Sei sulla pagina 1di 12

Engineering Structures 32 (2010) 18761887

Contents lists available at ScienceDirect


Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Case study: Analytical investigation on the failure of a two-story RC building
damaged during the 2007 Pisco-Chincha earthquake
Oh-Sung Kwon, Eungsoo Kim

Department of Civil, Architectural, and Environmental Engineering, Missouri University of Science and Technology. 1401 N. Pine Street, Rolla MO, 65409, USA
a r t i c l e i n f o
Article history:
Available online 29 December 2009
Keywords:
Pisco-Chincha earthquake
Post-earthquake analysis
Short column effects
Infill masonry wall
Confinement effect
Seismic performance
Reinforced concrete structures
a b s t r a c t
The Pisco-Chincha earthquake hit the western shore of Peru on 15 Aug 2007. More than five hundred
people perished and numerous residential structures made of masonry and adobe collapsed. Many
buildings were severely damaged by settlement due to liquefaction. In general, engineered reinforced
concrete structures experienced less damage compared with structures constructed with traditional
construction practice. This paper presents a case study on a two-story RC building in Ica, Peru which was
severely damaged during the earthquake. Field observations suggest that causes of the structural failure
are: (1) short column effects resulting from masonry infill walls; (2) an overloaded second floor; and
(3) insufficient stirrups in the columns. To test these hypothesized causes of column failure, analytical
models based on measured section dimensions and reinforcement bar configurations from the field
investigation are used. Nonlinear response history analyses are carried out with ground motions recorded
at a seismic station located 500 m from the reference structure. The analysis results show that the partial
infill walls and a lack of shear reinforcement were the leading causes of the column failures. The results
also showthat the overloaded columns do not significantly influence the seismic demand of the reference
structure. Stirrups in the columns mainly affect seismic capacity and have relatively no effect on seismic
demand for the ground motion that the building was subjected to.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
On August 15, 2007, the Pisco-Chincha earthquake, of magni-
tude 8.00.1 struck the coastal regionof central Peru, causing con-
siderable damage and loss of life. The earthquake was attributed
to the highly active source created by the subduction of the Nazca
oceanic plate under the South American continental plate. The
mechanism of the earthquake was complex with two major rup-
tures approximately 50 s apart. The loss of life was estimated at
600 and several hundreds were injured. The earthquake destroyed
over 50,000 buildings and severely damaged at least 20,000 others.
A team of researchers from the Mid-America Earthquake Center in
US, PeruJapan Center for Seismological Investigations and Disas-
ter Mitigation in Peru, and National University of Peru jointly in-
vestigated the resultant damage to building structures and bridges,
road failures, effects of land slide and liquefaction, and the seismic
hazards of the affected area [1]. The majority of structural failures
were in clay and brick masonry structures. However, several rein-
forced concrete (RC) structures also suffered major damage or col-
lapse, often due to soft story effects and lack of vertical continuity.

Corresponding author. Tel.: +1 573 341 4046.


E-mail address: ekwb8@mst.edu (E. Kim).
The lack of ductile detailing was clear and repetitive even in mod-
ern constructions. The field investigation provided a valuable op-
portunity to closely investigate one of the damaged RC structures,
which is the topic of this study.
The objective of this study is to investigate the failure
mechanism of one of the RC structures damaged from the
earthquake. The failure in this study refers to the loss of lateral
load resisting capacity due to the earthquake. The reference
RC structure is a two-story building which housed chemistry
labs and class rooms at the National University of Ica located
approximately 117 km from the epicenter of the earthquake as
shown in Fig. 1. Many residential structures in Ica are constructed
with adobe and masonry and were heavily damaged while most
engineered structures survived the earthquake without significant
damage. However, a few engineered structures, including the
one investigated in this study, suffered irrepairable damage. The
reference building inthis study was of highinterest to investigators
as it was one of a few engineered structures which were severely
damaged in the earthquake. In addition, ground accelerations of
the event were recorded at a seismic station located approximately
500 m from the building. With actual dimensions of structural
elements along with the close-to-actual input ground motion
data, the main causes of the structural failure can be identified
through numerical analyses. The numerical model also allows for
parametric studies to be conducted such that the relative effects of
0141-0296/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2009.12.022
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1877
Fig. 1. Location of the reference structure.
Fig. 2. First floor plan of the reference building.
several structural deficiencies on the seismic performance of the
building can be investigated.
In the following sections, the configuration of the building
and characteristics of recorded ground motions at nearby ground
motion station is introduced. The observed structural damages
from the field investigation and hypotheses on the causes of the
structural damages are introduced. Finally, the analytical model
of the building, the analysis parameters, and the analysis results
are presented followed by summary of key findings from this case
study.
2. Reference building and recorded ground motions
2.1. Configuration of the reference building
The reference building is a two-story reinforced concrete
structure consisting of 12 bays in the eastwest (EW) direction and
3 bays in the northsouth (NS) direction as depicted in Fig. 2. The
building was constructed as two independent modules separated
by slabs and beams at the interface of the modules. A stairwell was
also independently constructed. The height of the both stories is
4.1 m and the spans of each bay are 4.2 m 8.0 m. At the east
and west facades of the building, additional columns, CW2123
and CE2123 were constructed on the first floor to support gravity
loads of story-highinfill walls onthe secondfloor as showninFig. 2.
Fig. 3 illustrates the elevation of the west and north facades of the
building indicated in Fig. 2. Exterior walls facing east and west
include story-high infill walls as shown in Fig. 3(a). Partial infill
walls were constructed between most columns in north and south
facades of the building, Fig. 3(b). Openings for windows are present
inmost of the infill walls inthe northand southfacades while other
walls include openings used as entrances. Damage to the structural
elements is indicated in Fig. 3 which will be discussed in detail
in Section 3. Infill walls are made of clay bricks with thickness of
175 mm. On the second floor, slabs are extended as cantilevers on
which cladding was constructed.
Column dimensions of the building are 350 mm550 mm for
all columns except intermediate columns (350 mm 480 mm)
1878 O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887
Fig. 3. Elevation of west and north face of the building and damage patterns.
350
350
480
550 550
350
(a) Columns. (b) Beam.
Fig. 4. Section dimensions.
in the west and east facades as shown in Fig. 4. Eight #5 (diam-
eter = 15.8 mm) longitudinal reinforcement bars are used for
exterior columns and four #5 bars are used for interior columns.
From field observations of the columns, which experienced a loss
of cover concrete during the earthquake, it was found that an in-
sufficient amount of stirrups were used in the columns. Only one
#3 ( = 9.52 mm) stirrup was used at the end of each column and
smooth wires ( = 5 mm) were used to hold the longitudinal rein-
forcement bars in place. Beams were 350 mm650 mmincluding
a 250 mm slab thickness. Reinforcement bars of beams and slabs
could not be identified as the beams were not damaged during the
earthquake and their interiors were not exposed.
2.2. Recorded ground motions
The accelerometer at the ICA2 station in Fig. 1 was located
on the first floor of a two-story building similar to the reference
structure. Since the accelerometer was installed on the first
floor of the building, it was anticipated that the recorded
ground motion includes vibration components resulting from
soilstructure interaction. With limited information on soil
profiles, it was assumed that soil profiles at the reference building
and at the building housing the accelerometer are similar. Hence,
soilstructure interaction is not explicitly accounted for in the
analysis as it is implicitly considered by applying recorded motion
on the first floor of a similar building.
Fig. 5 shows the time history of the EW component ground
motion recorded at the ICA2 station. The ground motion has
two distinctive peaks at an interval of approximately 50 s. This
envelope of the acceleration time history was observed from all
recorded ground motions at other stations, from which it can
be inferred that the two distinctive peaks are from consecutive
ruptures at the source of the earthquake rather than local site
effects. Elastic response spectra with 2%, 5%, and 10% damping are
presented in Fig. 6. The ground motion has spectral acceleration of
0.70.8g between 0.1 s to 0.6 s of period range.
The earthquake lasted longer than 150 s. Damage likely resulted
from the many large amplitude cycles and the long duration of
the earthquake which were devastating to non-ductile structures.
Several intensity parameters of the ground motion are compared
with those of ground motions recorded from different earthquake
events. Peak values in the time domain, such as PGA, PGV, and PGD,
are the most frequently used parameters to define the intensity
of ground motions. However, these parameters lack many other
important features of ground motions such as number of cycles
and frequency content. The nonlinear response of structures can be
influenced by other ground motion parameters such as frequency
content, duration, number of large amplitude cycles, energy flux,
etc. The following ground motion parameters summarized in
Kramer [2] are evaluated for the ground motions recorded at
station ICA2 after baseline correction.
PGA = max |a(t)| (1)
v
max
/a
max
= PGV/PGA (2)
a
rms
=
_
1
t
r
_
t
r
0
[a(t)]
2
dt (3)
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1879
Fig. 5. Recorded acceleration from station ICA2 (EW component).
Fig. 6. Accelerationdisplacement response spectra of the EW component.
I
a
=

2g
_

0
[a(t)]
2
dt (4)
SED =
_
t
r
0
[v(t)]
2
dt (5)
where a(t), v(t), and t
r
are recorded acceleration, velocity, and
total duration of record, respectively. a
rms
is root-mean-square
of acceleration, v
max
/a
max
is peak velocity to peak acceleration
ratio, I
a
is Arias intensity, and SED is specific energy density. Since
the method adopted for baseline correction significantly affects
the velocity and displacement time history, PGV and PGD are
not compared. The evaluated intensity measures are presented in
Table 1. As a reference, the parameters of a few representative
earthquake ground motions are also presented. For the comparison
with the ground motion record at ICA2 station with PGA of 0.341g,
ground motions with similar PGA (around 0.3g) are selected.
It can be noted from Table 1 that while other parameters do
not show large differences between recorded ground motions,
the Arias intensity and specific energy density of the earthquake
record from ICA2 station are much larger than those from other
earthquake events. The Arias intensity and specific energy density
are greatly affected by earthquake duration in comparison with
other intensity parameters. Fig. 7 compares the energy flux of
the record from ICA2 station with records from other earthquake
events. The 2007 Pisco-Chincha earthquake is characterized by
a much larger energy flux when compared to other earthquake
events with similar PGA due to its prolonged ground motion
duration which may have been the primary cause of buildings with
poor ductility capacity.
3. Observed damage and assumed causes of the damage
The first-story columns of the west module of the structure in
Fig. 2 were heavily damaged by the earthquake. Many columns
Fig. 7. Energy flux comparison of the Pisco-Chincha earthquake with other
earthquake events.
completely lost their capacities to resist gravity loads. Field
investigation of the damaged structure suggest three possible
causes for the structural failure: short column effect, insufficient
stirrups and overloaded columns.
3.1. Short column effects due to infill walls
At the southand northfacades of the building, partial infill walls
were constructed between columns. Most infill walls remained
intact while some of the infill walls developed minor cracks and
crushing at the corners. Fig. 3 depicts the severity of column
damage with three damage indices ranging from DI
1
to DI
3
, which
correspond to minor cracks (DI
1
), loss of cover concrete and
exposure of longitudinal bars (DI
2
), and buckling of longitudinal
bars, fracture of stirrups, and loss of core concrete (DI
3
). These
damage indices qualitatively correspond to usable, repairable, and
irrepairable damage state. From Fig. 3(b), it can be observed that
columns CW16, CW18, CW19, and CW20 suffered more severe
damage (DI
3
) than columns CW15 and CW17. The columns with
less damage have partial infill wall on one side and no wall on
the other side of the columns. From this observation, it can be
presumed that the restraints of infill walls to the columns might
have had an influenced on the observed column damage.
Fig. 8 shows two examples of column failure from the
earthquake. In Fig. 8(a), the infill wall on the left side of the
column is higher than the infill wall on the right side, which lead
to non-symmetric deformation capacity of the column and the
formation of shear cracks in one direction. In Fig. 8(b), due to
the restraints from the infill walls, damage is concentrated at one
end of the column. The damage patterns of these columns clearly
show that the infill walls influenced the column damage. It is well
documented that short columns reduce structural periods, which
in general increase the seismic force demand. In addition, short
columns are subject to non-ductile shear failure rather than ductile
1880 O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887
(a) Column CW18. (b) Column CW16.
Fig. 8. Typical column failure modes of the reference building.
Table 1
Intensity measures of ground motions from the Pisco-Chincha earthquake.
Earthquake Station Component PGA (g) v
max
/a
max
(s) a
rms
(g) I
a
(m/s) SED(m
2
/s)
Pisco-Chincha, 2007 ICA2 NS 0.341 0.190 0.033 3.698 13074
LomaPrieta, 1989 Gilroy Array #2 0 0.367 0.091 0.044 1.197 875
Northridge, 1994 Beverly Hills Vertical 0.327 0.053 0.054 1.349 415
Kobe, 1995 Kakogawa 90 0.345 0.082 0.052 1.687 1625
Kocaeli, 1999 Duzce 180 0.312 0.192 0.051 1.085 7339
flexural failure. The effects of short column on shear capacity and
demand are analytically investigated in Sections 4 and 5.
3.2. Inappropriate stirrups
Several damaged columns experienced loss of cover concrete
and core fromwhich information pertaining to column dimensions
and number of reinforcement bars were extracted. Fig. 9 is a close-
up view of the two damaged columns with exposed stirrups and
longitudinal reinforcements. The field investigation revealed that
only one stirrup with regular deformed bar was used at the end of
the column. For the remaining parts of the columns, smooth wires
with diameters of 5 mm or less were used as stirrups instead of
regular deformed reinforcement bars.
Poor confinement characterized by the usage of smooth wires
stirrups resulted in crushing of the core concrete and subsequently
buckling of the longitudinal reinforcements. The combination
of increased shear force demand from column shortening and
reduced shear capacity frominappropriate stirrups are expected to
be the major contributing factor to the column failure. The effects
of confinement on the seismic demand and shear capacity are
investigated analytically in Section 4.
3.3. Overload on the second floor
Even though the structure consists of two identical modules as
shown in Fig. 2, the columns in the west module were significantly
more damaged than those in the right module. Fig. 10 shows that
columns CW06, and CW07 were severely damaged and crushed
due to the loss of core concrete. A close look at the plans of the
second floor and load carrying system suggests that the columns
in the west module of the building might have been overloaded
in comparison with those in the right module. From the field
investigation, it was found that the span surrounded by columns
CW06, CW07, CW14, and CW13 was heavily loaded with partition
walls. In addition, there was fairly heavy cladding on the canopy of
the span which can be seen in Fig. 10. The west and east facade
of the building also have story-high infill walls on the second
floor as shown in Fig. 3(a). However, the load from the infill walls
is distributed to additional intermediate columns CW21, CW22,
CE22, and CE 23 in Fig. 2. On the other hand, the shaded span in
Fig. 2 has no additional load carrying columns which can share
the load of infill walls on the second floor. It is hypothesized that
the failure of columns CW06 and CW07 may have resulted from
large overload on the second floor. Crushing and shortening of the
columns of this span might have redistributed gravity loads over
other columns CW04 and CW05 resulting in progressive column
failures. The effect of overload is analytically studied in Section 4.
4. Analytical model of the reference building
A 3D analytical model was developed to understand the causes
of the structural failures. The columns and beams are modeled
with fiber-based section elements in Zeus-NL [3]. Infill walls are
modeled as diagonal struts as shown in Fig. 11. In the following
sections, employed assumptions and numerical models of frame
elements and infill walls are presented.
4.1. Reinforced concrete frames
The building consists of two modules separated by a small gap.
Considering the very similar geometry of the two modules, it is
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1881
(a) Column CW05. (b) Column CW16.
Fig. 9. Exposed reinforcement bars from damaged columns.
Fig. 10. Overloads on columns CW06 and CW07.
assumed that the modules have similar fundamental periods and
pounding of the two modules is unlikely andcanbe ignored. Hence,
the west module of the building in Fig. 2 is modeled for response
history analysis. The material properties of the concrete and steel
reinforcements could not be identified fromthe field investigation.
Thus, the strengths of concrete and steel reinforcements are
assumed based on typical material properties commonly used in
Peru. Loaiza and Blondet [4] reported that concrete strength for
building structures in Peru ranges from 21 MPa to 35 MPa and
the steel yield strength is 410 MPa. Based on the study by Loaiza
and Blondet [4], the strength of concrete used in the analysis
is set to 27 MPa and the yield strength of steel at 410 MPa.
Dimensions of column and beam sections were obtained from the
field. The diameters and number of longitudinal and transverse
reinforcements of columns are as shown in Fig. 4. Beam elements
are modeled as T-beams to account for the contribution of slabs on
flexural stiffness and strength. The effective flange width of the T-
beams is determined according to ACI 318-08 [5]. The dimensions
and number of the beamreinforcement bars could not be obtained
from the field investigation as the beams were not damaged
by the earthquake. The beam reinforcements are determined by
designing the T-beam sections against gravity load.
Beam and column elements are modeled with fiber-based
section elements while accounting for material and geometric
nonlinearity. During the analysis, the sectional stressstrain state
is obtained through the integration of the inelastic material
response of the individual fibers describing the section. Eulerian-
based geometric nonlinearity is employed on the element level.
In doing so, the spread of inelasticity along the member length
and across the section depth as well as the effect of large member
deformations are all considered. Since the sectional response is
calculated at each loading step frominelastic material models that
account for stiffness and strength degradation, there is no need
for sweeping assumptions on the momentcurvature relationships
required by other analysis approaches. The concrete is modeled
using a nonlinear hysteretic uniaxial concrete model with constant
confinement based on the model presented by Mander et al. [6].
Steel is modeled with the bilinear elasto-plastic model with
kinematic strain hardening.
4.2. Masonry infill walls
Masonry walls are modeled with compression-only diagonal
struts as illustrated in Fig. 12. The strut strength can be determined
considering the three possible failure modes of infill walls: namely
compression failure of diagonal strut, sliding shear failure of the
masonry along horizontal mortar, anddiagonal tensile cracking [7].
Among these failure modes, the first and the second failure modes
are the most common failure modes for infill walls. In this study,
the shear strengths of the infill walls corresponding to the first
and the second failure modes are calculated for each infill wall.
The minimumof the shear strengths fromthe two failure modes is
consideredas the ultimate strength. The modeling of the infill walls
follows the method presented in Mostafaei and Kabeyasawa [8]
and is summarized below.
4.2.1. Compression failure of diagonal struts
Compressive strength of the masonry prism is a key parameter
inthe modeling of infill walls. Paulay andPriestley [7] proposedthe
following equation for the estimation of the compressive strength
of the masonry prism, f

m
.
f

m
= f
y
=
_
f

cb
_
f

tb
+f

j
__
/
_
U
u
_
f

tb
+f

cb
__
(6)
where is j/4.1h, j is thickness of mortar, h is height of the
masonry unit, U
u
is stress nonuniformity coefficient (U
u
= 1.5
following Hilsdorf [9]), f

tb
is tension strength of brick (= 0.1f

cb
), f

cb
is compression strength of brick, and f

j
is mortar compression
strength. Eq. (6) requires material parameters of brick and mortar,
1882 O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887
Fig. 11. Analytical model for the frame-infill wall system.
Fig. 12. Infill masonry walls and the equivalent diagonal strut parameters.
which could not be identified from the field investigation. Loaiza
and Blondet [4] reported that the strength of masonry prism of
typical masonry walls in Peru ranges from13 to 16 MN/m
2
. In this
study, the median value of this range, 14.5 MN/m
2
, is used for the
compression masonry prism strength.
Compression failure of infill walls occurs due to the compres-
sion failure of the equivalent diagonal strut. The horizontal com-
ponent of the diagonal strut capacity (shear force) is,
V
c
= zt
w
f

m
cos (7)
where f

m
is masonry compressionstrengthfor ungroutedclay brick
(14.5 MPa [4]), and z is equivalent strut width based on FEMA
306 [10].
4.2.2. Sliding shear failure mode
Shear strength for the sliding shear failure mode,
f
, can be
defined by the MohrCoulomb failure criteria:

f
=
o
+
N
(8)
where
o
is cohesive capacity of the mortar beds, is sliding fric-
tion coefficient along the bed joint, and
N
is vertical compression
stress in the infill wall. In terms of force,
V
f
=
o
t
w
l
m
+N (9)
where t
w
is infill wall thickness, l
m
is length of infill panel, and N
is vertical load in infill walls. The infill walls in the reference frame
are not load bearing walls. Thus N can be approximated as vertical
component of the diagonal compression force, R
c
sin , where R
c
is
diagonal compression force, Fig. 12. Eq. (9) can be rewritten as
R
c
cos =
o
t
w
l
m
+R
c
sin
or,
V
f
=

o
t
w
l
m
1 tan
. (10)
Typical ranges for
o
and are 0.1
o
1.5 MPa and 0.3
1.2 [8]. For analysis purposes,
o
= 0.04 f

m
= 0.04 (14.5) =
0.58 may be assumed [7]. Based on previously conducted experi-
mental work, Chen [11] reported that the frictional coefficient, ,
can be defined as:
= 0.654 +0.00525 f

j
(11)
where f

j
is mortar strength in MPa. Assuming f

j
= 4.9 MPa, fric-
tional coefficient is calculated as = 0.68. Hence, Eq. (10) yields
the following:
V
f
=
0.58t
w
l
m
1 0.68 tan
(12)
where V
f
, t
w
, and l
m
are expressed in N, mm, and mm, respectively.
4.2.3. Hysteretic behavior of a diagonal strut element
The shear strengths obtained from the sliding shear failure and
the diagonal compression failure may not exceed 0.83 MPa as
recommended in ACI 530-05/ASCE 5-05 [12].
V
max
/t
w
l
m
= 0.83 MPa. (13)
The two diagonal struts of infill walls provide resistance against
lateral load. In this study, it is assumed that forcedisplacement
relationship of each diagonal strut follows a tri-linear curve under
compression and zero tension strength as shown in Fig. 13. The
hysteretic behavior of diagonal struts is modeled with lumped
springs at the end of the strut and the strut is modeled with
rigid truss element. The tri-linear displacementforce relationship
consists of yield shear force (V
y
), maximum shear force (V
m
),
yield displacement (U
y
) and maximum displacement at peak force
(U
m
). The maximum displacement at the maximum lateral force is
estimated by Eq. (14) following Madan et al. [13].
U
m
=

m
d
m
cos
(14)
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1883
V
m
U
m
U
y
K
0
K
0

V
y
Fig. 13. Hysteretic behavior of diagonal strut elements.
where

m
is the masonry compression strain at the maximum
compression stress and assumed as 0.0018 and d
m
is the diagonal
strut length. The maximumdrift limitation of 0.8% is applied to the
U
m
/h
m
ratio, based on the results from [11,14]. The initial stiffness
K
o
is determined as follows according to Madan et al. [13]:
K
0
= 2(V
m
/U
m
) (15)
where V
m
is the maximum strength determined from Eqs. (11)
(13). The stiffness ratio, , is assumed as 0.2 and U
y
and V
y
, are
estimated based on V
m
, K
o
, and, .
4.3. Damping
The commonly used damping ratio for elastic analysis of
reinforced concrete structures is 5% of the critical damping.
When a structure behaves in a nonlinear range, the hysteretic
behavior of materials dissipates energy. Thus, the application of
the same amount of damping to elastic and inelastic systems is
not feasible [15]. Considering that the reference structure showed
significant excursion into the inelastic range, 2% of the critical
damping is assumed in the analytical model.
5. Analysis cases and result
The nonlinear time history analysis is adopted to evaluate the
seismic performance of the structure. There are several other
methods for the seismic performance evaluation. With the recent
trends on the adoption of the performance-based seismic design
and evaluation, the displacement-based approach is widely used
in recent studies instead of the traditional force-based evaluation.
The capacity spectrum methods in ATC-40 [16] and Chopra and
Goel [17] can provide approximate seismic performance utilizing
the structural resistance and seismic demand represented in the
accelerationdisplacement response spectra format. But to apply
the capacity spectrummethod, a structures pushover curve should
be able to represent the global forcedeformation relationship.
If a structure has very stiff elements, such as infill walls or
braces, or if the structure develop brittle failure due to shear,
the performance evaluation using capacity spectrum can be very
difficult. As the reference structure have many masonry walls
and the columns failed with brittle shear failure, the force-
based evaluation combined with nonlinear time history analysis is
adopted rather than displacement-based evaluation. The analysis
cases and results are presented in the following sections.
5.1. Analysis cases
The developed numerical model is analysed with several dif-
ferent parameters to determine the causes of the structural fail-
ures introduced in Section 3. Table 2 summarizes the analysis cases
considered in this study. Each row corresponds to different analy-
sis case andeachcolumnrepresents the controlledparameter. Case
1 is the as-built condition of the reference building which includes
masonry infill wall, columns with insufficient stirrups, and over-
loaded columns. The actual gravity load on the second floor in the
overloaded span shown in Fig. 2 is included in the model. The verti-
cal ground motion in addition to two components of the horizontal
ground motion is also included in the analysis case. The analytical
models for cases 2 through 5 are based on the model for case 1. Dif-
ferent parameters in cases 2 to 5 are controlled to investigate the
effects of these parameters. The effects of masonry infill wall on
seismic force and displacement demands are investigated in case
2 in which the first hypothesis on the causes of structural failure is
tested.
Case 3 is concerned with the effect of confinement on seismic
response and structural capacity. The stirrups in the columns have
two major functionalities, providing confinement which increases
concrete strength and ductility, and increasing shear strength.
In case 1, the confinement effect is ignored by employing a
confinement factor of 1.0 since the columns in the building do not
have structural grade stirrups. The contribution of 5 mm smooth
wire to the confinement effect is ignored. In case 3, it is assumed
that the columns meet the minimum requirement for stirrups for
seismic design according to ACI 318-08 [5] and the corresponding
confinement factors are determined by the method presented by
Mander et al. [6]. Based on ACI 318-08 [5], the column sections
are assumed to have a #4 ( = 12.7 mm) rectilinear hoop with
one crosstie arranged with a spacing of 100 mm. The confinement
factor for the column sections C1, C2, and C3 in Fig. 4 with
minimum required stirrups are 1.43, 1.32, and 1.32, respectively.
The effects of stirrups on seismic demand and shear strength are
compared in Section 5.3.
Case 4 evaluates the effect of overloading on the second floor. In
case 4, the gravity load from heavy cladding on the canopy of the
overloaded span shown in Fig. 2 is not considered which reduces
axial force on the columns CW06 and CW07. Since the fiber-
based section elements adopted to model columns and beams
estimate stiffness and strength from stressstrain relationship of
individual fiber. Therefore, the effects of different axial forces
on axial capacity of columns are automatically reflected during
response history analysis.
In case 5, the vertical ground motion is not included to
investigate the effect of vertical ground motion on seismic
response. The PGA of the vertical ground motion of the site is
0.19g, which may increase axial force on columns resulting in the
crushing of core concrete or resulting in low moment strengths in
columns. The vertical ground motion also can decrease the axial
force on columns which can affect the shear capacity of a concrete
section. The effects of the variation of axial force due to vertical
ground motion on the structural response are investigated through
a comparison of cases 1 and 5.
Response history analyses for the above five analysis cases
are carried out with the recorded ground motions at station
ICA2. For each analysis case, story displacement demand, axial
force demand, shear force demand, and shear force capacity
are evaluated at every time step. The employed elements for
beams and columns can simulate flexural deformation and flexural
failure of reinforced concrete elements with a high level of
confidence. But the load redistribution due to shear failure and
shear deformation cannot be represented with the model. Even
with more sophisticated finite element analysis approaches for RC
structures, the prediction of shear failure and post-failure behavior
of RC structures is still very challenging. Hence, in this study
the shear failure is evaluated through comparison of shear force
demand and shear force capacity at each time step. Analysis results
are presented in the following sections.
1884 O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887
Fig. 14. First-story relative displacement.
Table 2
Analysis cases.
Masonry infill wall Confinement effects Overload Vertical ground motion
Case 1 Yes No Yes Yes
Case 2 No No Yes Yes
Case 3 Yes Yes Yes Yes
Case 4 Yes No No Yes
Case 5 Yes No Yes No
Notes: Yes indicates that the effect of the parameter is considered, while analysis cases with No do not include the effect. Case 1 is as-is condition. Other cases are to see
the effects of several parameters on seismic performance of the reference building.
5.2. Relative story displacements
The maximum story displacements from the five analysis
cases are summarized in Table 3. Time history of the first-story
displacement from cases 1 and 2 are presented in Fig. 14. From
Table 3, it can be seen that cases 1, 3, 4, and 5 have very similar
maximumstory displacements. On the other hand, in analysis case
2, which does not include infill walls, the story displacement is
almost five times greater than the story displacements observed
in the other four cases. The structural models for cases 1 and 2
have fundamental periods of 0.47 s and 0.64 s, which corresponds
to the differences in stiffness with a ratio of 1.8. Hence, the large
differences in the maximum story displacement between case
2 and other cases originate from the large stiffness of the infill
walls. In case 3, the confinement factors for the minimumrequired
stirrups, which increases strength and ductility of concrete, are
included in the concrete material model. The analysis results show
that the maximum displacement from cases 1 and 3 are very
similar. This implies that the concrete does not develop large strain
to reach the increased strength due to the applied confinement
factor. This observationshows that the confinement factor does not
greatly contribute to the seismic demand with the levels of applied
ground motion used in this study.
Shear deformation and shear failure are not represented in
the analytical model. Thus, it is expected that the maximum
displacements may not be accurate. If columns fail due to shear,
the inter-story displacement fromthese cases may be greater than
the displacements presented in Table 3. Therefore, rather than
using displacements as failure criteria, the failure of the structure
is evaluated through comparison of shear strength capacity and
demand at each time step as presented in Section 5.3.
5.3. Comparison of shear force capacity and demand
The shear strength of RC members can be evaluated using
several formulas as suggested in ACI 318-08 [5]. The most
simplified code formula considers the contribution of concrete and
transverse reinforcement bars to the shear strength of the section.
More sophisticated equations consider the effect of span-to-depth
ratio through the applied shear force and moment demands.
Table 3
Comparison of the first-story displacement.
Case 1 Case 2 Case 3 Case 4 Case 5
Maximum (mm) 19.68 99.03 19.52 19.31 19.61
Minimum (mm) 15.61 65.23 15.51 15.33 15.69
Contribution of axial force demand to the shear force capacity is
also considered in the code formula in which the shear strength
capacity of concrete increases with the compressive axial force. In
this study, the formula that includes the effect of axial force on
concrete shear strengthis adoptedinorder toevaluate shear failure
of columns.
The failure of columns is evaluated through comparison of the
shear force capacity and demand at each time step. For analysis
cases 1, 2, 4, and 5, the contribution of transverse reinforcement
is not included in the shear capacity as the structure does not
have structural grade shear reinforcement. For analysis case 3, the
contributionof minimumrequiredshear reinforcement is included
in the shear strength evaluation. For all analysis cases, the effect
of axial load to shear strength of concrete, V
c
, is considered using
Eq. (16).
V
c
(t) = 0.17
_
1 +
N
u
(t)
14A
g
_
_
f

c
b
w
d (16)
where, A
g
is gross sectional area in mm
2
, N
u
is axial force in N
with positive sign for compression, f

c
is the concrete compressive
strength in MPa, b
w
is the web width in mm, and d is the effective
depth in mm. The variable is 1.0 for normal weight concrete and
0.75 for light weight concrete. The shear capacity contribution of
concrete for member subjected to axial tension is calculated as
V
c
(t) = 0.17
_
1 +0.29
N
u
(t)
A
g
_
_
f

c
b
w
d. (17)
Fig. 15 presents the temporal variation of shear force demand
and capacity of column CW19 from analysis case 1. The shear
force demand is a resultant force of two orthogonal shear forces
at the ends of a column. The shear force capacity varies with
time as the axial load on columns varies with time. It can be
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1885
300
250
200
150
100
S
h
e
a
r

f
o
r
c
e

o
r

c
a
p
a
c
i
t
y

(
k
N
)
Time (sec)
50
0
0 10 20 30 40 50 60
Fig. 15. Time history of shear force demand and capacity for CW19 in case 1.
Time (sec)
S
h
e
a
r

d
e
m
a
n
d

/

c
a
p
a
c
i
t
y
0 10 20 30 40 50 60
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Fig. 16. Time history of shear force demand to capacity ratio for CW19 in case 1.
Columns with infill walls
Fig. 17. Shear force demand-to-capacity ratio comparison of analysis cases 1 and 2.
seen from Fig. 15 that the variation of shear force capacity with
time is not significant in comparison with the variation of shear
force demand. Nonetheless, as both the shear force demand and
capacity of columns vary with time, the ratio of the shear force
demand-to-capacity, V
u
(t)/V
n
(t), is evaluated at each time step.
In Fig. 16 it can be seen that column CW19 in analysis case 1
may have failed as the demand-to-capacity ratio is greater than
1. However, as the material properties of the reference building is
assumed in this study, the absolute magnitude of shear demand to
the capacity ratio may not be used as a criterion to define failure
of a column. Since the objective of this study is to understand the
relative effects of various parameters to the failure of the building,
the relative demand-to-capacity ratios fromthe five analysis cases
are compared with each other.
Fig. 17 compares the maximumshear demand-to-capacity ratio
from analysis cases 1 and 2. From the figure, it can be seen that
the shear demand-to-capacity ratios are relatively uniform in case
2 while the ratios greatly vary in case 1. More specifically, the
columns adjacent to infill walls, marked with dashed rectangular
boxes in the figure, have noticeably higher demand-to-capacity
ratios than other columns. From this comparison, it can be seen
that the infill walls concentrate shear force demand to columns
in contact with the infill walls. But the infill walls reduce shear
force demand to other columns. A comparison of the demand-to-
capacity ratios of columns from analysis cases 1 and 3 in Fig. 18
shows that analysis case 3 has significantly lower shear demand-
to-capacity ratio than case 1 due to increased shear capacity in
case 3. Table 3 shows that case 3 has very similar displacement
demand to analysis case 1, which indicates that the shear demand
to columns in case 1 and case 3 are very similar. Hence, the
low demand-to-capacity ratio in Fig. 18 implies that the shear
capacities of the columns in case 3 are large due to the contribution
of shear force reinforcement. This result suggests that the columns
in the analysed structure may not have failed even if the minimum
required amount of shear reinforcement is used in the columns.
Figs. 19 and 20 compare the demand-to-capacity ratios from
analysis case 1 with analysis cases 4 and 5, respectively. In both
cases, the ratios are very similar to case 1. Thus even though there
was large overload on one of the spans and a vertical ground
motion with PGA of 0.2g, the effect of axial load to the shear
capacity of the studied reinforced buildings is not significant.
The comparisons of the shear demand-to-capacity ratios con-
sidering several parameters show that the infill walls concentrate
shear force demand to adjacent columns, the stirrups do not affect
1886 O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887
Fig. 18. Shear force demand-to-capacity ratio comparison of analysis cases 1 and 3.
Fig. 19. Shear force demand-to-capacity ratio comparison of analysis cases 1 and 4.
Fig. 20. Shear force demand-to-capacity ratio comparison of analysis cases 1 and 5.
seismic demand through confinement effects but significantly af-
fect the shear capacity of columns, and overload and the vertical
ground motion may not have significantly influenced the failure of
the columns.
6. Conclusion
To understand the main causes of the structural failure of a two-
story RC building damaged by the 2007 Pisco-Chincha earthquake,
numerical models are developed based on the measured section
dimensions andreinforcement bar configurations. Groundmotions
recorded at a station located 500 mfromthe reference building site
are used as input ground motions. Effects of masonry infill wall,
stirrups, overloading, and vertical ground motions on structural
failure are examined in five analysis cases. The following is a
summary of the findings from this study.
The first-story displacement is not significantly affected by
confinement effect, overloading on the second story, and
vertical ground motion. The main factor contributing to story
displacement are the masonry infill walls which increase the
lateral stiffness of the reference building.
The infill walls, especially partial story-height infill walls,
concentrate shear forces to columns in contact with the infill
wall while decreasing shear force demands to other columns.
When stirrups of columns are included in the analytical model,
the stirrups do not affect seismic response throughconfinement
effects. This is mainly due to the fact that the ground motion
level recorded at ICA2 station was not large enough to induce
increased concrete strength and ductility by confinement
factor. The stirrups however significantly decrease the shear
demand-to-capacity ratio.
The overload and vertical ground motions do not affect shear
force demand and capacity for the studied structure.
Some of the above findings are very specific to the studied
building and the input ground motion. For example, if a structures
failure is dominated by flexural failure, and if the ground motion
intensity is large enough, the increase in strength and ductility
due to confinement effects may have an effect on seismic
O.-S. Kwon, E. Kim / Engineering Structures 32 (2010) 18761887 1887
demand. This case study identifies main causes of the structural
failure of a two-story RC building damaged during the 2007
Pisco-Chincha earthquake. From the failure investigation of the
reference structure, it can be learned that the building should
not have experienced column failures if minimum required shear
reinforcement was used in the columns.
Acknowledgements
The field investigation after 2007 Pisco-Chincha earthquake
was supported by Mid-America Earthquake Center. The MAE
Center is an Engineering Research Center funded by the National
Science Foundation under cooperative agreement reference EEC
97-01785.
References
[1] Elnashai AS, Alva JH, Pineda O, Kwon O, Moran-Lanze L, Huaco G, Pluta G.
The Pisco-Chincha earthquake of August 15, 2007. Mid-America Earthquake
Center; 2008.
[2] Kramer SL. Geotechnical earthquake engineering. Englewood Cliffs (NJ):
Prentice Hall; 1996.
[3] Elnashai AS, Papanikolaou V, Lee DH. Zeus-NLA system for inelastic analysis
of structures. 2002.
[4] Loaiza C, Blondet M. World housing encyclopedia report. Country: Peru
Confined masonry houses. Earthquake Engineering Research Institute; 2002
[Report #51].
[5] ACI 318-08. Building code requirements for structural concrete and commen-
tary. Detroit (Michigan): American Concrete Institute; 2008.
[6] Mander JB, Priestley MJN, Park R. Theoretical stressstrain model for confined
concrete. J Struct Eng ASCE 1988;114(8):180426.
[7] Paulay T, Priestley MJN. Seismic design of reinforced concrete and masonry
buildings. John Wiley&Sons, Inc; 1992.
[8] Mostafaei H, Kabeyasawa T. Effect of infill masonry walls on the seismic
response of reinforced concrete buildings subjected to the Bam earthquake
strong motion: A case study of Bam telephone center. Earthquake Research
Institute, University of Tokyo; 2004.
[9] Hilsdorf HK. An investigation into the failure mechanism of brick masonry
under axial compression in designing. Eng Construct Masonry Product 1969;
3441.
[10] FEMA-306. Evaluation of earthquake damaged concrete and masonry wall
buildingsBasic procedures manual. Washington (DC): Federal Emergency
Management Agency; 1999.
[11] Chen YH. Seismic evaluation of RC buildings infilled with brick walls. Ph.D.
thesis. Tainan (Taiwan): National Cheng-Kung University; 2003 [in Chinese].
[12] ACI-530-05/ASCE 5-05. Building code requirements for masonry structures.
2005.
[13] Madan A, Reinhorn AM, Mander JB, Valles RE. Modeling of masonry infill
panels for structural analysis. J Struct Eng ASCE 1997;123(10):1295302.
[14] Mehrabi AB, Shing PB, Schuller MP, Noland JL. Experimental evaluation of
masonry infilled RC frames. J Struct Eng ASCE 1996;122(3):22837.
[15] Kwon O, Elnashai AS. The effect of material and ground motion uncertainty
on the seismic vulnerability curves of a RC structure. Eng Struct 2006;28(2):
289303.
[16] Applied Technology Council (ATC). Seismic evaluation and retrofit of concrete
buildings. ATC-40 report. 1996.
[17] Chopra AK, Goel RK. Capacity-demand-diagram methods for estimating
seismic deformation of inelastic structures. Report PEER-1999/02. 1999.

Potrebbero piacerti anche