beam.
Although dynamic tests are closer to the nature of progressive
collapse than static tests, very few were conducted mainly due to
complexity and safety concerns of the test setup. Moreover,
dynamic tests are force controlled; that is, each test can only be

© ASCE 04018235-1 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 conducted with a predefined value of applied load and can only
provide one corresponding displacement profile of the frame and
beam. Despite laboratory constraints, some research works have
been conducted on free-fall dynamic tests to compare with corre-
sponding static specimens that had the same structural design and
boundary condition (Tian and Su 2011; Qian and Li 2012;
Sagiroglu 2012; Orton and Kirby 2014). However, such studies
often used different loading methods, that is, CL displacement-
controlled method for the static tests but multipoint loading for
the dynamic tests. As a result, it was difficult to directly compare
the structural responses of the two different loading arrangements
although the specimens had the same design and boundary
conditions.
To study the resistance of RC beam-column structures against
a single-column-removal scenario under both types of loading
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.
configurations, that is, UDL and CL, an experimental program was
conducted at Nanyang Technological University. The first part
of the program, which investigated dynamic structural responses
under CL condition, is reported in Pham and Tan (2017). The
second part, which included two test series under UDL condition,
is presented in this paper. The first series included two specimens
under quasi-static condition, whereas the second series comprised
three specimens under free-fall dynamic condition. Total loads ap-
plied on the specimens were equally distributed into four points
along the double-span beam. To have a direct comparison with the
CL condition, the specimen design and boundary condition of the
tests conducted in this study, as well as those in Pham and Tan
(2017), followed the corresponding 2D static test series conducted
by Lim et al. (2015). Whereas Pham and Tan (2017) confirmed the
structural response and failures observed in quasi-static condition
under CL configuration, the UDL static experiments conducted in
this paper showed differences in structural behavior between the
two loading methods (CL and UDL), focusing on the degree of
mobilization of CA. Thereafter, the UDL dynamic test series veri-
fied the responses and failure modes observed in the static tests
and elucidated the dynamic effects between the two loading rates
(quasi-static and free-fall dynamic). Finally, the results from the
dynamic tests were compared with predictions using the Izzuddin
method (Izzuddin et al. 2008) to verify this approach.
Experimental Program
Specimen Design and Test Setup
Because experimental work on progressive collapse requires com-
binations of several structural members and their boundary condi-
tions, a scaled model for testing is normally applied in most
experimental studies. In the current research, a scale factor of 1=3
was used for the frame specimens. Because the expected domi-
nant failure modes involved flexure and CA, size effect would
not have a significant effect on structural behavior. Although the
contribution of floor slabs is important to resist progressive col-
lapse, especially through the mobilization of tensile membrane ac-
tion under large deformations, previous experimental studies on 3D
beam-slab systems had encountered a difficulty in differentiating
the contribution of CA (in beams) and tensile membrane action
(in slabs). Therefore, a study on 2D frame structures (conducted
in the current research) is important to quantify the enhancement
of CA in beams before any 3D beam-slab structural test is car-
ried out.
The specimen design of the static and the dynamic tests con-
ducted in this study consisted of a double-span beam with a cross
section of 100 × 180 mm2 and a center-to-center single span of
© ASCE 04018235-2
J. Struct. Eng., 2019, 145(1): 04018235
2.4 m. Thus, the double-span beam was 4.8 m. Side columns
with a dimension of 180 × 180 mm2 and beam extensions were
provided at two ends of the double-span beam to simulate a sub-
structure from an RC multistory building, as shown in Fig. 1. Three
deformed bars T10 with a diameter of 10 mm were arranged as top
reinforcement at each beam end. They were reduced to two T10 at
curtailment points 650 mm away from the beam-column interfaces.
Only two T10 bars were used for beam bottom rebars and were
continuous along the double-span beam. The respective lengths
of side columns and beam extensions were chosen at contra-
flexural points, that is, half-length for columns and 0.21 span for
beam extensions, so that only pin and horizontal supports were
needed at such locations. Mild steel bars of 6-mm diameter (R6)
were used as stirrups in the double-span beam and side columns.
To avoid shear failure, stirrup spacing near the joint interfaces was
reduced from 120 to 50 mm.
To simulate distributed loads acting on the structure, a four-
point loading method was used in both the static and the dynamic
tests. Distances between the four applied loads were chosen to be
equal along the double-span beam. An elastic analysis comparing
deformation shape and moment distribution of a structure subject to
UDL and four-point loading was carried out (Fig. 2). It showed that
the equivalent loading system produced an almost similar displace-
ment profile to that by UDL, whereby both methods were in rea-
sonable agreement in terms of bending moments at the beam ends
and the middle joints, as well as locations of contra-flexural point
along the double-span beam.
For the quasi-static tests, a steel-beam load-transfer system was
fabricated as shown in Fig. 3(a) to equally divide the load from an
actuator above to four roller points on the specimen. Such a setup
ensured that the steel beams could translate and rotate without
affecting the deformation shape of the double-span beam. In the
dynamic tests, steel plates with fixed equal weights were hung to
the specimens at four locations in accordance with the static tests,
as shown in Fig. 3(b). Two types of boundary conditions, that is,
full-restraint and partial-restraint, were applied for both test series.
For the full-restraint specimens, beam extensions were arranged at
two sides of the double-span beam and were horizontally restrained
by either an A-frame or a strong wall to simulate the loss of an
interior column in which horizontal restraints from adjacent struc-
tures were appropriately provided at both sides. Fig. 3(a) illustrates
the test setup for a typical static full-restraint specimen with a sym-
metric arrangement of load pins and horizontal load cells. With this
instrumentation, vertical and horizontal forces from the supporting
system were well captured. On the other hand, for the partial-
restraint specimens, the beam extension was only designed for the
left side [Fig. 3(b)]. As a result, the right-side column was only
horizontally supported at its top and bottom ends. This type of
boundary condition represented the removal of a next-to-outermost
column in which the horizontal restraint from one side is remark-
ably weaker than the other side. To simulate axial forces on the side
columns arising from gravity loads from the above floors in a build-
ing, preloaded compressive forces were applied onto these columns
before testing through a hydraulic jack placed on top. The value of
this column axial load, which was similar to the tests by Lim et al.
(2015), was 200 kN. In the free-fall dynamic tests, the middle joint
was suspended from a reaction frame by a quick-release device
[Item (2) from Fig. 3(b)], which could be suddenly yanked free
to release the connection. This type of release mechanism has been
successfully applied in a previous progressive collapse study by Liu
et al. (2016).
For the UDL static test series, two tests named as FR-U (full
restraint) and PR-U (partial restraint) were conducted. They corre-
sponded with two CL static tests of full restraint (FR) and partial
J. Struct. Eng.
 PL 15x180x180 PL 15x180x180
PVC pipes

80
Four 30 mm dia.
FULL-RESTRAINT SPECIMEN
R6@50 7
4T10 6

100
4R6 6a 3
760

650 650
90 60
2 PL 15x180x180
PL 15x100x180

40
3T10 1 2T10 3T10 5 5 PL 15

125
2 1 1 1 3
1560
180

180

100
125
R6@50 5 4
R6@50 7

2T10

40
1 2 3 8T13
Right-side of
4 4 8 partial-restraint specimen
620

4 PVC pipes Ø30

330 360 840 360 330

R6@50 5 R6@90 5 R6@120 5 R6@90 5 R6@50 5

300
PL 15x180x180 PL 15x180x180
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

400 180 2220 90 90 2220 180 400

2400 2400
6 4T10 4R6 6a 7 R6@50 R6@50 9
2 1 3T10 1 2T10 3 1 3T10 8 3T13
7 R6@50

180
180

180

180

180
5 R6@90 5 R6@120 5 R6@50 8 2T13
7a R6@100
4 2T10 4 2T10 4 2T10 8 3T13
180 180
100 100 100

Fig. 1. Specimen design.

(a) (b)

Fig. 2. Comparison between UDL and four-point loading methods: (a) uniformly distributed load; and (b) four-point load.

restraint (PR) from Lim et al. (2015). For the UDL dynamic test
series, three specimens, named as FDU1 to FDU3, were cast
and tested. FDU1 and FDU3 represented partial-restraint frames
and were tested under two applied load levels of 42 and 61 kN.
On the other hand, FDU2 was designed as a full-restraint specimen
and was tested with a total load of 55 kN. FDU1 was tested twice
with two different loads, whereas FDU2 and FDU3 were only
tested once. After the first release of FDU1 with an applied load
of 42 kN (FDU1-P/42), the second release (FDU1-P/61) caused
a complete collapse to this specimen. Hence, FDU3-P/61 was con-
ducted with the same boundary condition (partial restraint) and ap-
plied load (61 kN) as FDU1-P/61 to observe the dynamic behavior
again under pristine condition of the specimen. Table 1 summarizes
the parameters of both the static and the dynamic tests, as well as
information on the two static tests conducted by Lim et al. (2015).
In terms of concrete grade, the tests presented in this study had
higher cylinder strengths compared with tests from Lim et al.
(2015). Regarding reinforcement, both the current (Table 2) and
the previous studies had similar material properties.
Instrumentation
The instrumentation setup of the UDL static and the UDL dynamic
tests were similar to that of the previous CL static tests in Lim et al.
(2015) [and in Pham and Tan (2017)]. Reaction forces were mea-
sured by using load pins and horizontal load cells, as shown in
Fig. 3. Total horizontal reactions from one side of the specimen
were calculated by summing all the horizontal forces acting on that

© ASCE 04018235-3 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Test setup: (a) static test (full-restraint specimen); and (b) free-fall dynamic test (partial-restraint specimen).

Table 1. Specimen properties

Specimen Concrete cylinder strength (MPa) Restraint condition Dynamic test Total dynamic load (kN)
FR 32.0 Full restraint Static by Lim et al. (2015)
PR 32.0 Partial restraint Static by Lim et al. (2015)
FR-U 46.6 Full restraint Static test
PR-U 45.0 Partial restraint Static test
FDU1a 44.5 Partial restraint FDU1-P/42 42
Partial restraint FDU1-P/61 61
FDU2 44.5 Full restraint FDU2-F/55 55
FDU3 51.0 Partial restraint FDU3-P/61 61
a
Specimen FDU1 was tested twice with two different applied loads.

Table 2. Material properties for the static and the dynamic tests
Diameter Yield strength, Elastic modulus, Yield strain, Tensile strength, Ultimate strain,
Rebar (mm) fy (MPa) Es (GPa) εy (%) f u (MPa) εu (%)
R6 6 352 220 0.162 539 19.2
T10 10 554 196 0.288 653 11.5
T13 13 535 188 0.285 615 8.5

© ASCE 04018235-4 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 Static test Dynamic test

351

A1

250
650 920 650

750
Lh4 L1
180 300

180 300
L2 L3
Lh2 L5

Lh1 L4
350

350
Lv5 Lv4 Lv3 Lv2 Lv1 90 90 300 350 300 310 310 350 300

880
Lh3
450 200 200 520 200 200 450 L1 to L5: displacement transducers
440

A1: accelerometer
markers for slow-motion camera
Lv1 to Lv5: vertical displacement transducers
Lh1 to Lh4: horizontal displacement transducers
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

(a)
Static test Dynamic test
middle middle Hs1 Hs2
joint joint
end end
joint joint
To Ti T1 T1a T2 Tm T3 T4a T4 T4 T3 Tm T2 T1 Ti To

Bo Bi B1 B1a B2 Bm B3 B4a B4 B4 B3 Bm B2 B1 Bi Bo
Section m

Section m
Section 1

Section 2

Section 3

Section 4

Section 4

Section 3

Section 2

Section 1
180 470 460 460 470 180 650 460 460 650

(b)

Fig. 4. Setup of instrumentation: (a) displacement measurement; and (b) strain measurement.

side. Each test showed that the total horizontal forces from the larger and significantly enhanced the capacity, the structure
left and right sides were similar even for the partial-restraint cases. reached the first peak at a MJD of about 0.4dbeam .
Displacement transducers were arranged along the double-span • Stage II: After attaining the peak of CAA, concrete crushing
beam and at the end-joint regions to measure vertical and horizontal started occurring at compressive zones of the end joints, causing
deformations of the structure [Fig. 4(a)]. Strain gauges were placed structural capacity to gradually reduce. This stage finished at a
along the beam longitudinal bars at five critical sections [Fig. 4(b)]: MJD of one dbeam when structural response started increasing
end-joint interface (Section 1), rebar curtailment point near the again (for both FR-U and PR-U), indicating a transformation
end joint (Section 2), rebar curtailment point near the middle joint from CAA to CA. In both stages, there were little discernible
(Section 3), middle-joint interface (Section 4), and mid-single-span differences between full- and partial-restraint specimens.
section (Section m). Dynamic data acquisitions at a sampling rate • Stage III: In this stage, horizontal reaction changed from com-
of 1,000 Hz were used in the dynamic tests. To measure the release pression to tension, and the structural capacity kept increasing
time of the device, two strain gauges (HS1 and HS2) were attached under the mobilization of CA. The first failure mode took place
to the hooks on top of the middle joint. when the first top rebar from one end joint fractured, hereafter
called “first top rebar fracture,” causing a sudden drop in struc-
tural capacity. This failure occurred earlier in PR-U (at 275-mm
Static Test Results of FR-U and PR-U MJD) than in FR-U (318 mm). From this point on, structural
responses of FR-U and PR-U differed significantly. Under in-
creasing displacement, the load-carrying capacity of FR-U in-
Overall Behavior creased but that of PR-U decreased until all the top rebars at
Total applied loads and total horizontal reactions measured from the end joint from one side fractured completely, hereafter called
the two UDL static tests are presented in Fig. 5 together with those “one-side top rebar fracture.” The increase of MJD after this
from the relevant CL tests of FR and PR (Lim et al. 2015) for com- point led to no further noticeable increase of structural capacity.
parison. Focusing on the static tests FR-U and PR-U, the general All the critical points of response from the UDL static tests are
frame behavior subjected to UDL can be summarized into three listed in Table 3, together with the relevant values from the CL
stages: static tests. It was clearly observed that vertical applied loads of
• Stage I: From the beginning of loading when the structure was the UDL tests under both boundary conditions were similar to
elastic under flexure until it gained its first peak due to compres- each other until rebar fracture took place, and were slightly more
sive arch action (CAA). In this stage, the structure was about to than twice those of the CL tests [Fig. 5(a)]. In addition, MJDs
attain yield at a middle-joint displacement (MJD) of 0.25dbeam , corresponding to different stages of response (yielding point,
marked by the yielding of tensile rebars at both Sections 1 and 4 end of Stage I, and end of Stage II) were comparable between
[Fig. 4(b)]. When the compression force in the beam became the two loading methods, as shown in Fig. 5(a) and Table 3.

© ASCE 04018235-5 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. General responses of static tests: (a) vertical applied load; and (b) total horizontal reaction.

Table 3. Summary of static test results

Bottom rebar One-side top rebar
Yielding point Maximum CAA fracture First top rebar fracture fracture Horizontal force (kN)
Beam-end
MJD Applied MJD Applied MJD Applied MJD Applied rotation MJD Applied Maximum Maximum
Test (mm) load (kN) (mm) load (kN) (mm) load (kN) (mm) load (kN) (degree) (mm) load (kN) compression tension
FR-U 44 53.9 74 56.45 N/A N/A 318 77.9 9.1 368.0 100.3 −28.3 81.9
PR-U 42 55.4 76 59.20 N/A N/A 275 67.8 9.1 286.3 44.6 −25.8 49.5
FR 44 24.7 70 27.40 230 26.4 549 70.9 8.6 a a
−28.5 148.8
PR 56 25.4 80 26.90 329 33.9 N/A N/A N/A N/A N/A −20.0 54.3
Note: N/A = top/bottom rebar fracture did not occur in the test.
a
All top rebars fractured together.

However, bottom rebar fracture at the middle joint, which occurred
in the CL static tests of Lim et al. (2015) and in all reported CL tests
in the literature, did not occur in the UDL static tests. Hence, it can
be considered as the key difference between the two loading sce-
narios. In terms of horizontal reaction, the UDL and the CL static
tests had relatively similar values [Fig. 5(b)], and maximum com-
pression forces during the CAA stage between the two loading
methods agreed well with each other (Table 3). However, the maxi-
mum MJD from FR-U was noticeably smaller than that from FR.
As a result, the maximum tension force of the horizontal reaction in
FR-U was 45% smaller than that in FR. Regarding partial-restraint
specimens, the MJD and the horizontal reaction of PR-U at the
maximum capacity was 16% and 9% smaller than those of PR,
respectively. However, after attaining the stage of bottom rebar
fracture in the PR test, the partially restrained side column started
moving inward under CA. Because the structural capacity had no
sign of increase, the test was stopped to prevent excessive column
lateral movement. As a consequence of this early termination,
the maximum applied load and horizontal reaction of PR were ap-
proximately 48% and 37% of the corresponding values from FR.
In contrast, the difference between PR-U and FR-U was less (68%
and 60% for applied load and horizontal reaction, respectively).
Damage Patterns and Failure Modes
Fig. 6 describes the final damage patterns and failure modes of the
UDL static tests together with the corresponding CL static tests.
During Stages I and II, it was observed that top-surface flexural
cracks at the end joint (Section 1) did not propagate significantly
toward the curtailment points (Sections 2 and 3) in the UDL tests.
In the CL tests, this top-surface crack propagation was obvious.
However, bottom-surface cracks from the CL tests only occurred
near the middle joint (Section 4), much fewer than those at the
middle joint of the UDL tests. To further study the development
of tensile zones in the structure in the flexure/CAA stage, compres-
sion and tension zones of FR-U and FR were plotted according to
strain gauge readings from beam longitudinal bars (Fig. 7). It was
assumed that, at this stage, the beam sections still remained plane
and strain distribution along each section was linear. During the
elastic range (0.1dbeam ), tension zones from the two loading cases
were relatively similar. As MJD increased to 0.25 and 0.4dbeam , the
tension zone near the end joint in FR quickly expanded toward the
middle joint. However, in FR-U, the trend was opposite, with
the tension zone near the middle joint quickly propagating toward
the end joint. This difference in the development of tensile region
accounted for the difference of crack distribution between the two
loading methods during flexure/CAA.
At a beam section with a distance Li from the side-column cent-
roid, bending moment M i was calculated according to recorded
reaction forces and corresponding vertical deformation δ i at this
section, as shown in Fig. 8(a). As a result, M i at the end-joint inter-
face and at the locations of displacement transducers [Lv1 to Lv5
in Fig. 4(a)] was calculated to obtain the bending moment distri-
bution along the beam. Fig. 8(b) shows this distribution at an MJD
of 0.25dbeam (yielding state) for both FR-U and FR tests. It agrees
well with the development of concrete tension zone presented in

© ASCE 04018235-6 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 1
2
m
4 3
(a)

1
2
m
4 3
(b)
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

1
2
m
3

4
(c)

1
2
m
3
4
(d)

Fig. 6. Final failure modes of static tests: (a) FR-U; (b) PR-U; (c) FR; and (d) PR.

(a) (b)

Fig. 7. Tensile zone development in static tests: (a) FR-U (UDL); and (b) FR (CL).

Fig. 7. The bending moment distribution in FR-U followed a para-
bolic shape, whereas that of FR was linear with a steep curvature
at midspan. Compared with FR, FR-U shows a more uniform
distribution of bending moment at the middle-joint region. This
explains why flexural cracks were more equally spaced out in
FR-U [Fig. 6(a)].
After the MJD had exceeded one beam depth, the horizontal
reaction gradually changed to tension, and there were only few
new cracks occurring in the UDL tests. Instead, opening of existing
cracks was more dominant. When the top rebars started fracturing
at one end joint, bottom-surface cracks near the middle joint propa-
gated toward Section 2. This crack propagation can be explained by

© ASCE 04018235-7 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 LH3 LH3

LH2 LH2

LH1 LH1

LV LV
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

(a)
Section m

Section m
Section 2

Section 3

Section 4

Section 4

Section 3

Section 2
Section 1

Section 1
450 400 520 400 450 450 400 520 400 450

Lv5 Lv4 Lv3 Lv2 Lv1 Lv1 Lv2 Lv3 Lv4 Lv5

middle
end end
joint joint joint

(b)

middle
end end
joint joint joint

(c)

Fig. 8. Moment distribution of the specimens under CL and UDL conditions: (a) bending moment at typical sections of RC frames; (b) moment
distribution at MJD ¼ 0.25dbeam (kN m); and (c) moment distribution at MJD ¼ 2dbeam (kN m).

the moment distribution of the beam at an MJD of 2dbeam , that is,
at a large deformation stage [Fig. 8(c)]. After the first top rebar
fracture of FR-U at 318 mm (1.8dbeam ), bending moment at the end
joint reduced, shifting contra-flexural point toward the end joint,
which caused propagation of bottom-surface cracks. On the other
hand, for FR, the bottom rebars had completely fractured at this
stage of deformation causing nearly zero bending moment at the
middle joint and negative bending moment along the entire beam.
As a result, top-surface cracks extensively developed in FR during
Stage III. There was no major failure (concrete crushing or rebar
fracturing) at the middle-joint regions of the UDL tests. On the
other hand, in the CL tests (FR and PR), top-surface cracks propa-
gated throughout curtailment regions after the bottom rebars had
fractured. Full-depth tension cracks also occurred in these tests,
illustrating that some beam sections were already under pure
tension. In contrast, there were no full-depth cracks for FR-U and
PR-U, indicating lower development of tensile force. Regarding
displacement profiles, whereas the CL tests showed a rigid-link
type of deformation, the UDL tests exhibited the usual parabolic
shape.
When examining the two UDL static tests [Figs. 6(a and b)],
there were fewer flexural cracks at the end joint in PR-U compared
with FR-U, suggesting greater strain concentration at the beam-
column interface of PR-U. Therefore, the top rebars started fractur-
ing earlier in PR-U at 275 mm (1.5dbeam ) than in FR-U at 318 mm
(1.8dbeam ). Moreover, because the collapse of PR-U came from the
fracture of top rebars at one end joint, its partially restrained column
underwent little inward movement. This inward column movement
was more significant in PR.
Structural responses of UDL and CL static tests were also com-
pared at the cross-sectional level by studying the top and bottom
reinforcement strains along the double-span beam (Fig. 9). In FR,
top rebars in curtailment points (T2, Tm, and T3) yielded quickly
after the fracture of bottom rebars at an MJD of 230 mm [Fig. 9(a)],

© ASCE 04018235-8 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 MJD (mm) MJD (mm)
5000 5000
4000
4000 yield strain

Strain of reinforcement (µ)

Strain of reinforcement (µ)

3000
yield strain 2000
3000
1000
2000 0
T1
T2 -1000 0 100 200 300 400 500 600
1000 Tm
T3 -2000 B1
T4 B2
-3000 yield strain Bm
0
0 100 200 300 400 500 600 -4000 B3
B4
-1000 -5000
(a) (b)
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

MJD (mm) Deflection (mm)

5000 5000
T1 4000
T2 yield strain
Strain of reinforcement (µ)

4000

Strain of reinforcement (µ)

T3 3000
yield strain T4 2000
3000
1000
2000 0
-1000 0 100 200 300 400
1000 -2000
B1
B2
-3000 Bm
0 yield strain B3
0 100 200 300 400 -4000
B4
-1000 -5000
(c) (d)

Fig. 9. Comparison of rebar strains between CL and UDL static tests: (a) top rebars of FR; (b) bottom rebars of FR; (c) top rebars of FR-U; and
(d) bottom rebars of FR-U.

accounting for the occurrence of top-surface cracks at this region PR-U, and FR. Because top rebars did not fracture in PR, the
[Fig. 6(c)]. On the other hand, in FR-U, these rebars remained elas- rotation result of this test was not considered. It is shown that the
tic until the end of the test [Fig. 9(c)]. As a result, negative-moment beam-end rotations at the first top rebar fracture, denoted as θmax ,
plastic hinges did not form at the curtailment points in the UDL were relatively similar among all the three tests [approximately 9°
tests [Figs. 6(a and b)]. Thus, curtailment of top rebars at the middle (Table 3)], although the failure occurred at different MJDs when
zone of the single span (Fig. 1), which had a significant effect comparing the UDL and the CL tests. The horizontal reactions of
on deflected shape under CL condition, had almost no effect the two loading cases at this state are described in Fig. 11 and de-
under the UDL case. Concerning the beam bottom reinforcement pend on the resultant forces from top (Fatop ) and bottom (Fabot )
[Figs. 9(b and d)] in FR and FR-U, the development of rebar tensile rebars at Section 1 of Fig. 4(b). From Fig. 9(b and d), Fabot (rep-
strain at Section 3 (B3) during Stages I and II was limited in FR resented by Strain gauge B1) was in tension in FR and in compres-
than in FR-U. It agreed well with the bending moment in Fig. 8(a), sion in FR-U at this state. As a result, horizontal reactions of
which shows a more uniform distribution of negative moment near FR (H CL ) and FR-U (HUDL ) are calculated using Eqs. (1) and (2),
the middle joint in FR-U. Before the top rebar fracture in FR (MJD respectively, as follows:
of 549 mm), the bottom rebar at Section 1 (B1) already changed
from compression to tension, showing that the entire beam section H CL ¼ cosðθmax ÞðFatop þ Fabot Þ ð1Þ
at the end joint was in tension at this stage. However, in FR-U, B1
was still in compression until the end of the test (MJD of 368 mm),
indicating that tensile force was not yet fully developed throughout HUDL ¼ cosðθmax ÞðFatop − Fabot − Fc Þ ð2Þ
the end-joint section.
where Fc = resultant force of concrete region in compression at
beam bottom surface.
Beam-End Rotation and Horizontal Reaction at Top
At the state of top rebar fracture, it is assumed that Fatop had
Rebar Fracture
reached the ultimate tensile capacity of reinforcement for both load-
The behaviors of the UDL and the CL static tests at the state of final ing cases. As a consequence, if the maximum beam-end rotation
collapse(i.e., top rebar fracture) were investigated via relative rota- θmax was assumed to be equal in both cases, HCL would be obvi-
tion θ between the beam end and the end joint; θ was measured in ously larger than HUDL . In the static tests when the specimens
the tests by two displacement transducers Lm1 and Lm2 arranged failed, HCL and HUDL were 148.8 and 81.9 kN, respectively. There-
at this location, as shown in Fig. 10(a). The development of this fore, although having the same beam-end rotation at the final stage
rotation with respect to MJD is presented in Fig. 10(b) for FR-U, of top rebar fracture, CA could not be fully utilized in the UDL tests

© ASCE 04018235-9 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 end
joint

Lm2

Lm1
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 10. Comparison of beam-end rotations at first top rebar fracture: (a) measurement method; and (b) rotation responses.

end end
joint joint

(a) (b)

Fig. 11. Total horizontal reaction and rebar forces at the end joint at state of first top rebar fracture: (a) CL case; and (b) UDL case.

unlike in the CL tests because part of the end-joint section was still
under compression.
Analytical Relationship of Structural Behaviors
between UDL and CL Conditions
On the basis of the results of the beam-column tests under the
UDL (section “Static Test Results of FR-U and PR-U”) and CL
(Lim et al. 2015) cases, the relationship between behaviors from
these two loading configurations was developed. The structural re-
sponses at the initial yielding stage, large-deformation stage, and
final failure stage were compared. The structural capacity of a
double-span beam at the yielding point under both UDL and CL
conditions can be simply calculated by assuming the formation
of plastic hinges at both the middle and the end joints, as shown
in Fig. 12(a). Hence, neglecting the effect of axial compression
force and basing on analytical relationships between applied loads
and moment distributions of the two cases, Eqs. (3) and (4) show
that the total applied load from the UDL case (PUDL ) is twice that of
the CL case (PCL )
where M mid and M end = plastic bending moment capacities at the
middle and the end joints, respectively. They can be simply calcu-
lated on the basis of the equation proposed by Paulay and Priestley
(1992) as follows:
M ¼ As f y ðd − d 0 Þ ð5Þ
where As and fy = area and yield strength of tension rebars, respec-
tively; and d and d 0 = distances from tension and compression
rebars to the extreme compressive fibre of concrete, respectively.
Based on Eqs. (3)–(5), the PCL and PUDL at yielding state of the
specimens used in this study (Fig. 1) were equal to 28.5 and
57.0 kN, respectively, which are relatively similar to test results
(24.7 and 53.9 kN in Table 3).
When the MJD is relatively large and the beam axial force starts
to significantly affect structural response, that is, during CAA and
CA stages, on the basis of the equilibrium of moment from half of
the structure as illustrated in Fig. 12(b), the relationships between
total applied loads and horizontal reactions can be described as
follows:

PCL ð2LÞ 2ðM mid þ M end þ HCL δ CL Þ

¼ M mid þ M end ð3Þ PCL ¼ ð6Þ
4 L

qð2LÞ2 PUDL ð2LÞ 4ðM mid þ M end þ HUDL δ UDL Þ

¼ ¼ M mid þ M end ð4Þ PUDL ¼ ð7Þ
8 8 L

© ASCE 04018235-10 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 (a)
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

(b)

Fig. 12. Relationship between CL and UDL responses: (a) moment distribution at yielding point; and (b) force equilibrium at large deformation.

where H CL and HUDL = horizontal reactions for the CL and UDL Measuring Location (m)
cases, respectively; and δ CL and δUDL = MJDs from the CL and 0 0.5 1 1.5 2 2.5
UDL cases, respectively. 0
From the static tests results [Fig. 5(b)], HCL and H UDL were -50
relatively similar for the same MJD (i.e., δCL ¼ δ UDL ) before
Vertical deflection (mm)

any rebar fracture occurred. Applying this observation, from -100

Eqs. (6) and (7), PUDL is equal to 2PCL , which was also observed -150
in the static test results for large MJDs, that is, Stages II and III
-200
[Fig. 5(a)].
Regarding maximum MJD of the static tests at the failure state -250
of top rebar fracture (θ ¼ θmax ), if the displacement profile of
-300
the CL case follows a linear function defined in Eq. (8), then the FR-U (test)
corresponding MJD of this loading case at θmax can be calculated -350 FR-U (analytical)
according to Eq. (9) as follows: -400
δ
y ¼ CL ðx − LÞ ð8Þ Fig. 13. Displacement profiles under UDL.
L

maxðδ CL Þ ¼ tanðθmax Þ · L ð9Þ

To analytically predict the displacement profile of UDL tests,
a second-order function presented in Eq. (10) is used, which agrees
reasonably well with actual displacement profiles (Fig. 13). The
corresponding MJD at θmax applying the second-order function
is presented in Eq. (11). Hence, if it is assumed that θmax is similar
for both loading cases at the stage of first top rebar fracture,
maxðδ UDL Þ is equal to 0.5 of maxðδ CL Þ. The test results from
FR-U and FR indicated that this ratio was 0.58 (Table 3), which
shows that the proposed method to predict the maximum displace-
ment under UDL based on that under CL is conservative
δ UDL 2
y¼ ðx − L2 Þ ð10Þ
L2
horizontal reaction is similar between the two loading conditions.
This relationship is valid not only at the yielding stage but also at
the CAA and the CA stages, until fracture starts occurring at either
top rebars (UDL) or bottom rebars (CL). At the final stage of
catenary action (top rebar fracture), maximum deformation of the
UDL case can be conservatively equal to half of that from the CL
case. The relationship on structural behavior between the two load-
ing cases is important for studying frame responses under UDL
because no static experimental study has been previously con-
ducted for this loading method.
Dynamic Test Results of FDU1 to FDU3

L Overall Responses
maxðδ UDL Þ ¼ tanðθmax Þ · ð11Þ
2
Before conducting the dynamic tests, axial loads were applied to
In summary, at the same MJD, the vertical load–carrying ca- the side columns via a hydraulic jack installed on each side column.
pacity of the UDL case is twice that of the CL case, whereas total At the same time, temporary supports for the steel-plate systems

© ASCE 04018235-11 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 Table 4. Summary of dynamic test results
Initial total Damping Maximum Maximum
Applied vertical Release Vibration ratio of vertical horizontal
load reaction time, period, vibration Maximum Residual reaction reaction
Test (kN) (kN) tr (ms) T d (ms) tr =T d phase (%) MJD (mm) MJD (mm) (kN) DLIF (kN)
FDU1-P/42 42 20.5 70 285 0.25 3.5 36 30.4 54.4 1.30 −0.3a
FDU1-P/61b 61 25.6 70 Collapse 74.9
FDU2-F/55 55 27.8 22 305 0.07 3.5 250 248.0 61.1 1.11 21.2
FDU3-P/61c 61 29.0 65 305 0.21 3.5 485 480.0 76.7 1.26 75.1
a
Only compression force was mobilized.
b
Second release of specimen FDU1.
c
Axial force of side column lost after the test.
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

underneath the specimen [Item (9) in Fig. 3(b)] were slowly re-
moved. As a result, self-weights from the steel plates were first car-
ried by the single span, causing initial readings in the pin supports,
as shown in Table 4, which summarizes important results of the
dynamic tests. After readings from all data channels were stable,
the rope from the quick-release device was suddenly yanked free,
releasing the specimen in a free-fall movement. If there was no fail-
ure, the specimen vibrated freely in its final deformation position
and slowly came to rest under damping effect. Otherwise, the struc-
ture lost its bearing capacity and dropped to the floor with a loud
thud. Overall deformations of all the dynamic tests are shown in
Fig. 14(a). Among the four dynamic tests, only FDU1-P/42 with
42 kN of loads sustained a small deformation. The others were
under higher loads of 55 and 61 kN and attained maximum MJDs
larger than one beam depth, indicating that the structures already
entered into Stage III of response. The release duration tr of the
quick-release device, measured by strain gauges attached on the
middle-joint hooks, was compared with fundamental period T d
of the specimens during the vibration stage (Table 4). It was found
that tr =T d did not exceed 0.25 for all the surviving specimens,
indicating that the release duration was significantly smaller than
the natural period of vibration (Biggs 1964). The damping ratio of
the oscillating phase was defined on the basis of the logarithmic
decrement method and was approximately equal to 3.5% for all
specimens [Table 4 and Fig. 14(b)].
On the basis of the load-pin and load-cell readings, the total
vertical reactions and total horizontal reactions from the left side
of all dynamic tests were defined and plotted in Fig. 15. Whereas
the horizontal reaction of FDU1-P/42 was purely negative showing
that the specimen was still under flexure/CAA stage, the horizontal
reactions from the other tests changed from negative to positive,
illustrating the onset of CA. For FDU1-P/61, which was the second
test of specimen FDU1 and was the only one that totally collapsed,
both the vertical and the horizontal reactions dropped significantly
at 0.67 s, denoting a complete failure of the specimen. Thereafter,
the specimen hit the strong floor at 1.035 s. For specimens that
survived the free fall, their maximum vertical reactions were di-
vided by their applied loads to calculate the dynamic load increase
factor (DLIF), which measured the inertial effect of applied weights
on the structures. This factor ranged from 1.11 to 1.30, as shown
in Table 4.
Structural Damages and Failure Modes
Structural damages, failure modes, and deflected shapes of all the
UDL dynamic tests had good resemblance with the corresponding
UDL static tests, as shown in Fig. 16. The first release of the partial-
restraint specimen FDU1 under 42 kN of applied load only caused
minor flexural cracks to the structure. There was no sign of concrete
crushing, indicating that the specimen had not yet reached its maxi-
mum capacity under CAA. After the first test, FDU1 was pulled
back up to its original position. The second release of this specimen
was under 61 kN, which led to a complete collapse. The specimen
experienced multiple severe damages including the fracture of
top rebars at the left-side end joint and the fracture of all sectional
rebars at the right-side end joint. However, its partially restrained
column experienced less inward movement effect, similar to the
corresponding static test PR-U. For FDU2-F/55, multiple cracks
occurred at the end joints and in the two sides of the middle joint,
together with concrete crushing at bottom surfaces of the end joints.

Time (s) Time (s)

0 1 2 3 4 0 1 2 3 4
0 0

FDU1 -P/42 -5
-100
-10
-200 FDU2 -F/55 -15
MJD (mm)

MJD (mm)

-300 -20
-25
FDU1 -P/61

-400
FDU3 -P/61 -30
-500
specimen collapsed -35
-600 -40
(a) (b)

Fig. 14. General responses of dynamic tests: (a) MJD time history; and (b) damping ratio (FDU1-P/42).

© ASCE 04018235-12 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 Time (s) Time (s)
90 100
FDU1-P/42 FDU1-P/42
80 FDU1-P/61 FDU1-P/61
FDU2-F/55
80 FDU2-F/55

Horizontal reaction (kN)

70 FDU3-P/61

Vertical reaction (kN)

FDU3-P/61
60
60
50 40
40 20
30
0
20 0 1 2 3 4
10 -20
specimen hit the floor specimen hit the floor
0 -40
0 1 2 3 4
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 15. Structural reactions from dynamic tests: (a) total vertical reaction; and (b) total horizontal reaction (left side).

(a)

(b)

(c)

(d)

Fig. 16. Damage patterns and failure modes of dynamic tests: (a) FDU1-P/42; (b) FDU1-P/61; (c) FDU2-F/55; and (d) FDU3-P/61.

However, no rebar fracture was observed. For FDU3-P/61 which
had identical boundary condition and applied load with FDU1-
P/61, excessive vertical deflection (up to 485 mm) was captured,
together with severe concrete cracking and crushing. Top rebars at
the left-side end joint also fractured after the test. Under large
pulling-in effect from the double-span beam, the partially restrained
side column underwent an excessive deformation (up to 47 mm)
and experienced severe concrete crushing at its outer face. How-
ever, the specimen sustained the applied load, and no collapse took
place. Such behavior of large inward deformation for the side
column was not observed in the corresponding UDL static test
for PR-U [Fig. 6(b)], but occurred in the CL static test for PR
[Fig. 6(d)]. Nonetheless, after completing FDU3-P/61, the axial
force on the partially-restrained side column was only 35% of its
original value, owing to a widening gap between the hydraulic jack
and the steel plate on top of the column. This gap was created by
excessive inward movement of the end joint, which caused mis-
alignment with the vertical axis. As a result, it was questionable if
FDU3 had indeed survived the free fall since the column load had
dropped significantly.

© ASCE 04018235-13 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 Table 5. Summary of strain rates and DIF for rebars
Tensile rebars at Section 1 (end joint) Tensile rebars at Section 4 (middle joint)
DIF DIF
Maximum strain rate Yield Ultimate Maximum strain rate Yield Ultimate
Test (1=s) up to yielding strength strength (1=s) up to yielding strength strength
FDU1-P/42 0.224 1.175 1.056 0.06 1.143 1.046
FDU1-P/62 0.404 1.189 1.060 0.079 1.149 1.048
FDU2-F/55 0.490 1.194 1.062 0.097 1.154 1.050
FDU3-P/61 0.670 1.202 1.064 0.067 1.145 1.047

Strain Rate Effect on Material Strength Comparisons of Structural Responses between UDL
Static and UDL Dynamic Tests
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

The material dynamic increase factors (DIFs) for rebar yield

strength due to strain rate effect was calculated on the basis of em-
pirical equations proposed by Malvar (1998) [Eqs. (12) and (13)].
Strain rates were defined for beam reinforcement at Sections 1
and 4 [Fig. 4(b)] up to the yield strain value. If the maximum strain
rates used for yield strength were also applied to calculate rebar
ultimate strength, the corresponding DIF was computed following
Eq. (14). The results for maximum strain rates and DIFs are pre-
sented in Table 5. Basically, the strain rates of tensile rebars at
Section 1 (end joint) were greater than those at Section 4 (middle
joint). The maximum DIFs for rebar yield strength of Sections 1
and 4 were 1.20 and 1.15, respectively, because, under UDL con-
dition, bending moment was developed at a faster rate at the end
joint than at the middle joint. Regarding rebar ultimate strength, the
enhancement was negligible, as the largest DIF was only 1.06.
In other words, strain rate effect on reinforcement had some effect
on the flexure/CAA capacity of the structure via increasing rebar
yield strength, but only marginal influence on CA capacity, which
relies on rebar ultimate strength
 α
ε̇
DIF ¼
10−4
fy
For yield strength∶ α ¼ αfy ¼ 0.074 − 0.040
414
fy
For ultimate strength∶ α ¼ αfu ¼ 0.019 − 0.009
414
In the static tests, all data were collected corresponding to MJD; in
dynamic tests, they were recorded with respect to time. Therefore,
to compare the behavior of the UDL static tests with the corre-
sponding UDL dynamic tests at the structural level, data from dy-
namic tests such as vertical and horizontal reactions were plotted
with reference to MJD instead of time. Fig. 17 shows the compar-
isons between the dynamic tests under the applied loads of 42 and
55 kN with the static tests of FR-U and PR-U. Due to the initial
effect of applied weights distributing on the single span before the
release, which was mentioned in Fig. 15(a) and Table 4, vertical
reactions from the dynamic tests had initial nonzero values at zero
MJD compared with vertical applied loads from the static tests.
Other than that, responses of the vertical and horizontal reactions
from both test series had reasonably good agreement. For the dy-
namic tests under 61 kN (Fig. 18), they also displayed a similar
trend of response to the relevant static test for PR-U. The peaks
of CAA from the dynamic tests were larger than those from the
static tests by 8% and 14% for the full- and the partial-restraint
specimens, respectively. This could be due to the enhancement in
rebar yield strength, which led to an increase of plastic bending
ð12Þ
moment resistances M mid and Mend , as explained in the section
“Strain Rate Effect on Material Strength.”
FDU1-P/61 and FDU3-P/61 had a sudden drop in structural
capacity (for both vertical and horizontal reactions) at an MJD of
ð13Þ 320 and 324 mm, respectively. It was assumed to be the state when
top rebars at the end joints started fracturing during these two
dynamic tests, which were comparable with PR-U and FR-U. The
beam-end rotations at this state were calculated as 9.4° and 9.6° for
FDU1 and FDU3, respectively. It is noteworthy that such rotations
ð14Þ
had good agreement with those from the UDL and the CL static

MJD (mm) MJD (mm)

120 120
PR-U (vertical) FR-U (vertical)
100 100 FDU2-F/55 (vertical)
Applied load/Reaction (kN)

FDU1-P/42 (vertical)
Applied load/reaction (kN)

PR-U (horizontal) FR-U (horizontal)

80 FDU1-P/42 (horizontal) 80 FDU2-F/55 (horizontal)
60 60
40 40
20 20
0 0
0 50 100 150 200 0 100 200 300 400
-20 -20
-40 -40
(a) (b)

Fig. 17. Structural responses of dynamic tests with load levels of 42 and 55 kN: (a) full-restraint specimens; and (b) partial-restraint specimens.

© ASCE 04018235-14 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 MJD (mm) MJD (mm)
90 100
80 80 PR-U

Vertical load/reaction (kN)

Horizontal reaction (kN)

70 FDU1-P/61
60 FDU3-P/61
60
50 40
40 20
30
PR-U 0
20 0 100 200 300 400 500
FDU1-P/61
-20
10 FDU3-P/61
0 -40
0 100 200 300 400 500
(a) (b)
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.

Fig. 18. Structural responses of dynamic tests with 61 kN: (a) vertical load/reaction; and (b) horizontal reaction.

tests (Table 3). After this failure, the structural responses of MJD (mm)
FDU1-P/61 dropped to zero, probably due to initial damages 70
FDU1-P/61 (collapsed)
caused by the first release. On the other hand, the structural 60
response of FDU3-P/61 did not drop to zero at the end of the FDU3-P/61

Applied load (kN)

50 FDU1-P/42
test, denoting that a complete collapse did not occur, although FDU2-F/55
the column axial load dropped to one-third of its original value. 40
30
Comparison between UDL Dynamic Tests Conducted
20
in This Study and CL Dynamic Tests in Pham and Tan
10
The load levels used in the dynamic tests in this study and in Pham
and Tan (2017) were different. Whereas the load levels in Pham and 0
Tan (2017) were 20, 25, and 30.5 kN for partial restraint and 20, 29, 0 200 400 600
and 34 kN for full restraint, those used in the UDL dynamic tests Pseudo-static (FR-U) Pseudo-static (PR-U)
were 40 and 61 kN for partial restraint and 55 kN for full restraint. Pseudo-static (FR) Pseudo-static (PR)
The reason for this difference is the difference in the load-carrying Dynamic tests
capacities between the CL and UDL cases observed for quasi-static
tests. In the free-fall dynamic tests, the loading levels were chosen Fig. 19. Comparison between pseudostatic predictions and actual
to maximize the behavior observed, from small deformation (com- dynamic tests.
pressive arch action) to large deformation (catenary action). There-
fore, the results from the CL dynamic tests could not be directly
compared with those from the UDL dynamic tests. In terms of boundary condition on UDL tests because the maximum MJD
damage patterns and rebar fracturing sequence, the CL dynamic from the UDL tests was significantly smaller than that from the
tests in Pham and Tan (2017) closely followed the CL static tests CL tests. When comparing the maximum dynamic capacities
in Lim et al. (2015), which were quite different from the UDL between the UDL and the CL tests predicted from the Izzuddin
dynamic tests. method, FR-U was 1.8 times larger than FR, and PR-U was 2.1
times larger than PR.
The maximum MJD versus applied load results from all the dy-
Comparison of Load-Carrying Capacity between namic tests were compared with the dynamic responses predicted
Actual Dynamic Tests and Simplified Energy-Based by the Izzuddin method, as shown in Fig. 19. Among the three load
Method levels applied in the dynamic tests, 42 kN was close to the predicted
dynamic capacity at the first peak of CAA (MJD of 0.4dbeam ),
The energy-based method proposed by Izzuddin et al. (2008) was 55 kN was almost equal to the ultimate dynamic capacity of both
applied to predict the pseudostatic behavior on the basis of the FR-U and PR-U, and 61 kN was slightly larger than these load
static responses of the UDL and the CL static tests, and the results limits (Fig. 19). The simplified method provided reasonable and
are illustrated in Fig. 19. It is shown that FR-U and PR-U had rel- conservative results compared with the actual dynamic capacities
atively identical dynamic responses with comparable maximum at load levels of 42 and 55 kN. For the applied load of 61 kN,
dynamic capacities. For FR-U, although the static peak of the one- whereas FDU1-P/61 showed a complete collapse, FDU3-P/61 sus-
side top rebar fracture was 29% larger than that of the first top rebar tained this load and possessed a higher MJD capacity than the pre-
fracture (Table 3), the difference for the corresponding dynamic diction from Izzuddin method. However, due to the loss of column
peaks was only 5%. Therefore, the state of first top rebar fracture axial force during the test, FDU3-P/61 could also be considered as
can be conservatively considered as the final failure mode for both having collapsed, making the load level of 61 kN exceed the actual
FR-U and PR-U. When comparing the difference of maximum dynamic capacity of the specimen. Therefore, for safety, predic-
dynamic capacity between the partial- and the full-restraint condi- tions from the Izzuddin energy-based method in the CA stage
tions, FR-U was only 7.5% larger than PR-U, whereas FR was 28% should be considered as the ultimate capacity that a structure can
larger than PR. Again, this emphasizes the weaker influence of reach under a sudden column removal scenario.

© ASCE 04018235-15 J. Struct. Eng.

J. Struct. Eng., 2019, 145(1): 04018235

 Conclusions
In this paper, a quasi-static and a free-fall dynamic test series were
conducted by applying equivalent UDL conditions. The results
from the UDL static tests were compared with previous CL static
tests on similar specimens. In addition, dynamic response caused
by the actual sudden removal of a supporting column was also in-
vestigated in the UDL dynamic tests. The following conclusions
are highlighted:
1. Using four-point loading method for RC frames reasonably
simulated the UDL condition in actual structures.
2. Differences between behaviors in CL and UDL methods in-
cluded load-carrying capacity, sequence of failure (no bottom
rebar fracture in UDL tests), deformation profile (linear in CL
but parabolic in UDL), and maximum deflection. The analytical
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.
calculation in the UDL case confirms a 100% increase for ver-
tical applied load but a 50% reduction in maximum deflection
compared with the CL case. However, the two loading methods
provided similar horizontal reactions for the same displacement.
Moreover, at the state of top rebar fracture, the UDL and CL
methods had the same beam-end rotation. This can be used as
the failure criterion for predictions.
3. Catenary action, which was observed to significantly affect the
structural response under CL condition, showed a lesser degree
of mobilization under UDL condition, leading to a weaker in-
fluence of boundary condition on structural response.
4. The release device used in the dynamic tests could simulate well
the sudden loss of a supporting column. Dynamic effects of
applied loads were observed when comparing with the corre-
sponding UDL static. Material DIFs due to strain rate effect
showed some increase for flexure/CAA capacity but little effect
on CA capacity.
5. The dynamic tests displayed similar structural response, failure
modes, and displacement profile to those of corresponding UDL
static tests, showing that the increase of loading rate from a
quasi-static (10−4 =s) to a free-fall environment (up to 0.67=s)
did not change the mechanism or failure mode. The dynamic
tests, although following a force-controlled manner, had a
relatively similar trend of reaction responses to those of the
static tests.
6. Dynamic capacities predicted by using the Izzuddin method
provided accurate and conservative predictions compared with
actual dynamic responses. Nonetheless, in engineering practice,
the maximum load during CA stage calculated by this method
should be considered as the ultimate capacity of a structure
under sudden column loss scenarios.
Acknowledgments
This study was financially supported by the research grant GPC:
MHA 191/9/1/345 provided by Ministry of Home Affairs, Singa-
pore, as well as from the Grant 107.01-2018.01 provided by
National Foundation For Science and Technology Development
(NAFOSTED), Vietnam. The authors greatly appreciate the finan-
cial supports by both organizations.
References
Bao, Y., S. Kunnath, S. El-Tawil, and H. Lew. 2008. “Macromodel-based
simulation of progressive collapse: RC frame structures.” J. Struct. Eng.
134 (7): 1079–1091. https://doi.org/10.1061/(ASCE)0733-9445(2008)
134:7(1079).
Bao, Y., H. S. Lew, and S. K. Kunnath. 2014. “Modeling of reinforced
concrete assemblies under column-removal scenario.” J. Struct. Eng.
© ASCE
J. Struct. Eng., 2019, 145(1): 04018235
140 (1): 04013026. https://doi.org/10.1061/(ASCE)ST.1943-541X
.0000773.
Biggs, J. M. 1964. Introduction to structural dynamics. New York:
McGraw-Hill College.
Choi, H., and J. Kim. 2011. “Progressive collapse-resisting capacity of RC
beam–column sub-assemblage.” Mag. Concr. Res. 63 (4): 297–310.
https://doi.org/10.1680/macr.9.00170.
Izzuddin, B., A. Vlassis, A. Elghazouli, and D. Nethercot. 2008.
“Progressive collapse of multi-storey buildings due to sudden column
loss. Part I: Simplified assessment framework.” Eng. Struct. 30 (5):
1308–1318. https://doi.org/10.1016/j.engstruct.2007.07.011.
Kang, S. B., and K. H. Tan. 2016. “Analytical model for compressive
arch action in horizontally restrained beam-column subassemblages.”
ACI Struct. J. 113 (4): 813–826. https://doi.org/10.14359/51688629.
Lew, H., Y. Bao, S. Pujol, and M. A. Sozen. 2014. “Experimental study
of reinforced concrete assemblies under column removal scenario.”
ACI Struct. J. 111 (1–6): 881–892.
Lim, N. S., C. K. Lee, and K. H. Tan. 2015. “Experimental studies on 2-D
RC frame with middle column removed under progressive collapse.”
In Proc., FIB Symp. 2015. Lausanne, Switzerland: International
Federation for Structural Concrete.
Liu, C., T. C. Fung, and K. H. Tan. 2016. “Dynamic performance of
flush end-plate beam-column connections and design applications in
progressive collapse.” J. Struct. Eng. 142 (1): 04015074. https://doi
.org/10.1061/(ASCE)ST.1943-541X.0001329.
Lu, X., K. Lin, Y. Li, H. Guan, P. Ren, and Y. Zhou. 2017. “Experimental
investigation of RC beam-slab substructures against progressive col-
lapse subject to an edge-column-removal scenario.” Eng. Struct. 149:
91–103. https://doi.org/10.1016/j.engstruct.2016.07.039.
Malvar, L. J. 1998. “Review of static and dynamic properties of steel
reinforcing bars.” ACI Mater. J. 95 (5): 609–616.
Orton, S. L., and J. E. Kirby. 2014. “Dynamic response of a RC frame under
column removal.” J. Perform. Constr. Facil. 28 (4): 04014010. https://
doi.org/10.1061/(ASCE)CF.1943-5509.0000464.
Paulay, T., and M. Priestley. 1992. Seismic design of reinforced concrete
and masonry buildings. New York: Wiley.
Pham, A. T., and K. H. Tan. 2017. “Experimental study on dynamic
responses of reinforced concrete frames under sudden column removal
applying concentrated loading.” Eng. Struct. 139: 31–45. https://doi.org
/10.1016/j.engstruct.2017.02.002.
Pham, X. D., and K. H. Tan. 2013. “Experimental study of beam–slab sub-
structures subjected to a penultimate-internal column loss.” Eng. Struct.
55: 2–15. https://doi.org/10.1016/j.engstruct.2013.03.026.
Pham, X. D., and K. H. Tan. 2015. “Experimental response of beam-
slab substructures subject to penultimate-external column removal.”
J. Struct. Eng. 141 (7): 04014170. https://doi.org/10.1061/(ASCE)ST
.1943-541X.0001123.
Qian, K., and B. Li. 2012. “Dynamic performance of RC beam-column
substructures under the scenario of the loss of a corner column—
Experimental results.” Eng. Struct. 42: 154–167. https://doi.org/10
.1016/j.engstruct.2012.04.016.
Qian, K., B. Li, and Z. Zhang. 2016. “Influence of multicolumn removal on
the behavior of RC floors.” J. Struct. Eng. 142 (5): 04016006. https://
doi.org/10.1061/(ASCE)ST.1943-541X.0001461.
Ren, P., Y. Li, X. Lu, H. Guan, and Y. Zhou. 2016. “Experimental
investigation of progressive collapse resistance of one-way reinforced
concrete beam–slab substructures under a middle-column-removal
scenario.” Eng. Struct. 118: 28–40. https://doi.org/10.1016/j.engstruct
.2016.03.051.
Sadek, F., J. A. Main, H. S. Lew, and Y. Bao. 2011. “Testing and analysis of
steel and concrete beam-column assemblies under a column removal
scenario.” J. Struct. Eng. 137 (9): 881–892. https://doi.org/10.1061
/(ASCE)ST.1943-541X.0000422.
Sagiroglu, S. 2012. “Analytical and experimental evaluation of progressive
collapse resistance of reinforced concrete structures.” Ph.D. disserta-
tion, Dept. of Civil and Environmental Engineering, Northeastern Univ.
Sasani, M. 2008. “Response of a reinforced concrete infilled-frame
structure to removal of two adjacent columns.” Eng. Struct. 30 (9):
2478–2491. https://doi.org/10.1016/j.engstruct.2008.01.019.
04018235-16 J. Struct. Eng.
 Sasani, M., M. Bazan, and S. Sagiroglu. 2007. “Experimental and analyti-
cal progressive collapse evaluation of actual reinforced concrete struc-
ture.” ACI Struct. J. 104 (6): 731–739.
Sasani, M., and J. Kropelnicki. 2008. “Progressive collapse analysis of
an RC structure.” Struct. Des. Tall Special Build. 17 (4): 757–771.
https://doi.org/10.1002/tal.v17:4.
Sasani, M., and S. Sagiroglu. 2010. “Gravity load redistribution and
progressive collapse resistance of 20-story reinforced concrete structure
following loss of interior column.” ACI Struct. J. 107 (6): 636–644.
Stinger, S. M., and S. L. Orton. 2013. “Experimental evaluation of dispro-
portionate collapse resistance in reinforced concrete frames.” ACI
Struct. J. 110 (3): 521.
Su, Y., Y. Tian, and X. Song. 2009. “Progressive collapse resistance of
axially-restrained frame beams.” ACI Struct. J. 106 (5): 600–607.
Tian, Y., and Y. Su. 2011. “Dynamic response of reinforced concrete beams
following instantaneous removal of a bearing column.” Int. J. Concr.
Downloaded from ascelibrary.org by Sergio Barreiro on 01/07/19. Copyright ASCE. For personal use only; all rights reserved.
Struct. Mater. 5 (1): 19–28. https://doi.org/10.4334/IJCSM.2011.5
.1.019.
© ASCE 04018235-17
J. Struct. Eng., 2019, 145(1): 04018235
Valipour, H., N. FarhangVesali, and S. Foster. 2013. “A generic model for
investigation of arching action in reinforced concrete members.” Constr.
Build. Mater. 38: 742–750. https://doi.org/10.1016/j.conbuildmat
.2012.09.046.
Yi, W. J., Q. F. He, Y. Xiao, and S. K. Kunnath. 2008. “Experimental
study on progressive collapse-resistant behavior of reinforced concrete
frame structures.” ACI Struct. J. 105 (4): 433–439..
Yu, J., and K. H. Tan. 2013a. “Experimental and numerical investigation
on progressive collapse resistance of reinforced concrete beam column
sub-assemblages.” Eng. Struct. 55: 90–106. https://doi.org/10.1016/j
.engstruct.2011.08.040.
Yu, J., and K. H. Tan. 2013b. “Structural behavior of rc beam-column
subassemblages under a middle column removal scenario.” J. Struct.
Eng. 139 (2): 233–250. https://doi.org/10.1061/(ASCE)ST.1943-541X
.0000658.
Yu, J., and K. H. Tan. 2017. “Structural behavior of reinforced concrete
frames subjected to progressive collapse.” ACI Struct. J. 114 (1):
63–74. https://doi.org/10.14359/51689424.
J. Struct. Eng.