Engineering Structures 34 (2012) 95110

Contents lists available at SciVerse ScienceDirect

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

Dynamic collapse test on eccentric reinforced concrete structures

with and without seismic retrot
Yousok Kim a,, Toshimi Kabeyasawa a, Shunichi Igarashi b
a
b

Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi Bunkyo-Ku, Tokyo 113-0032, Japan
Structural Quality Assurance Inc., 2-7-10 Samikicho Chioda-Ku, Tokyo 101-0061, Japan

a r t i c l e

i n f o

Article history:
Received 31 May 2010
Revised 5 September 2011
Accepted 6 September 2011
Available online 4 November 2011
Keywords:
Reinforced concrete
Torsional response
Seismic retrot
Shear failure
Axial collapse
Shake table

a b s t r a c t
Reconnaissance reports on structures damaged or collapsed by severe earthquakes have revealed several
common characteristics in their structural members and systems, such as insufcient reinforcement
details in beam-column joints and transverse connements, low aspect ratios, soft and or weak stories,
and eccentric plans. Dynamic tests were carried out to investigate the collapse process of reinforced concrete structures that had seismically decient reinforcement details (light transverse reinforcement) and
seismic systems (soft/weak stories and eccentric plans). A comparison of collapse behaviors with and
without seismic retrots also veried the effectiveness of the SRF (super reinforced with exibility)
strengthening method, which was developed to prevent the loss of axial load carrying capacity even at
excessive lateral deformation. The columns of one specimen were strengthened with polyester ber belts
and its shear walls with polyester ber sheets, while the members of the other were not. Each specimen
was designed following old (1970s) reinforcement detail practice in Japan, and is a one-third-scale reinforced concrete structure with considerable stiffness and strength eccentricity in the rst story. The specimens were composed of independent column frame and shear wall frame. Torsional response resulting
from the eccentricity in the 1st story induced a displacement concentration on the weak frame, and eventually the independent columns of the RC specimen failed in shear and lost their axial load carrying
capacity. On the other hand, the SRF specimen survived not only an identical earthquake load to the
one that caused the RC specimen to fail, but also three additional earthquake loads, although signicant
strength deterioration and considerable lateral and vertical deformation were generated at the end of the
test. The following conclusions were drawn from the comparison of the two specimens responses: axial
column collapse cannot be predicted from vertical responses since the vertical behavior of bare RC columns was not discernibly different from that of SRF columns until axial collapse was initiated, and the
SRF strengthening method is effective in conning the column and preventing the cracking progress, thus
modifying the failure mode of the RC columns from brittle shear failure after exural yield to exural
dominant behavior.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Collapsing buildings have caused numerous casualties during
recent major earthquakes, even in countries with advanced
earthquake engineering technologies. Understanding the collapse
mechanisms of reinforced concrete (RC) structures subjected to
earthquake attack is a crucial aspect of one of the goals of earthquake engineering, which is to protect peoples lives and safety,
and has been a challenging task in many researches [13]. Many
RC structures have suffered severe damage in devastating earthquakes in the past, leading to unrecoverable damages and the
collapse of structures. Reconnaissance reports describing these

Corresponding author. Tel.: +81 3 5841 5783; fax: +81 3 5841 1765.
E-mail address: yskim@eri.u-tokyo.ac.jp (Y. Kim).
0141-0296/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2011.09.017

damaged or collapsed structures have revealed several common

characteristics in their structural members such as insufcient
reinforcement details in beam-column joints, transverse connement designed following old seismic codes, and low aspect ratio
(shear span to depth ratio) members that are susceptible to shear
failure [47]. In addition to these seismic element deciencies,
the unbalanced layout of structural members in elevation or plan
(i.e. buildings with soft/weak stories or eccentric plans) caused
poor seismic performance leading buildings to collapse.
Although research themes in earthquake engineering are turning to innovative technologies for new structures, it is still important to continue developing economical retrotting methods for
existing buildings that have seismically decient members and
systems in order to reduce injuries and loss of life and property.
To perform safely in an earthquake, vertical members (columns
and walls) must maintain their gravity load carrying capacity even

96

Y. Kim et al. / Engineering Structures 34 (2012) 95110

when earthquake intensity exceeds the members design limits. In

some old reinforced concrete buildings, the columns can lose their
axial load carrying capacity due to inadequate transverse reinforcement, then the buildings collapse at the weakest story or
completely pancake. The development of a simple and economical
strengthening method to prevent these brittle columns from
gravity load collapse during strong earthquakes would be very
benecial.
A new strengthening method for RC columns called the super
reinforced with exibility method (SRF) focuses particularly on
maintaining the columns axial load carrying capacity even under
excessive lateral deformation. It was developed and has been
demonstrated using static tests on columns with a variety of
cross-sections under different loading patterns [811]. In the rst
phase of this testing, eight specimens were tested, with 14 in a second phase. The specimens were constructed to simulate reinforced
concrete columns in old buildings in Japan or Turkey with light
transverse reinforcement and relatively low member aspect ratios.
Test results showed that the columns strengthened by the SRF
method could maintain relatively high gravity loads through drift
ratios of more than ten percent, while the bare specimens failed
in shear at small drift ratios, simultaneously losing axial load carrying capacity. Through such testing, the retrot scheme has been
improved to be effective in preventing loss of capacity against not
only axial loads but lateral load reversals as well. In the shake table
test presented in this paper, we simultaneously tested two eccentric wall-frame specimens with identical section details and material properties. One specimen was strengthened using the SRF
method while the other was not.
This study investigates several characteristics of structures that
suffer from severe damage or collapse during earthquakes, including: (1) the lack of sufcient shear strength in RC columns designed
following 1970s Japanese reinforcement detail practice, which
leads to shear failure after exural yielding and the loss of axial
load carrying capacity; (2) irregular layout along the height (i.e.
soft/weak rst stories); and (3) asymmetric 1st story plans composed of independent column and wall frames that generate considerable stiffness and strength eccentricity. In addition, another
aim of this experimental program is to conrm the effectiveness
of SRF strengthening under dynamic loading conditions to support
static tests that have demonstrated its value.
The primary objectives of this experiment, therefore, are to
investigate and understand the collapse process of reinforced concrete structures with both seismically decient sections (poor
transverse reinforcement) and structural member layout (soft/

weak stories and eccentric plans) and to conrm the effectiveness

of the proposed SRF retrot scheme by comparing the seismic
behaviors and collapse modes of two specimens that are identical
except for being strengthened or not. We also investigate the inuence of stiffness and strength eccentricity on elastic and inelastic
earthquake responses under severe earthquake loads by observing
the seismic behaviors of the two specimens.
2. Shake table test description
2.1. SRF strengthening method
The polyester ber reinforcing method used here was originally
developed to improve vertical members ability to sustain axial
loads under large lateral deformation and to prevent the collapse
of structures in severe earthquake loadings. The important characteristics of the SRF material are its toughness, durability, heatresistance and exibility.
The results of tensile tests on SRF sheets and belts are given in
Fig. 1 and the average material properties obtained from three
tensile tests are summarized in Table 1. These results show an almost linear relationship between stress and strain until the sheet
and belt fail at very large strains of 16% and 35%, respectively.
The tensile strength of belt is almost twice as high as that of sheet,
while the Youngs moduli are almost same in both materials and
lower than that of concrete. Much higher tensile force is expected
from belt (3 mm thick) than sheet (0.9 mm thick).
Externally bonded steel or ber reinforced polymer (FRP) are
also very effective in strengthening the structural members of
existing buildings which cannot satisfy the current design
demands [12,13]. FRP strengthening method, of which advantages
over steel plate bonding method are easy installation due to the
light weight, chemical resistance and lower labor cost, was proposed to overcome the shortcomings of steel plate bonding method. In special, CFRP (carbon ber reinforced polymer) retrot
method enhanced both stiffness/strength and ductility of structural members. Improved structural behaviors are resulted from
the high strength and stiffness of CFRP materials. Furthermore,
adhesive which is essential in the external plating/wrapping method is also stiffer and stronger than concrete. However, some experimental researches have also revealed that peeling off and shear
cracks at the plate ends resulted in premature, brittle failure of
RC members externally bonded with FRP plates [1416]. Therefore,
it has been recommended that in strengthening applications, the
external FRP should fail in tension after yielding the internal steel

400

400
Belt

300

Stress (Mpa)

Stress (Mpa)

Sheet

200
100

300
200
100

0
0

10

20
Strain (%)

30

40

10

20
Strain (%)

30

40

Fig. 1. Tensile test results of SRF materials.

Table 1
Material properties of SRF belt and sheet.

Belt
Sheet

Thickness (mm)

Width (mm)

Tensile strength (MPa)

Strain at max. strength (%)

Elasticity modulus (MPa)

3
0.9

50

358.1
169.7

34.42
16.17

0.85  103
0.76  103

97

Y. Kim et al. / Engineering Structures 34 (2012) 95110

but before failure of the concrete in the compressive zone, since

this would ensure a more ductile failure mode.
On the other hand, SRF strengthening method, which adopts almost similar strengthening process to FRP method (i.e. plating or
wrapping through adhesive), do not show the sudden failure resulted from premature failure or debonding from concrete. The
main difference between SRF and FRP is strengthening material
properties. That is, polyester belt/sheet and urethane adhesive
which are applied in SRF strengthening method are much more
exible than concrete and steel. Therefore, its adherend such as
concrete surface is not damaged due to the fracture or peeling of
strengthening materials, which can be observed from stiff and
strong strengthening materials used in FRP retrot method.
The other different characteristic of SRF to FRP is that SRF
materials (polyester belt/sheet) resist only in tension not in
compression. In shear resistant behavior of RC members, two different stresses are generated, which are tensile and compressive
stresses across and along shear crack, respectively. In case of simply supported beam subjected to exural bending moment, two
different stress conditions which are tension in lower part and
compression in upper part of beam side are also generated in the
same surface. While the SRF materials resist in tensile direction
in which it exerts very high and ductile resistance, very low compressive stress is generated in SRF materials.
However, in FRP materials which are very stiff and resist both in
compression and tension, high compressive stress is concentrated
and cause debonding or peeling of FRP from concrete surface. It
was noted that peeling and shear cracks at the plate ends were
responsible for causing premature, brittle failure of FRP materials.
RC members strengthened with SRF materials show very high
performance in terms of ductility, since the SRF materials including
adhesives can endure more than 10% tensile strain without any
sudden failure. Furthermore, axial deformation can be expected
beyond 2% of column height, which means that axial load carrying
capacity at extremely large deformation can be preserved in the
vertical members strengthened by SRF materials and prevents
structures subjected to severe earthquake attack from collapse
(e.g. pancake collapse).
From the comparisons of SRF with FRP strengthening method
described above, it can be said that SRF retrot scheme is more
effective for improving post-peak behavior and axial load carrying
capacity of vertical members compared to FRP strengthening

X2

Y1

1100

Y2
st

st
1 1Floor

Y1

744

5344

1500 500
2500

600
1500

2700

ReX

eY
reX

800

nd

600
X

Two specimens with identical properties and dimensions were

constructed simultaneously. Fig. 2 shows plan and elevation of
the specimens. They had wall and column frame construction in
the rst story but had wall frames only in the second story. A large
damage concentration was expected in the 1st story independent
columns due to the irregular layout of structural members in both
plan and elevation. Each specimen was constructed in two parts
(connection level: 2044 mm, Fig. 2(b)) because of height limitations imposed on specimens constructed outside and moved into
the shake table testing facility.The specimens were built at 1/3
scale based on a prototype structure and thus should meet similitude requirements. Adding steel plates on the top of the specimen
preserves the axial stresses and shear coefcients of the prototype
six-story building. These steel plates (WS, 148.3 kN) together with
two concrete masses W1 and W2 (284.6 kN) produce an axial load
stress of 0.15 AgFc (Ag: gross area of the column section, Fc: concrete
strength, 18 MPa) in the rst story independent columns.
The total height of each specimen is 4544 mm, which is the sum
of the base (500 mm), load cells (244 mm), the 1st story (800 mm),
W1 (1100 mm), the 2nd story (800 mm) and W2 (1100 mm), as
shown in Fig. 2. The top of concrete mass W1 does not correspond
to the 2nd oor height of the prototype building, although it will be
referred to as the 2nd story throughout this paper. The 2nd story
response described here, therefore, just represents a response
above the 1st story where the upper story has a symmetric plan
with two walls.
An asymmetric plan comprised of columns and a wall frame in
the rst story generates considerable stiffness eccentricity for seismic motion in the X direction. The stiffness eccentricity in the rst
story is 0.71, as calculated from Eq. (1). The stiffness of the columns
and wall frame, 9.1  104 N/mm and 4.57  105 N/mm, respectively, were obtained from elastic analysis. The stiffness of the wall
frame in the 1st story is almost ve times greater than that of the
column frame.

2 Floor

500

2.2. Specimens

steel weight (WS)

Y2

800 500 600 800

2200

350 1500 350

X1

method at large deformation, although considerable enhancement

of initial stiffness and ultimate strength cannot be expected as
those of FRP strengthening method.

600

1500
2700

(a)

600

W2

box for accelerometer

2FL
W1

connection level

1FL
load cell
West
East
Loading direction

View from east

(b)
Fig. 2. Specimen plan and elevation. (a) Plan. (b) Elevation.

98

Y. Kim et al. / Engineering Structures 34 (2012) 95110

rP
P
P
2
K X Y
K X Y 2
P K Y X , eY: diswhere eY j lY  g Y j, lY P K , r eX
K
X

tances between centers of stiffness and mass, lY : center of stiffness,

g Y : the center of mass, K X , K Y : the frame stiffness in the X and Y
direction, respectively, Y Y  lY , X X  lX , X, Y: member
coordinate in the X- and Y-axis.
The independent columns in the rst story were designed
following 1970s Japanese reinforcement detail practice (Fig. 3(a)).
Their shear strength at exural yield and shear strength calculated
from Arakawas [17] empirical equations are 62.8 and 60.7 kN,
respectively. These calculated strengths show that the difference
between their shear and exural capacities is small. The shear
strength of the shear wall, however, calculated from Hirosawas
empirical equation [17] is 430 kN and its shear strength at exural
yield is 770 kN. As with the stiffness difference, the shear strength
of the wall frame is almost four times that of the independent column frame. The cross-sectional details of the wall and columns are
shown in Table 2, while the material properties of the concrete and
steel, obtained from cylinder and coupon tests, are summarized in
Tables 3 and 4, respectively.
One of the two specimens was strengthened with polyester
ber sheets and belts, while the other was not. The strengthened
specimen will hereafter be referred to as the SRF specimen whose
independent columns and shear wall with boundary columns in
the 1st story were strengthened. The latter will be called the RC
specimen throughout this paper.
The reinforcing methods and material properties of independent columns and walls are somewhat different (Fig. 4). Independent columns in the Y1 frame were spirally wrapped with a
single layer of 3-mm-thick polyester belt material. Fig 5 shows
the SRF column retrotting process, which is similar to that done
in previous static tests. The shear wall, with its boundary columns,
was wrapped with a double layer of polyester sheeting 0.9 mm
thick and 0.8 m high that covered the entire clear height of the
wall. Epoxy-urethane adhesive was used in both cases, applied
between the concrete surface of the members and the polyester

main bar

12-D10

hoop

2-D4@50

main bar

Table 3
Material properties of concrete.

Superstructure
Base

rB (MPa)

ec (l)

Ec (MPa)

rt (MPa)

24.1
25.37

1894
2060

21556
23096

2.38
2.22

Table 4
Material properties of steels.

D4
D6
D10

Es (MPa)

ry (MPa)

ey (l)

156490
185288
175137

188.4
439.1
352.4

1210
2372
2011

materials. The shear wall had two layers of sheeting, so adhesive

was also applied between the rst and second layers. It should
be noted that in this experiment, it took only a half-day for four
non-technicians (students) to reinforce two independent columns
and one shear wall in the 1st story without using any special
devices or machines.
2.3. Experimental setup and instrumentation
Fig. 6(a) shows the arrangement of the specimens on the shake
table, which is symmetric about the tables centroid. The two specimens, therefore, were subjected to identical earthquake loadings
until the RC specimen collapsed.
As described previously, the upper and lower sections of the
specimens were constructed separately. Two sub-structures were
assembled on the shake table during setup as shown in Fig. 7.
The lower part of each specimen was mounted on the shake table
rst and then connected to the upper part by tensioning high
strength steel bars that penetrated through holes in W1. Steel
weights (WS) were connected to top of W2 in the same manner.

12-D10

hoop: 2-D4@50 (1F)

hoop: 2-D6@50 (2F)
No.7
No.6
No.5

strain gauge
strain gauge

No.4

No.3
No.2
No.1

(a)

(b)

(c)

Fig. 3. Column section details. (a) Independent column. (b) Attached column to shear wall. (c) Hoop.

Table 2
Section details of members (unit: mm).
Floor

Column

(unit: mm)

Wall

(unit: mm)

B (width),D (depth)
Longitudinal reinforcement
Transverse reinforcement

200, 200
12-D10
2-D6@50

Thickness
Vertical and horizontal reinforcement

50
D6@100

B (width),D (depth)
Longitudinal reinforcement
Transverse reinforcement

200, 200
12-D10
2-D4@50

Thickness
Vertical and horizontal reinforcement

50
D6@100

99

Y. Kim et al. / Engineering Structures 34 (2012) 95110

Fig. 4. SRF specimen. (a) Independent column frame (SRF specimen vs. RC specimen). (b) Wall frame (SRF specimen vs. RC specimen).

Fig. 5. SRF column reinforcing process.

H-shaped steel beam

Steel frame

2nd Floor

SRF
specimen

:displacement transducer

West

East
:accelerometer

RC
specimen

1st Floor
load cell

(a)

(b)

(c)

Fig. 6. Instrumentation on 1st and 2nd oor plans. (a) Specimen arrangement on shake table. (b) Specimen overview and instrumentation. (c) Instrumentation on 1st and 2nd
oor plans.

The responses of the specimens (acceleration, displacement,

the strain of the steel bars and shear and axial forces) when subjected to earthquake excitation were recorded with a sampling
rate of 1000 Hz using accelerometers, displacement transducers,
strain gauges and load cells. Instrument placement is shown in

Fig. 6(b and c). To record horizontal and vertical accelerations

of concrete masses W1 and W2, accelerometers were installed
in boxes located inside of W1 and W2 as shown in Fig. 1. In addition, input acceleration was recorded by accelerometers on the
shake table.

100

Y. Kim et al. / Engineering Structures 34 (2012) 95110

Fig. 7. Specimen setup on shake table.

Laser displacement sensors, which require no physical connection between the target point and sensor, were used to measure
horizontal and vertical displacements of the 1st story where large
displacements were expected, while strain-type displacement
transducers were installed in the 2nd story. Strain gauges were
attached to the longitudinal and transverse column reinforcements
in the 1st story (Fig. 2). Bi-directional load cells capable of measuring both axial and lateral forces in the X direction were installed at
the bottom of the independent columns of the rst story. Instruments were located symmetrically about the geometric center of
the specimen in order to compare torsional responses between
columns (Y1 frame) and wall frames (Y2 frame). Identical
instrumentation was installed in both specimens.
In addition to installing data recording instrumentation, protection must be provided to prevent the specimen from falling down
onto the shake table after collapse, since this experiment was
designed to simulate the structures collapse. In this test, an
H-shaped steel column (in Y1 frame) was built to contain the rst
oor where axial collapse of columns was expected to occur.
H-shaped steel beams spanned the top of the steel frames, and
steel wires connected to the specimen hung from the beams.
2.4. Base motion input plan
Records of four different historical earthquakes were employed
as seismic excitations; the Miyagi-ken Oki earthquake recorded at
Tohoku university in 1978 (TOH), the Imperial Valley earthquake
recorded at El Centro in 1940 (ELC), the Hyogo-Ken Nambu earthquake recorded at the Japan Meteorological Agency in 1995 (JMA)
and the Chile earthquake recorded at Via del Mar in 1985 (CHI).
Northsouth components of these earthquake records were
applied unidirectionally to the specimens in the X direction.
Table 5 shows the amplitudes of the input base motions, scaled
on the basis of preliminary analysis results, under which the RC
specimen experienced gradual damage ranging from elastic behavior through nal collapse. The
p duration of the base motion was
compressed by a factor of 1= 3 to satisfy the similitude law, keeping the acceleration scale factor at unity. Here, the number in each
loading name (e.g. 12.5 in TOH12.5 or 50 in JMA50, etc.) indicates
the maximum velocity (kine, cm/s) corresponding prototype

records. Fig. 8(a and b) show the time history and acceleration
response spectra of each input base motion.
Although the shapes of the acceleration response spectra of the
actual input base motions recorded from the shake table are
slightly different than those of the original data, the difference is
acceptable. White noise excitations, whose maximum acceleration
is about 0.3 m/sec2, were also performed before the earthquake
loadings to evaluate changes in the natural frequency of the damaged specimens.
3. Test results
3.1. Lateral displacement response
Fig. 9 shows the drift ratio of the columns (Y1 frame) and wall
frame (Y2 frame) of the RC specimen subjected to loading ELC37.5.
Displacement through the 1st story was recorded from laser
displacement sensors set between the top of the specimen base
and the bottom of concrete mass W1. The 2nd story displacement
was recorded between the top of W1 and the bottom of W2. As
shown in Fig. 9(a), drift ratio of the Y1 frame in the 1st story is
almost 10 times larger than that of the Y2 frame because the
considerable eccentricity in the 1st story induced large torsional
responses. In the second story, which was symmetrical, the drift
ratio of the Y1 frame is slightly larger (about twice) than that of
the wall frame, which might also have been affected by the torsional response generated in the 1st story. The other loading inputs
produced similar results as shown in Fig. 9 (i.e. deformation concentration on the Y1 frame) in both the RC and SRF specimens.
The maximum lateral drift ratio of the Y1 and Y2 frames are compared in Fig. 10. The difference of the two frames maximum drift
ratio is almost constant throughout all the input stages and in both
specimens. Note that different scales on the vertical axis for the Y1
and Y2 frames are used in Figs. 9(a) and 10.
A comparison of the drift ratio of the two specimens when subjected to the JMA50 loading is shown in Fig. 11. The drift ratio of
the RC specimen is slightly larger than that of the SRF specimen
for both the Y1 and Y2 frames. This is consistent with the results
observed from all the other input stages, except for TOH12.5,
which can be seen in Fig. 12 where ratios comparing the maximum

Table 5
Base motion input plan.

Earthquake data

Scale factor

Max. acceleration to
specimen (m/sec2)

Max. velocity to
specimen (m/s)

Max. acceleration of
prototype (m/sec2)

Max. velocity of
prototype (m/s)

TOH12.5
TOH25
ELC37.5
JMA50
CHI50-1
TAK125*
CHI63*
CHI50-2*

0.3
0.6
1.1
0.6
0.7
1
0.9
0.7

0.77
1.55
3.76
4.92
6.19
6.06
7.96
6.19

0.07
0.14
0.22
0.29
0.28
0.71
0.36
0.28

2.58
2.58
3.42
8.21
8.84
6.06
8.84
8.84

0.41
0.41
0.35
0.85
0.71
1.24
0.71
0.71

Only for SRF specimen.

(0.013)
(0.25)
(0.38)
(0.5)
(0.5)
(1.25)
(0.63)
(0.50)

101

Acceleration (m/s 2)

Y. Kim et al. / Engineering Structures 34 (2012) 95110

TOH12.5

TOH25

ELC37.5

JMA50

CHI50-1

0
-5

Acceleration (m/s 2)

30
TAK125

60

90
Time (sec.)

120

CHI63

175

CHI50-2

0
-5
0

20

75

130

Time (sec.)

(a)
TOH25

ELC37.5

15

0.5
period (sec.)

15

15

10

10

10

0.5
period (sec.)

0.5
period (sec.)

0.5
period (sec.)

proto-type
shake table record
0

0.5
period (sec.)

(b)
Fig. 8. Input base motions. (a) Time history. (b) Acceleration response spectra (h = 0.03).

0.002
Y 1 f rame
Y 2 f rame

0.01
0

0.001
0

-0.01

-0.001
Time (sec.)

-0.02
2

-0.002

10

12

Drift ratio of Y2 frame

Drift ratio of Y1 frame

0.02

14

(a)
-4

-4

x 10
5

5
Y 1 f rame
Y 2 f rame

2.5
0

2.5
0

-2.5

-2.5
Time (sec.)

-5
2

-5

10

12

Drift ratio of Y2 frame

x 10
Drift ratio of Y1 frame

14

(b)

0.04

Y 1 f rame
Y 2 f rame

0.004

0.03

0.003

0.02

0.002

0.01

0.001

0
TOH12.5

ELC37.5

(a)

CHI50-1

0.05
0.04

Y 1 f rame
Y 2 f rame

0.004

0.03

0.003

0.02

0.002

0.01

0.001

0
TOH12.5

ELC37.5

CHI50-1

(b)

Fig. 10. Maximum drift ratios of Y1 and Y2 frames. (a) RC specimen. (b) SRF specimen.

Max. drift ratio of Y2 frame

0.05

Max drift ratio of Y1 frame

Fig. 9. Y1 and Y2 frame comparison of RC specimen drift ratio under ELC37.5 load input. (a) 1st Story. (b) 2nd Story.

Max. drift ratio of Y2 frame

TAK125

20

15

25

CHI50-1

20

10

JMA50

20

Max drift ratio of Y1 frame

Sa (m/s 2)

102

Y. Kim et al. / Engineering Structures 34 (2012) 95110

0.02
RC
SRF

Drift ratio

0.01
0
-0.01
Time (sec.)

-0.02
2

10

12

14

(a)

-3

x 10
2
RC
SRF

Drift ratio

1
0
-1
Time (sec.)

-2
2

10

12

14

(b)

RC / SRF

RC / SRF

Fig. 11. Comparison of RC and SRF drift ratios. (a) Y1 frame. (b) Y2 frame.

0
TOH12.5

ELC37.5

CHI50-1

TOH12.5

ELC37.5

(a)

CHI50-1

(b)

drift ratio of the RC and SRF specimens are displayed for the various loadings. The difference between two specimens is most outstanding in loading CHI50-1 where the RC specimen collapsed.
Since the SRF specimen exhibited smaller deformations than the
RC specimen when subjected to identical earthquake loadings, this
test showed the SRF retrotting method to be also effective under
dynamic loading conditions.

3.2. Torsional response

As shown previously, the displacement of the specimens is
dependent on the torsional response generated by eccentricity in
the 1st story. Although considerable torsional responses inducing
displacement concentration on the weak 1st story Y1 frame were
observed in both specimens for all inputs, these responses differed
for every loading input as well as for the strengthened or bare condition of the frame. Quantitative assessment of torsional response
in each input stage is needed to evaluate the characteristics of torsional behavior in both the elastic and inelastic ranges. The effectiveness of the SRF strengthening method can also be further
veried by quantitatively comparing those two specimens torsional responses.
The extent of torsional response is evaluated by the index r,
(Fig. 13(a)) representing the distance of a members response center from its center of gravity. This index can be calculated from Eq.
(2), which is the relationship between lateral displacement and
rotation angle at the center of mass. Thus r can be thought of as

Displacement

Fig. 12. Maximum RC/SRF specimen displacement ratios. (a)Y1 frame. (b) Y2 frame.

Rotation angle

(a)

(b)

Fig. 13. Response center distance, r. (a) Denition of r. (b) Calculation of r.

the slope of a line tted from the relationship between angle of

rotation and translational displacement at the center of gravity
(Fig. 13(b)).

d
h

here, d is the displacement of the center of a specimen, obtained by

averaging the displacements of the Y1 and Y2 frames, and h is
calculated from displacement difference divided by the distance
between the two frames.
For example, the value of r is zero in pure torsion and innite for
pure translation. Thus, the torsional response becomes dominant
as the value of r decreases.

103

Y. Kim et al. / Engineering Structures 34 (2012) 95110

RC specimen
SRF specimen

translation
r (m)

0.8

torsion
0.6
TOH25

ELC37.5

JMA50

small

CHI50

TAK125

specimen damage

CHI63

CHI50

large

Fig. 14. Change of response center distance, r.

(430kN, about 30% shear strength), while shear force resisted by

the independent columns exceeded their capacity.

Fig. 14 shows variations in r-values for both specimens

produced by the different input loadings. It can be observed that
r becomes smaller as load levels increase in both specimens, which
means that the torsional response becomes more dominant in the
inelastic range rather than in the elastic range. This may be explained by the fact that the strength eccentricity of the specimen,
which governs the characteristics of torsional response in the
inelastic range, is so high that the wall did not yield even though
the columns did. As a result, the increased strength eccentricity
in the inelastic range might cause the specimen to exhibit more
torsional behavior, which was observed in both specimens. The r
value of the SRF specimen is slightly larger than that of the RC
specimen because the SRF strengthening reduced damage and prevented the columns from losing their lateral load carrying capacity
as compared to the RC specimen.

3.4. Hysteretic relations

The relationships between lateral displacement and shear force
in the Y1 frame, subjected to ve consecutive earthquake loadings
(TOH12.5 through CHI50-1) are shown in Fig. 16, where the scales
of all the plots are identical. The solid lines in the gure indicate
the calculated shear strength (121.5 kN) of the two independent
columns, while dotted lines show their exural strength
(125.5 kN).
Fig. 17 shows changes in period experienced as the loading
stages progressed. As a system identication scheme for calculation of period, the ARMAX method [18] was used. Accelerations
recorded from the shake table and mass W2, subjected to white
noise inputs were used as input and output values, respectively.
As the load level increased, the stiffness degraded (or the period
was lengthened) and lateral drift grew. This is more pronounced
in the RC specimen than in the SRF specimen. The specimens initial period and stiffness were similar, although the SRF specimen
was slightly stiffer and had a shorter period than the RC specimen.
As described previously, torsion-induced concentrated displacement in the Y1 frame and ultimately caused the collapse of the
RC specimens independent columns. Shear forces shown in this
gure were obtained from load cells installed at the bottom of
the columns in the Y1 frame. During the TOH12.5 and TOH25 loadings, the relationships between lateral drift ratio and shear force
are almost linearly elastic. In the RC specimen, initial crack was observed on the wall panel but not on the independent columns after
TOH25, which was unexpected and might be due to out-of-plane
deformation resulting from the torsional response. SRF sheets
and belts covering the surfaces of the SRF specimen made it impossible to observe crack.
During the ELC37.5 loading input, a diagonal crack that had
developed in the wall panel during the previous loading step grew,

3.3. Shear force distribution

The base shear force was computed by summing the external
forces in the following manner: the masses of W1, W2 and WS
were multiplied by their accelerations, while the shear forces carried by the shear walls were obtained by subtracting the shear
force on the independent columns recorded by the load cells from
the base shear force. Fig. 15(a and b) illustrate the shear forces carried by the columns and the wall in the RC and SRF specimens,
respectively. The ratio of the column shear force to the base shear
force is shown in Fig. 15(c). Here, the shear forces are the values at
the moments when maximum base shear force was attained in
both directions (positive and negative).
From this gure, it can be seen that the shear force carried by
the columns is relatively smaller than that carried by the wall,
and they degrade gradually as the load level increases. The columns of the SRF specimen carried a larger shear force than the
RC columns, which attests to the efcacy of the SRF strengthening
method. It is also noteworthy that the maximum shear force carried by the shear wall is smaller than its calculated shear strength

550

50

Columns
Wall

25
0

SRF
RC

25
50

-550
TOH25

(a)

JMA50

(b)

TAK125

CHI50-2

TOH12.5

ELC37.5

CHI50-1

(c)

Fig. 15. Shear force carried by columns and wall. (a) RC specimen. (b) SRF specimen. (c) Ratio of column shear to base shear.

104

Y. Kim et al. / Engineering Structures 34 (2012) 95110

150

dashed:flexural strength
solid:shear strength

Shear force (KN)

100

SRF
RC

50
0
-50
-100 TOH12.5
TOH25
ELC37.5
JMA50
-150
-0.01 0 0.01 -0.01 0 0.01 -0.02
0
0.02 -0.05
Lateral drift ratio (%)

CHI50-1
0

0.05

Fig. 16. Hysteretic relation between shear force and lateral displacement.

Period (sec.)

0.3
0.25

SR F specim en
R C specim en

0.2
0.15
0.1
Initial

TOH 12.5

TOH 25

ELC 37.5 JMA50

C H I50-1 TAK125

C H I63

Fig. 17. Change in period by loading input.

but no cracks were observed on the independent columns of the RC

specimen. Yielding of the columns longitudinal reinforcement was
observed in both specimen, but no transverse reinforcement
yielded. The hysteretic behaviors of both independent columns in
the RC and SRF specimens are compared in Fig. 19(a), where the
circular and square markers indicate the instants when the longitudinal reinforcement rst yielded in the RC and SRF columns,
respectively. Varying axial load effects on the shear force carried
by the columns can be observed in both compressive and tensile
directions. The longitudinal reinforcements yielded in tension.
After the JMA50 loading, the rst cracking was observed on the
independent columns, although this is not well explained, considering the number of steel bars that had yielded as determined from
strain gauge records. However, there is a possibility that crack was
concentrated at the critical section nearest top of the load cell (Figs.
1b and 18b) which is rough surface and therefore could not be
found by visual observation after ELC37.5 loading. In addition,
the yielding of reinforcement was also recognized from the strain
gauge attached at the bottom of longitudinal reinforcement (same
height with critical section).
The crack patterns that developed on the columns and the wall
during the JMA50 loading is illustrated in Fig. 18. In both
specimens, all the longitudinal reinforcements yielded and the

maximum shear force was recorded. The maximum shear force

of the SRF specimen is higher than that of the RC specimen. The
recorded maximum shear force in the RC specimen was almost
the same as its calculated strength, while the SRF specimen carried
higher loads than its calculated strength (Fig. 16).
Yielding of transverse reinforcements was also observed in both
specimens. Fig. 19(b) shows the relationship between shear force
and lateral drift ratio for each independent column, together with
the instants the transverse reinforcements yielded during the
JMA50 loading. It can be seen from this gure that the difference
of the two specimens maximum column shear force is larger in
compression than in tension, where the transverse reinforcement
yielded. Furthermore, the maximum shear strength developed in
an SRF column (east column) increased even after the transverse
reinforcement yielded, while in the RC column the maximum shear
force is almost same as the shear force recorded at the moment the
transverse reinforcement yielded, meaning that the shear strength
of the RC column did not increase after transverse reinforcement
yielding. From these results, it might be concluded that SRF
strengthening is effective in conning the concrete, which is a
primary role of transverse reinforcement, and therefore SRF is
more effective in improving compressive response than tensile
response.
During the CHI50-1 loading, the stiffness and strength of the RC
specimen signicantly degraded under reversed cyclic loadings,
resulted in collapse at an elapsed time of about 20 s. In the SRF
specimen, on the other hand, the hysteretic relations were stable,
without strength deterioration.
All the transverse reinforcements of the RC columns yielded in
this loading, but only two transverse reinforcements yielded in SRF
columns as recognized from the strain gauge records. The difference in the number of yielded transverse reinforcements between
the two specimens also supports the effectiveness of the SRF
retrot method.

Fig. 18. Crack patterns. (a) Shear wall. (b) Independent columns.

105

Y. Kim et al. / Engineering Structures 34 (2012) 95110

SRF
RC

Shear force (kN)

50

50

-50

-50

SRF
RC

West column
-0.005

Shear force (kN)

100

0
Drif t ratio

East column

0.005

-0.005

(a)
100

SRF
RC

50

50

-50

-50

-100

-100

West column

-0.02

-0.01

0
Drif t ratio

0.01

0.02

-0.02

0
Drif t ratio

0.005

SRF
RC

East column
-0.01
0
0.01
Drif t ratio (%)

0.02

(b)
Fig. 19. Shear force-drift hysteretic relation for each column. (a) ELC37.5 input. (b) JMA50 input.

Fig. 20 shows the columns in both specimens after the CHI50-1

loading input. The front and side views of the collapsed RC specimen are shown in Fig. 21, where the W1 mass landed on the safety
frame (H-shaped steel frame) installed around the 1st oor and
rested inclined toward the Y1 frame after the collapse of the independent columns. These gures indicate that the RC specimen
would have fallen onto the shake table if the safety frame (H-shape
steel frame) had not been installed.

3.5. RC column collapse process

To investigate the collapse process of the east column in the RC
specimen during the CHI50-1 input, time histories and hysteretic
relations of both horizontal and vertical responses for a period of
10 s (from 12 to 22 s) are illustrated in Fig. 22. The responses of
the SRF specimens east column are also plotted and compared
with the RC specimen. Two reference times, 16.7 and 19.77 s,

Fig. 20. Column conditions after CHI50-1 loading input. (a) RC specimen. (b) SRF specimen.

Fig. 21. Collapsed RC specimen. (a) Front view. (b) Side view (from West).

106

Y. Kim et al. / Engineering Structures 34 (2012) 95110

(m/sec 2)

5 (a) Base motion

0
-5

()

3000 (b) Strain of hoop (No.4)

SRF

2000

RC

1000
0
100

(c) Shear force

(kN)

50
0
-50
40

(d) Horizontal displacement

(mm)

20
0
-20
-40
400

(e) Axial force

(kN)

200
0
-200

(mm)

-400
4

(f) Vertical displacement

2
0
-2
12

Shear force (KN)

100

12sec.-16.7sec.

19.77

22

19.77sec.-22sec.

(g)

SRF
RC

50

-50
-40

400
Axial force (kN)

16.7
Time (sec.)
16.7sec.-19.77sec.

-20
0
20
Horizontal displ. (mm)
12sec.-16.7sec.

40-40

-20
0
20
Horizontal displ. (mm)
16.7sec.-19.77sec.

(h)

40-40

-20
0
20
Horizontal displ. (mm)
19.77sec.-22sec.

40

40-40

-20
0
20
Horizontal displ. (mm)
19.77sec.-22sec.

40

40-40

-20
0
20
Horizontal displ. (mm)

40

SRF

200

RC

0
-200
-400
-40

Vertical displ. (mm)

-20
0
20
Horizontal displ. (mm)
12sec.-16.7sec.

40-40

-20
0
20
Horizontal displ. (mm)
16.7sec.-19.77sec.

(i)

2
0
-2
-4

SRF
RC

-6
-40

-20
0
20
Horizontal displ. (mm)

40-40

-20
0
20
Horizontal displ. (mm)

Fig. 22. Responses during CHI50-1 loading input. (a) Base motion time history. (b) Strain time history of hoop (No. 4 in Fig. 2). (c) Shear force time history. (d) Horizontal
displacement time history. (e) Axial force time history. (f) Vertical displacement time history. (g) Horizontal displacement vs. shear force. (h) Horizontal displacement vs. axial
force. (i) Horizontal displacement vs. vertical displacement.

shown by the black and white triangles in Fig. 22, were selected to
divide the responses into three parts.
At 16.7 s, when the highest shear force experienced during the
CHI50-1 input (i.e. 70% of the maximum shear strength developed
during the JMA50 loading) was recorded, considerable stiffness and

strength degradation were initiated and hoop strain at the midheight of the column started to increase (Fig. 22b). The critical
shear cracking associated with the yielding of the hoop might have
caused the residual hoop strains and shear strength decay. Furthermore, the loss of interface shear transfer along the widened shear

107

Y. Kim et al. / Engineering Structures 34 (2012) 95110

cracks which was resulted from yield and fracture of hoop might
have caused the fatal loss of axial load carrying capacity in column.
From the above results, it should be noted that the inelastic hoop
strain, accumulated over cyclic load reversals, might be one of
the main causes of shear and axial failure of RC columns. The horizontal hysteretic relationship of the RC column in Fig. 22g shows
relatively stable behavior until 16.7 s and a remarkable pinching
effect between 16.7 and 19.77 s. In addition, the decay of stiffness
and strength in the RC column during this time zone resulted in the
considerable increase in the lateral drift at 19.77 s compared to
that at 16.7 s. On the other hand, the lateral drift of SRF column
at 19.77 s is only slightly larger than its rst peak drift at 16.7 s,
since the decay of stiffness and strength were very small
(Fig. 22b and g). At 19.77 s, when the mid-height hoop might have
ruptured (Fig. 22b), almost all the lateral stiffness and strength of
the RC column were lost. It can be seen from Fig. 22h that the maximum axial force of RC column observed from 16.7 to 19.77 s is
higher than that of SRF column, which might be due to larger lateral drift of RC column than that of SRF specimen. However, after
19.77 s, axial force in RC column could not exceed that of SRF
one, although lateral drift of RC column was much larger than that
of SRF one. These results indicate that axial failure of RC column
initiated soon after shear failure around 19.77 s at which the lateral
drift ratio was 0.036. That is, the loss of axial load-carrying capacity, which was precipitated by shear failure, induced collapse of the
RC specimen.

Fig. 23 shows the RC column at each reference time, showing

the collapse process, which was captured on video.
The SRF column was stable throughout the entire duration of
the CHI50-1 input, although lateral strength and stiffness both
deteriorated slightly compared to the previous input (JMA50). Both
specimens showed that shear strength in the positive direction
(compression) was higher than that in the negative direction (tension), arising from the variable axial force. The lateral strength difference between compression and tension is more pronounced in
the SRF column rather than in the RC column. This result, as mentioned previously, indicates that the SRF retrot method is more
effective in compression than tension because of its ability to conne the concrete, which becomes more apparent with increased
variable axial loads.
There was no distinguishable difference in the two specimens
vertical responses, both vertical displacement and axial force, until
the 19.77 s reference time (Fig. 22e and f). Relatively large compressive deformation was generated just before 19.77 s and downward deformation drastically increased with the progress of axial
failure in the RC columns, which was followed by collapse of the
whole specimen. Comparing two specimens vertical responses in
this test, it could be concluded that axial collapse cannot be predicted from the vertical responses since the RC column exhibited
no discernable difference in its vertical behaviors compared to
the SRF column until column collapse was initiated. Furthermore,
it is also observed from the time history response (Fig. 22) that

Fig. 23. RC specimen collapse process. (a) 16.7 s; (b) 19.77 s; (c) 20.03 s; (d) 20.3 s; (e) 20.9 s.

No.7 Y ield strain

No.7 Y ield strain

No.6 Y ield strain

No.6 Y ield strain

No.5 Y ield strain

No.5 Y ield strain

No.4 Y ield strain

No.4 Y ield strain

No.3 Y ield strain

No.3 Y ield strain

No.2 Y ield strain

No.2 Y ield strain

No.1 Y ield strain

No.1 Y ield strain

0
2000 4000
Max. strain ( )

(a)

10
15
Time (sec.)

20

0
2000 4000
Max. strain ( )

20
40
Time (sec.)

(b)

Fig. 24. Height-wise distribution of transverse steel strain. (a) RC specimen. (b) SRF specimen.

108

Y. Kim et al. / Engineering Structures 34 (2012) 95110

phase deviation between the two specimens until 19.77 s is significantly smaller vertically than laterally. This might be due to the
fact that there is no neighboring vertical member to sustain axial
load redistributed from the collapsed column since there is just
one span in each direction (only two columns per frame).
Fig. 24 illustrates the height-wise strain distribution recorded
during the CHI50-1 input. As described before, the transverse steel
strain at the mid-height of RC column (No. 4 in Fig. 2) started to increase at 16.7 s with rupture following at 19.77 s when the sliding
of the shear failure plane might have occurred. However, transverse reinforcements at the bottom and top of the column (Nos.
1, 2, 6, and 7 in Fig. 2) did not yield and the strains of the Nos. 3
and 4 hoops slightly exceeded their yield strain. These strain records indicate that shear damage to the RC columns was concentrated at mid-height rather than at the plastic hinge regions at
the bottom and top of the columns. In the SRF column, however,
which was strengthened by the polyester belting, shear damage,
represented by transverse steel strain, was severe at the bottom
and top but small at mid-height. The hoop at mid-height did not
yield and its strain was about 7.76  104 m/m, far below the yield
strain. This result also substantiates the effectiveness of the SRF
retrot methods conning effect and prevention of crack progress.
That is, using the SRF strengthening method changed the failure
mode of RC columns with light transverse reinforcement from
shear failure after exural yield to exural dominant behavior.
3.6. Retrot effectiveness
After removing the collapsed RC specimen from the shake table,
the SRF specimen was subjected to three additional base motions:
TAK125, CHI63 and CHI50-2. Fig. 25 shows the relationship between lateral force and displacement of the Y1 frame in the SRF
specimen for those three loadings.
In the TAK125 input, the maximum strength had decreases
slightly from the previous input stage (CHI50-1), but the hysteretic

relationship showed relatively stable behavior. In the CHI63 input,

strength deterioration and pinching effects can be observed from
the hysteretic relationship. Furthermore, residual deformation
was generated in the positive direction, which might be due to
the characteristics of that particular earthquake load. The gravity
load distribution between the two columns also changed due to
permanent lateral deformation in the positive direction. The gravity load on the east column increased and that on the west column
decreased by about 50 kN as shown in Fig. 26.
In the nal input, CHI50-2, maximum lateral strength degraded
by almost one-third of the maximum strength recorded during the
JMA50 loading cycle. Permanent deformation in the positive direction could not be recovered during this input stage, and the maximum lateral drift exceeded 10% of story height. Nevertheless, the
SRF specimen showed stable behavior against axial collapse
although its members lost almost all their lateral stiffness and
strength.
Fig. 27(a) shows the relationship between axial force and vertical displacement of the SRF columns. A remarkable compressive
deformation was observed from loading CHI63 through CHI50-2
(Fig. 27(b)), and maximum axial deformation in the compressive
direction was 19.57 mm during CHI50-2 a 2.5% axial deformation
ratio. The columns in the Y1 frame of the RC specimen started to
collapse drastically at an axial deformation of 5 mm (axial deformation ratio of 0.063%), while the SRF specimen showed a gradual
increase in compressive deformation of about 4-times to that of RC
column and nally survived throughout all loading stages.
The state of the SRF columns after testing is shown in Fig. 28,
from which it can be seen that severe damage was concentrated
at the ends of the columns. At mid-height, however, the strain
time-history for the SRF column shown in Fig. 29 indicated that
the maximum strain only slightly exceeded the yield strain of its
transverse reinforcement. From these results, it is obvious that
the midsections of the retrotted columns were not damaged
while both ends suffered severe damage.

150
dashed:flexural strength

100

0
-50
-100
TAK125

-150
-0.1

-0.05

0.05

0.1

CHI50-2

CHI63
-0.1

-0.05
0
0.05
Lateral drift ratio (%)

0.1

-0.1

-0.05

0.05

Fig. 25. Hysteretic relations of SRF columns subjected to additional earthquake loads.

Axial force (kN)

Shear force (KN)

solid:shear strength
50

200

200

100

100

-100

-100

-200

-200
0

10

20
30
Time (sec.)

(a)

40

50

10

20
30
Time (sec.)

40

50

(b)

Fig. 26. Axial force carried by east and west columns in SRF specimen. (a) West column. (b) East column.

0.1

109

Y. Kim et al. / Engineering Structures 34 (2012) 95110

Vertical dispalcement(mm)

10
5
0
-5
-10
-15

TAK125
CHI63
CHI50-2

-20
-25

-300 -200 -100 0 100 200 300

Axial f orce (KN)

-300 -200 -100 0 100 200 300

Axial f orce (KN)

(a)
Vertical displ. (mm)

10

TAK125

CHI63

CHI50-2

-10

-20
0

20

40

60

80
Time (sec.)

100

120

140

(b)
Fig. 27. Vertical responses of SRF specimen. (a) Hysteretic relations between vertical displacement and axial force. (b) Vertical displacement of east column.

strain ( )

Fig. 28. SRF specimen after test.

1500

1500

1000

1000

500

500

-500

CHI50-1 TAK125
0

50

CHI63

100
Time (sec.)

CHI50-2
150

(a)

-500
200

CHI50-1 TAK125
0

50

CHI63

100
Time (sec.)

CHI50-2
150

200

(b)

Fig. 29. Strain time histories at mid-height of SRF column. (a) West column. (b) East column.

4. Conclusions
Two identical specimens having several seismic deciencies
including low transverse reinforcement ratios, weak stories and
eccentric plans were tested on a shake table. One specimen was

reinforced using the SRF strengthening method. The following

conclusions can be drawn from the results of the earthquake
simulations.
The rst story plan of the specimens was eccentric, having one
side with independent columns only, while the other side included

110

Y. Kim et al. / Engineering Structures 34 (2012) 95110

a shear wall. Throughout all the input stages, the lateral displacement of the side having only independent columns was much
larger than that of the shear wall side in both specimens. The
column-only side displacement was almost 10 times the shear
wall-side displacement in the 1st story, about twice that of the
2nd story, which had a symmetric plan. In addition, lateral displacement in the RC specimen is larger than that of the retrotted
specimen, where shear strength and stiffness increased slightly by
virtue of the SRF retrot, which may have decreased its defor
mation.
The response center distance (r) showed that torsional response
is more pronounced in the inelastic range than in the elastic range,
which may be due to the increased strength eccentricity that resulted from damage concentration on the independent column
frame. The comparison of r between the two specimens indicated
that the torsional response of the SRF specimen is smaller than that
of the RC specimen. This was also due to the SRF strengthening,
which reduced the damage on the SRF columns.
The shear forces carried by independent columns are relatively
smaller than those carried by the wall and they decrease gradually
with increasing load level as column damage increases. The SRF
columns carried larger shear forces than the RC columns. Furthermore, the maximum shear strength of the SRF columns increased
even after transverse reinforcement yield, while the maximum
shear carried by the RC columns was very near that recorded at
the moment the transverse reinforcement yielded. These results
indicate that the SRF strengthening method is effective in conning
the concrete, a primary role of transverse reinforcement.
During the CHI50-1 input, the SRF columns showed stable hysteretic relations without any considerable damage while the RC
columns experienced severe shear strength deterioration and axial
load failure along with inelastic load reversal and nal collapse.
Vertically, however, comparing the axial force and vertical displacement of the two specimens revealed that axial collapse could
not be predicted from the vertical responses since bare RC columns, showing signs of drastic collapse, exhibited no discernibly
different vertical behaviors than the SRF columns until axial collapse was initiated.
The SRF specimen was subjected to three additional severe
earthquake loads. It was capable of sustaining its gravity load
and survived all loading stages despite considerable lateral
strength deterioration (about one-third of maximum strength), a
high lateral drift ratio (about 10%) and a large compressive axial
deformation ratio (about 2.5%) during the nal loading.
The height-wise strain distribution was recorded from strain
gauges attached to transverse reinforcements in the RC columns.
The strain was highest at mid-height and lowest at both ends. This
condition was reversed by strengthening columns with the SRF
polyester belting, so that the strain was highest at both ends and
lowest at mid-height. This changed the failure mode of the column
from shear failure after exural yield in the RC column to exural
failure showing minor damage at mid-height and concentrated
damage on both ends of the SRF column.

Acknowledgements
This study was carried out as part of the Development of postearthquake performance evaluation using practical accelerometers, research project (Grant No. 12308018, PI: Toshimi Kabeyasawa) under the support of the JSPS and Monbukagakusho. The
dynamic test was conducted at NIED, Tsukuba, with Dr. Nobuyuki
Ogawa and Mr. Atsushi Kato of NIED. The support and cooperation
we received in conducting the tests are gratefully acknowledged.
References
[1] Kim Y, Kabeyasawa T, Kabeyasawa T, Matsumori T. Dynamic collapse analysis
of the six-story full-scale wall-frame building, Proceedings of the 2007
Structures Congress, Long Beach, CA, USA, May 1619, 2007.
[2] Lee H-S, Ko D-W. Seismic response characteristics of high-rise RC wall
buildings having different irregularities in lower stories. Eng Struct
2007;29(11):314967.
[3] Kabeyasawa T, Kabeyasawa T, Matsumori T, Kim Y. Full-Scale Dynamic
Collapse Tests of Three-Story Reinforced Concrete Buildings on Flexible
Foundation at E-Defense, The 14th World Conference on Earthquake
Engineering, Beijing, Paper ID: S15-002, Oct 2008.
[4] Mitchell D, DeVall RH, Saatcioglu M, Simpson R, Tinawi R, Tremblay R. Damage
to concrete structures due to the 1994 Northridge earthquake. Can J Civil Eng
1995;22:36177.
[5] Watanabe H. Behavior of reinforced concrete buildings during the HyogokenNambu earthquake. Cem Concr Compos 1997;19(3):20311.
[6] Sezen H, Whittaker AS, Elwood KJ, Mosalam KM. Performance of reinforced
concrete buildings during the August 17, 1999 Kocaeli, Turkey earthquake, and
seismic design and construction practise in Turkey. Eng Struct
2003;25(1):10314.
[7] Ahmadizadeh M, Shakib H. On the December 26, 2003, southeastern Iran
earthquake in Bam region. Eng Struct 2004;26(8):105570.
[8] Kabeyasawa T, Tasai A, Igarashi S. An economical and efcient method of
strengthening reinforced concrete columns against axial load collapse during
major earthquake. The Third Workshop on Performance-based Engineering on
Reinforced Concrete Building Structures, Seattle, WA. PEER Report 2002/02:
371384.
[9] Kabeyasawa T, Tasai A, Igarashi S. A New Method of Strengthening Reinforced
Concrete Columns against Axial Load Collapse during Major Earthquake.
Proceedings of EASEC-8, CD-ROM, Singapore, 2001.
[10] Kabeyasawa T, Tasai A, Igarashi S. An Economical and Efcient Method of
Preventing Old Reinforced Concrete Buildings from Collapse under Major
Earthquake, Proceedings of 7NCEE, CD-ROM in print, Boston, 2002.
[11] Kabeyasawa T, Tasai A, Igarashi S. Test and Analysis Of Reinforced Concrete
Columns Strengthened With Polyester Sheet, Proceedings of 13th World
Conference on Earthquake Engineering, Vancouver, B.C., Canada, August 16,
2004 (Paper No. 408).
[12] Altin S, Anil , Kara ME. Improving shear capacity of existing RC beams using
external bonding of steel plates. Eng Struct 2005;27(5):78191.
[13] Sheikh SA. Performance of concrete structures retrotted with bre reinforced
polymers. Eng Struct 2002;24(7):86979.
[14] Bo Gao, Christopher K.Y. Leung, Jang-Kyo Kim. Prediction of concrete cover
separation failure for RC beams strengthened with CFRP strips, Eng Struct
2005; 27: 177189.
[15] Bakis CE, Bank LC, Brown VL, Cosenza E, Davalos JF, Lesko JJ, Machida A, Rizkalla SH,
Triantallou TC. Fiber-reinforced polymer composites for construction
State-of-the-art review. J Compos Construct ASCE 2002;6(2):7387.
[16] Buyukozturk O, Hearing B. Failure behaviour of precracked concrete beams
retrotted with FRP. J Compos Construct 1998;2(3):13844.
[17] The Japan Building Disaster Prevention Association, Standard for Seismic
Evaluation of Existing Reinforced Concrete Buildings, The Japan Building
Disaster Prevention Association (in Japanese), 2001.
[18] Safak Erdal. Identication of linear structures using discrete-time lters. J
Struct Eng ASCE 1991;117(10):306485.