Engineering Structures 73 (2014) 13–23

Contents lists available at ScienceDirect

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

Effect of axial loads in the seismic behavior of reinforced concrete walls

with unconfined wall boundaries
C. Alarcon, M.A. Hube ⇑, J.C. de la Llera
Department of Structural and Geotechnical Engineering, Pontificia Universidad Católica de Chile, and National Research Center for Integrated Natural Disaster Management
CONICYT/FONDAP/15110017, Vicuña Mackenna 4860, Santiago, Chile

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

Article history: About 2% of Reinforced Concrete (RC) buildings taller than nine stories suffered serious damage in
Received 21 August 2013 structural walls during the 2010 Chile earthquake. The observed damage involved mostly crushing of
Revised 18 February 2014 concrete, buckling of vertical reinforcement, and opening of the horizontal reinforcement. This damage
Accepted 22 April 2014
is attributed to poor concrete confinement in the web and boundaries, inadequate horizontal reinforce-
Available online 22 May 2014
ment detailing, and high axial loads. This research aims to reproduce the observed damage and evaluate
the influence of axial loads in the seismic behavior of RC walls with unconfined boundaries. To achieve
Keywords:
these objectives, three identical wall specimens were tested. The wall specimens were designed with
Reinforced concrete
Axial load
characteristic wall detailing obtained from data of five damaged buildings. These wall specimens were
Wall tested under equal lateral displacement cycles and subjected to different axial load ratios. The
Experiment flexural-compressive failure mode exhibited by damaged walls during the earthquake was reproduced
Reinforcement buckling in these tests. Experimental results indicate that high axial load has a significant effect on the seismic
Seismic behavior performance and failure mode of RC walls. Indeed, it triggers a dangerous brittle concrete crushing failure
Chile which occurs immediately after spalling of the concrete cover.
Failure Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction buildings is about 3% in each direction [2,3]. The good building

On February 27th, 2010, the central south region of Chile was
struck by an Mw = 8.8 earthquake [1], one of the strongest ever
measured. More than 80,000 residences were destroyed and more
than 100,000 suffered substantial damage caused by the
earthquake and tsunami. The general performance of tall
Reinforced Concrete (RC) wall buildings with 9 or more stories
was acceptable; only about 2% of the newer building inventory
presented severe damage [2,3] and just one of these buildings
collapsed. This damage was most likely attributed to poor concrete
confinement, inadequate horizontal reinforcement detailing, high
axial loads, and low wall thickness [2,4,5]. Closed inspection to
wall boundaries revealed lack of special boundary reinforcement,
relatively large spacing of the horizontal web reinforcement, as
well as 90° hooks in the horizontal reinforcement, which are
inadequate for seismic detailing of RC walls [4].
Most of Chilean residential buildings rely on a structural system
with a large number of RC walls to resist gravity and lateral loads.
The ratio of wall cross sectional area to floor plan area in these
⇑ Corresponding author. Tel.: +56 2 2354 4207; fax: +56 2 2354 4243.
E-mail address: mhube@ing.puc.cl (M.A. Hube).
performance in the 1985 Chile earthquake was attributed to the
high stiffness and strength provided by this structural system
and it was concluded that confinement in RC walls was not
required in such buildings [6]. However, after 1985, Chilean con-
struction practice evolved, transforming these buildings into taller
structures with thinner walls, which leads to an increase in axial
loads and stresses. In many cases, buildings with 20 stories were
built with 150–200 mm wall thicknesses and axial load ratios—
ALR = N=fc0 Ag , where N is the axial load, fc0 the concrete compressive
strength, and Ag the gross cross section of the wall—that could
range from 0.20 to 0.50 [3]. The building survey performed by
Jünemann et al. [2] indicates that most of damaged buildings in
2010 were mainly new structures constructed after year 2000
and that high axial load was a relevant factor that may have trig-
gered the observed brittle damage.
In design, RC walls are intended to develop a ductile flexural
behavior consistent with the strength reduction factor R, and
hence, brittle modes of failure should be avoided. Hidalgo et al.
[7] tested 26 squat walls in order to study the shear failure mode
in Chilean RC walls. Because shear strength was the goal, these
walls were tested conservatively without axial load to provide a
lower bound of the shear strength, since ACI 318 [8] neglects this

http://dx.doi.org/10.1016/j.engstruct.2014.04.047
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
14 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23
effect. The effect of high axial loads in the flexural behavior of RC
walls was studied experimentally by Zhang and Wang [9] and Su
and Wong [10]. They concluded that the axial load affects the
cracking pattern, failure mode, and ductility of the walls. These
conclusions were obtained by testing RC walls with transverse
boundary reinforcement, which increases ductility by providing
concrete confinement, which is not the case of Chilean walls. Dur-
ing a compressive failure of a well confined RC wall, the axial
strength is not reduced after spalling of the cover concrete; the
deformation capacity and strength of the concrete core increases
due to confinement, and the inelasticity may be distributed along
the wall height. However, if confinement is poor, damage in gen-
eral will not be redistributed along the boundary element and
may localize in a reduced region of the wall characterized by a brit-
tle flexural-compressive failure, induced by crushing of concrete
and buckling of vertical reinforcement bars as it was observed in
some walls of damaged Chilean buildings (Fig. 1). This research
aims to reproduce the damage observed in walls during the 2010
Earthquake and to experimentally evaluate the cyclic behavior of
RC walls subjected to high ALRs. Findings from this project are
intended to be considered in future RC walls design provisions in
Chile in order to avoid the brittle failure mode observed in 2010.
More details of this research project are available elsewhere [11];
experimental data is open to the community and will be available
in the NEES Project Warehouse.
2. Survey of damaged buildings
A survey of critical walls of damaged RC buildings was
conducted in this study to obtain representative characteristics
of damaged walls. These characteristics were used to define a
Fig. 1. Typical flexural-compressive failure of RC walls during 2010 Chile
Earthquake involving crushing of concrete, vertical reinforcement buckling, and
horizontal reinforcement opening.
prototype wall for conducting the experimental campaign pre-
sented herein. General characteristics of five seriously damaged
RC buildings and relevant properties of their walls were consid-
ered. Mostly, rectangular walls were included in this survey
because their behavior is much simpler to interpret and hence
experiments were limited to such sections as a first approximation
to this complex problem; rectangular walls represent 32% of the
total walls of the selected buildings.
Five damaged buildings, whose number of stories ranged from
13 to 20, were considered in this survey. Four of them were located
in the city of Concepción (CM, AH, PR and AA); whose characteris-
tics were obtained from the survey performed by Westenenk et al.
[12], and one in Santiago (EM). The specified concrete strength is
fc0 = 20 MPa for PR and EM buildings, and fc0 = 25 MPa for CM, AH,
and AA buildings. The wall characteristics considered in the survey
were: wall thickness, M/Vlw ratio (where M is the moment at the
base of the wall, V the shear load and lw the wall length), ALR,
and reinforcement ratios. For this purpose, 27, 20, 14, 22 and 4 crit-
ical walls of the first two stories and basement–where damage was
usually concentrated–were considered for CM, AH, PR, AA and EM
buildings, respectively.
The wall thicknesses of the four buildings located in Concepción
range between 150 and 200 mm, and between 170 and 250 mm for
the EM building. The vertical loads and the M/Vlw ratios were
obtained from finite element models of the buildings using ETABS
[13]. The moment (M) and the shear (V) of walls corresponds to the
seismic demand obtained from a modal response spectrum analy-
sis according to the Chilean Code [14]. The M/Vlw ratio is an impor-
tant property for the wall behavior, and if this ratio is small, the
wall is considered squat and probably will exhibit a shear mode
of failure [7]. The average of the mean M/Vlw ratios of the critical
walls in the damaged buildings is 2.02, which means that these
walls cannot be considered squat, and hence flexural behavior is
relevant. However, some of the selected walls in the buildings con-
sidered were squat with M/Vlw ratios less than 0.5.
The ALRs of the surveyed walls of the five damaged buildings
for gravitational load, including dead load and 25% of live load
(D + 0.25L) are shown in Fig. 2(a) where the mean and standard
deviation of the average ALRs for the five buildings is 0.18, and
0.10, respectively. The axial load in walls induced by an earthquake
is estimated using modal response spectrum analysis according to
NCh433 [14]. Fig. 2(b) shows the ALRs of walls of damaged build-
ings under gravitational plus earthquake loads (D + 0.25L + E),
where the earthquake loads have been divided by the strength
reduction factor. The mean and standard deviation of the average
ALRs for the five buildings is 0.27 and 0.09, respectively. The mean
average ALR for these loads is 50% larger than the value for gravi-
tational loads. Also shown in Fig. 2(b) is the fact that about 10% of
the walls are subjected to ALRs higher than 0.35, which is the limit
imposed after the 2010 Chile Earthquake [15] for ultimate axial
loads. Also, only 20% of the walls exhibit ALRs lower than 0.15
and the rest 70% of the walls exhibit ALRs between 0.15 and
0.35. This latter range of ALRs is considered for the experimental
program of this research. However, if the load combination
1.2D + 1.0L + 1.4E is considered in this analysis according to
NCh3171 [16], the mean of the average ALR of the five buildings
increases to 0.38 and the percentage of walls with ALR higher than
0.35 increases to 37%.
The actual axial load demand in RC walls during the 2010 earth-
quake may have exceeded the one estimated from the design code.
A comparison between the elastic response spectra, from the two
ground motions recorded closest to the four buildings located in
Concepción, and the elastic design spectra is shown in Fig. 3. The
ordinates of elastic design spectra (Sae) and response spectra at
the fundamental period of the buildings are summarized in Table 1.
The fundamental period of the buildings and the soil type, were
 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23 15

0.6 0.6

0.5 0.5

ALR (D+0.25L+E)
ALR (D+0.25L)
0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0 0
CM AH PR AA EM CM AH PR AA EM

(a) (b)
Fig. 2. ALR in RC walls for (a) D + 0.25L and (b) D + 0.25L + E . ALR increase in average 50% when earthquake loads are added to the gravity loads.

25 the selected walls of the five damaged buildings are summarized in

Concepción EW
Concepción NS Table 2. The highest average ql is 2.49% for building AH, which has
Acceleration Spectrum (m/s2)

20
15
10
5 ings is 0.45%, which is about 1.8 times the minimum ratio of
0
0 0.5
Fig. 3. Comparison between elastic design spectra and response spectra of two
stations recorded near Concepcion.
obtained from Westenenek et al. [12]. Table 1 also summarizes the
strength reduction factor (R⁄) according to the design code [14] and
the factored inelastic design spectrum (1.4Sae/R⁄). If an elastic
behavior of the buildings is assumed, which is a reasonable
assumption before the brittle damage occurred in RC walls, the
response spectrum ordinates are 11.9, 5.4, 2.1, and 12.1 times
the factored inelastic design spectrum values for buildings CM,
AH, PR and AA, respectively.
The mean vertical boundary reinforcement (q), vertical distrib-
uted reinforcement (ql) and horizontal reinforcement (qt) ratios of
San Pedro EW a single wall with a maximum ratio of 3.27%, which is about 13
San Pedro NS
Sae soil type II
times the minimum ratio of 0.25% required by ACI 318 [8]. The
Sae soil type III cross section detailing of this wall is shown in Fig. 4. The high
average ql ratio in building AH seems to compensate for the low
average q of 0.22%. EM building also has a high ql ratio of 1.34%
compared to the other three buildings (CM, PR and AA), whose
average ratio is only 0.53%. The average qt ratio of the five build-
0.25% required by the ACI 318 code [8].
1 1.5 2 2.5 3
Period (s) 3. Experimental program
Three identical RC wall specimens identified as W1, W2 and W3
were designed with representative characteristics obtained from
the survey described in section 2. The design of the wall specimens
and the test setup including instrumentation and load application
protocol are described in this section. The tests were aimed to
reproduce the observed failure in walls during the 2010 Chile
Earthquake and to evaluate the influence of axial load in the
seismic behavior of the walls.
3.1. Wall geometry and reinforcement
For the wall specimen design, a 1/2 scale was selected for the
wall prototype due to laboratory limitations. The wall cross section

Table 1
Comparison between inelastic design spectrum and response spectrum for buildings located in Concepción.

Building Building period (s) Soil type Sae (m/s2) Response spectrum (m/s2) R⁄ Inelastic design spectrum (m/s2)
CM 0.80 II 4.05 7.74 8.8 0.65
AH 0.78 III 10.49 12.56 6.3 2.31
PR 0.50 III 12.11 7.06 5.2 3.29
AA 0.71 II 4.79 9.51 8.5 0.79

Table 2
Average reinforcement ratios of walls of damaged buildings considered for the wall
specimen design.

Building Number of walls q (%) ql (%) qt (%)

CM 27 0.57 0.61 0.32
AH 20 0.22 2.49 0.68
PR 14 0.45 0.50 0.50
AA 22 0.40 0.48 0.32
EM 4 0.51 1.34 0.45
Average 17.4 0.43 1.08 0.45 Fig. 4. Cross section of selected wall in AH building showing large ql ratio
(dimensions in mm).
16 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23
Fig. 5. Geometry and reinforcement detailing of wall specimens (dimensions in
mm).
is 700 mm wide, 100 mm thick, and the wall height is 1600 mm
(Fig. 5). Thus, the h/lw ratio of the wall prototype is 2.3. The wall
was designed with base and top RC beams to connect it to the
loading frame. The lateral load was applied at mid height of the
top beam (at 1750 mm from the base), and the resulting
M/Vlw ratio of the wall specimen is 2.5. This ratio is slightly
higher than the 2.02 average ratio of the surveyed walls but it is
in the actual range of practice and also ensures a flexural mode
of failure.
The vertical boundary reinforcement is 4 /10 mm (10 mm
diameter) bars (q = 0.45%) and the distributed vertical reinforce-
ment is /8 mm bars spaced at 140 mm in two layers (ql = 0.72%).
The horizontal distributed reinforcement is /5 mm bars spaced
at 90 mm (qt = 0.44%). A different steel was selected for the
horizontal reinforcement because of the available bar diameters,
which allowed to use an s/db ratio of 9.0, where s is the spacing
of the horizontal reinforcement and db is the diameter of the verti-
cal bars. This s/db ratio is within the range of 8–11 observed in
damaged RC walls [4]. The s/db ratio is an important parameter
since buckling of vertical reinforcement was found to be a critical
issue and the horizontal reinforcement in actual walls was bent
at the wall edges and restrained the buckling of the vertical
reinforcement. The reinforcement ratios of the wall specimens
are similar to the mean ratios obtained from the survey of
damaged buildings except for ql (0.72% for wall specimens), which
is smaller than the average of the survey (1.08% in Table 2) because
the average ratios in buildings AH and EM are excessively larger
than those of the other three buildings. The horizontal reinforce-
ment detailing follows typical Chilean construction practice which
consists on hoops with 90-degree hooks outside the vertical bars
which are not anchored into the concrete core and become ineffec-
tive after spalling of the concrete cover [3]. The reinforcement of
the base and top beams, not detailed in Fig. 5, consists of /
12 mm longitudinal bars with intermediate /10 mm longitudinal
bars and /8 mm stirrups spaced at 100 mm.
3.2. Materials
The concrete of the wall specimens was specified with a charac-
teristic strength of 20 MPa and with a maximum aggregate size of
11 mm. For the reinforcement, A630-420H steel (fy = 420 MPa) was
specified for /8, /10 and /12 mm bars, and AT560-500H steel
(fy = 500 MPa) for the transverse reinforcement (/5 mm bars).
The properties of concrete and steel were measured using standard
laboratory tests.
Concrete testing considered five 150 by 300 mm standard cylin-
der samples that were tested a day before the first wall test, at an
age of 260 days. The average concrete strength was fc’ = 27.4 MPa,
which is considered as the compressive concrete strength (fc0 ) for
the ALRs applied in the tests. The measured secant modulus of
elasticity at 0:4fc0 was Ec = 32,700 MPa. This modulus of elasticity
pffiffiffiffi33% larger than that proposed by ACI 318 (Ec = 4700-
is
fc0 = 24,600 MPa) [8].
For the reinforcing steel, three bars of each diameter (/5, /8
and /10) were tested in order to obtain the properties of the steel
bars (Table 3). The /5 mm AT560-500H steel bars are not as ductile
as the A630-420H steel bars. However, the Chilean code [14]
allows this type of steel for transverse reinforcement of RC walls
if the wall is designed using capacity design principles to prevent
a shear failure. Strains during tests were impossible to measure
for /5 mm bars and only yield strength, ultimate strength, and
ultimate strain were measured for these bars.
3.3. Test setup and instrumentation
The wall specimens were placed in a steel loading frame (Fig. 6)
where the wall is fixed at the base and pinned at the top. The hor-
izontal 500 kN actuator was pinned at both ends and attached to
the top RC beam with 4 steel bars that were bolted against
400  300  30 mm steel plates at each side of the wall. At the left
end, the horizontal actuator was connected to a steel I-beam which
was bolted to the steel frame. A 5 kN concrete counterweight was
connected to the clevis of the horizontal actuator at the wall side
using two pulleys. This counterweight (not shown in Fig. 6) was
used to hang the actuator and eliminate the vertical reaction
induced by its weight in the wall. The vertical 700 kN actuator
was bolted to the steel frame and connected to the wall using roll-
ers to allow horizontal displacement of the top RC beam. Therefore,
the P-delta effect was not included in the test setup. The out-of-
plane displacement of the walls was restrained with rolling
supports that were connected to a steel I-beam at each side of
the top RC beam.
The walls were instrumented with 2 load cells, 14 displacement
transducers, and 16 strain gauges attached to the reinforcement
bars (Fig. 7). Loads cells were connected to both horizontal and
vertical actuators to measure the applied loads. Six displacement
transducers were installed to measure the displacement and
 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23 17

Table 3 1 Load Cell

Average properties of reinforcing steel.
Displacement transducer
3 2
Parameter /5 mm /8 mm /10 mm 5 4 Transversal displacement
Steel AT560-500H A630-420H A630-420H
Yield strength (MPa) 608.9 445.6 469.2 6
Ultimate strength (MPa) 667.7 598.9 675.7
WEST EAST HT
H1W H1E
Modulus of elasticity (GPa) – 225.8 224.7
SECTION A-A
Yield Strain – 0.0020 0.0021
WEST EAST
Hardening strain – 0.0139 0.0138
Ultimate strain 0.057 0.151 0.166
Strain hardening modulus (MPa) – 4134 5431 H2W H2E
10 13 V1T V7T
150
200
11 14 V1M H1W H1E V7M
150
7 8 A A
12 15 200
150 V2 V6
rotation of the top RC beam (transducers 3, 4, 5, 6, 7 and 8), six to 16 V1 V3 V4 V5 V7
measure the wall curvature (transducers 10–15), and two to mea- 9
sure the displacement of base RC beam (transducers 9 and 16). The
strain gauges were used in each wall to measure strains in the rein-
forcing steel bars and were installed during the construction of the (a) (b)
walls.
Fig. 7. Instrumentation of wall specimens: (a) 2 load cells and 14 displacement
transducers, and (b) 16 strain gauges.
3.4. Load application and control

Constant axial load and increasing lateral displacements were Table 4

applied to the wall specimens. The walls were tested with ALRs Applied axial loads for wall specimens result in ALRs of 0.15, 0.25 and 0.35 which are
of 0.15, 0.25 and 0.35. Table 4 shows the ALR and the applied axial within the range obtained from 70% of the surveyed walls. The limit imposed in the
2011 Chilean design code [15] corresponds to the maximum ALR of 0.35 for wall W3.
load for the three wall specimens.
The three wall specimens were subjected to a constant verti- Wall ALR (N=fc0 Ag ) Axial load (kN)
cal load that was maintained throughout the test by controlling W1 0.15 287.4
the hydraulic pressure. Following the application of the axial W2 0.25 479.0
load, the walls were subjected to horizontal displacement with W3 0.35 671.6
increasing amplitude and two cycles at the same amplitude.
The displacement cycles were designed based on the horizontal
yield displacement of the wall which was estimated as
50
Dy = 5.5 mm for the three wall specimens. The yield displacement
Actuator displacement (mm)

2 40
is obtained from Dy ¼ hw /y =3, where hw = 1750 mm is the height
30
of the free cantilever wall, and /y is the estimated yield curvature
20
[11]. The ductility factors D/Dy that where intended to be applied
10
at peak displacements of the three walls were 0.5, 1, 1.5, 2, 3, 4,
0
5, 6, and 8. The actuator command displacement loading protocol
− 10
is shown in Fig. 8, where the actuator displacement was
− 20
increased by 20% to account for the flexibility of the loading
− 30
frame measured in a previous calibration test with a wall of
− 40
the same characteristics.
− 50

0 20 40 60 80 100 120 140 160

4. Test results Time (min)

This section describes the observed damage and failure patterns Fig. 8. Horizontal actuator command displacement with two cycles at each peak
amplitude of 3.3, 6.6, 9.9, 13.2, 19.8, 26.4, 33, 39.6 and 52.8 mm.
of the three tested walls. Load–displacement relationships are

EAST WEST

Steel I beam
700 kN
Actuator Load cell Rolling Support
IN Beam 500 kN Actuator Steel plate
with rollers

Plate Plate
Steel bar
1750
mm Wall
specimen

LATERAL VIEW

Fig. 6. Test setup (lateral view shown in larger scale).

18 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23
horizontal and
diagonal cracks
vertical crack
(a) W3
(d) W1
Fig. 9. Observed behavior of wall specimens: cracking process and failure. Horizontal and diagonal cracks were followed by vertical cracks at the initiation of cover concrete
spalling. The walls experimented a flexural-compressive concrete crushing failure with vertical reinforcement buckling, horizontal reinforcement opening, and out-of-plane
wall buckling.
compared among walls and the plastic hinge approach is used to
estimate ultimate wall displacements and rotations. In the follow-
ing discussion, the lateral wall displacement (D) and drift (D/hw)
are obtained from the displacement transducer located in the wall
specimens at the level of the horizontal actuator (transducer 3 in
Fig. 7(a)) and not from the actuator displacement.
4.1. Cracking process and failure mode
The behavior and failure mode of the three wall specimens with
ALR of 0.15, 0.25 and 0.35 was controlled by flexural-compressive
interaction due to the relatively high M/Vlw ratio of 2.5. The crack
initiation and propagation was similar for the three walls at the
first cycles of the tests (Fig. 9) but the ultimate stage was affected
by ALR (Fig. 10).
The first cracks were observed at 8.1 mm (0.5% drift) lateral
wall displacement for the three walls. These first cracks were
diagonal cracks along the height of the walls which were
initiation of cover
concrete spalling
(b) W3
(c) W3
(e) W2
followed by horizontal flexural cracks near the base of the walls
on the tensile zone (Fig. 9(a)). The initiation of yielding occurred
in the vertical bars at the boundary. Fig. 11 shows the strain his-
tory of strain gauges V1M (left toe) and V7M (right toe) which
were located 200 mm above the concrete base in the vertical bars
at the boundary, V1 and V7, respectively (Fig. 7(b)). For wall W1,
first yielding was measured in compression in bars V1 at a wall
displacement of 9.7 mm (0.6% drift) and V7 bars also yielded in
compression in the following loading reversal. For walls W2
and W3 first yielding also occurred in compression in bars V1
at a wall displacement of 4.9 mm (0.3% drift) and 4.6 mm (0.3%
drift), respectively. Bars V7 also yielded in compression in the
following cycle. In walls W2 and W3 yielding occurred before
the first observed crack. This early yielding in compression seems
to be attributed to the high compressive strain of about 0.1% (i.e.
half of the yield strain of the steel bars) which was recorded after
the application of the axial load. Fig. 11 also shows that bars V1
and V7 in walls W2 and W3 showed larger compressive strains
 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23 19

52.5
failure
3 complete spalling of concrete cover that exposed the reinforce-
displacement (mm)

vertical
(a) W1 horizontal

first crack
spalling ment occurred at the wall boundaries generally at about
26.3 1.5
crack 26 mm wall displacement (1.5% drift) for the three walls. After

Drift (%)
0 0
reaching this drift level, wall W3 with ALR of 0.35 failed imme-
diately after spalling of the cover concrete at 26.5 mm of wall lat-
yielding
− 26.3 −1.5 eral displacement (1.5% drift). Wall W2 with ALR of 0.25 survived
longer and failed at the last reversal cycle of that same amplitude
− 52.5 −3
0 2000 4000 6000 8000
at 31.3 mm peak lateral wall displacement (1.8% drift). On the
other hand, wall W1 with ALR of 0.15 could attain much more
Time (sec)
deformation after concrete spalling as its failure occurred at
52.5 3 48 mm displacement (2.8% drift). For the three wall specimens,
displacement (mm)

vertical the test ended when the walls were not able to carry the applied
(b) W2 horizontal

spalling
crack
26.3 first 1.5 axial load. A brittle compressive failure along the entire length of
crack

Drift (%)
the walls together with a sideway out-of-plane buckling was
0 0
observed in the three walls (Fig. 9(c) and (d)). Not only the bars
yielding
− 26.3 −1.5
closest to the wall edges exhibited buckling; but also the second
failure bars (Fig. 9(e)). Additionally, the horizontal reinforcement opened
− 52.5 −3 due to the poor detailing with the 90° hooks. Therefore, the low
0 2000 4000 6000 8000
concrete confinement provided by the horizontal reinforcement
Time (sec) became completely ineffective.
The observed failure mode of the three wall specimens
52.5 3
(Fig. 12) was similar to that observed in some walls damaged
displacement (mm)
(b) W3 horizontal

vertical spalling
26.3 first crack 1.5 during the 2010 Chile earthquake (Fig. 1). The horizontal and
failure
crack vertical reinforcement bars in the three wall specimens did not
Drift (%)

0 0 fracture during the tests. The fracture of vertical reinforcement

yielding
− 26.3
− 52.5
0 2000 4000
Time (sec)
Fig. 10. Behavior of the wall specimens through time showing that high axial load
triggered an early reinforcement yielding in compression for walls W2 and W3 and
reduced the lateral deformation capacity from 48 mm in wall W1 to 31.3 mm
(34.8% reduction) and 26.5 mm (44.8% reduction) in walls W2 and W3, respectively.
than that from wall W1, which is attributed to the higher ALR in
walls W2 and W3.
After yielding of the longitudinal reinforcement, the horizontal
flexural cracks located near the base of the wall propagated
toward the center of the wall specimens. Also, additional flexu-
rally induced diagonal cracks were visible along the height of
the walls. However, in wall W3 these diagonal cracks were less
and had lower thickness than those of W1. Vertical cracks at
the wall base in the position of bars V1 and V7 were also visible
after yielding, generally at a wall displacement of 17.7 mm (1.0%
drift), which seems to indicate that vertical reinforcement started
to experiment bar buckling due to their large s/db = 9 ratio
and started to push against the concrete cover (Fig. 9(b)). The
in actual buildings may be attributed to additional cycles in dam-
−1.5
aged walls caused by the long duration of the earthquake or a
−3 dynamic effect not considered in these tests. Since a constant
6000 8000 vertical load was applied to the walls, they were not able to
transfer this load to adjacent structural elements and the walls
crushed in compression.
It is apparent from Fig. 10 that as the axial load increases, the
‘‘distance’’ between spalling of the concrete cover and brittle fail-
ure of the wall reduces drastically. This observation enables us to
state that the observed failure in many walls during 2010 Chile
earthquake was brittle and occurred just a few cycles after spalling
of the concrete cover. This is certainly inconsistent with the
assumption of large strength reduction factors (R) in the design
building codes.
The tests also suggests that spalling of walls occurs at a drift
of 1.5%, which contradicts the estimated drift demand of 1% for
buildings during 2010 Chile earthquake [4]. So additional
research needs to be conducted, such as T-wall sections, or
building irregularities, to understand what occurred in damaged
buildings.
4.2. Load–displacement relationships
The measured load–displacement relationships of walls W1,
W2 and W3 are shown in Fig. 13 and relevant results are

−3 −3
x 10 x 10
4 4
2 2
V1M Steel strain

V7M Steel strain

0 0
−2 −2
−4 −4
−6 −6 W1

−8 −8 W2
W3
−10 −10
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400
Time (sec) Time (sec)

Fig. 11. Strain versus time for strain gauges V1M and V7M for specimens W1, W2 and W3 (first four displacement amplitudes) show that yielding occurred first in
compression.
20 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23
(a) W1
Fig. 12. Failure mode in the walls specimens consisted on crushing of concrete, buckling of vertical reinforcement, and opening of horizontal reinforcement, which was
similar to the observed damage in 2010 Chile earthquake shown in Fig. 1.
summarized in Table 5. Wall W1 failed under concrete crushing at
48 mm (2.7% drift) wall displacement, seven cycles after reaching
the peak strength. A strength reduction of 5.9% occurred in the pre-
vious completed cycle of 39.3 mm (2.2% drift) wall displacement
amplitude. The strength reduction measured in subsequent cycles
of equal amplitude is attributed to buckling of vertical reinforce-
ment. Wall W2 experienced a concrete crushing failure at
31.3 mm (1.8% drift) wall displacement. The failure occurred at
the second cycle at this amplitude and it was triggered by a
strength reduction of 6.6% relative to the strength of the preceding
cycle. When vertical and horizontal cracks at the base of wall W3
were still developing, suddenly the concrete reached high com-
pressive stresses that exceeded concrete capacity and triggered a
concrete crushing failure at 26.5 mm (1.5% drift) wall displace-
ment. Prior to failure, at 24.4 mm (1.39% drift) wall displacement
cycles (1.39% drift), a 6.7% strength reduction was measured in
the second cycle of that amplitude.
The dissipated work, estimated as the enclosed areas of the
load–displacement relationships for the last completed cycles are
5.66 kN m for wall W1, 4.87 kN m (14% reduction) for wall W2,
and 3.67 kN m (35% reduction) for wall W3. The equivalent viscous
damping ratios for these cycles, computed as the ratio of dissipated
work over 4p times the elastic work of the wall, are 17%, 15%, and
14% for walls W1, W2 and W3, respectively. Please note that true
equivalent viscous damping ratios for the building are much smal-
ler than these values because the elastic energy of the structure is
much larger than the sum of the elastic energies of the walls. In any
case, the relevant trend is that the equivalent viscous damping
ratios decrease as ALR increases.
The comparison between the analytical and experimental
strength is summarized in Table 6. In this table, Vn is the shear
strength and Vi is the interaction strength computed using
ACI318 [8] with the material properties of Section 3.2. For the
interaction strength, a lever arm of hw = 1750 mm is considered.
It is important to note that the axial load of wall W3 (Table 4)
is 97% of the axial load corresponding to the balance point. The
ACI shear strength overestimates the experimental strength 52%
on average because the walls are slender and their strength is
controlled by flexure. On the other hand, the ACI flexural
strength is on average 75% of the experimental strength. This
low estimation is attributed to the location of the critical region,
which was located above the base and thus the lever arm is less
than hw.
(b) W2 (c) W3
4.3. Analytical estimations and plastic hinge length
A fiber model for RC sections was developed using MATLAB
[17] in order to perform a flexural analysis of the wall speci-
mens. Fig. 14 shows the discretization of the cross-section of
the wall and the strain profile, where plane sections remain
plane. The stress–strain constitutive models proposed by Karthik
and Mander [18] were adopted for concrete and steel consider-
ing the material properties described in Section 3.2. At midway
between stirrups, the effective confined concrete area is only
3.3% of the concrete core which results in a low confinement
ratio of K ¼ fcc0 =fc0 ¼ 1:01, where fcc0 is the compressive concrete
strength of the confined core [19]. For a given curvature, the
bending moment of the fiber section is obtained by finding the
neutral axis depth such that the applied axial load (Table 4) is
achieved.
The ultimate displacement and rotation of the wall specimens
are estimated using the simplified plastic hinge approach
(Fig. 15) where a plastic hinge length of lp = 2.5tw = 250 mm is
considered [20], and the ultimate and yield curvatures are
obtained from the fiber model. This plastic hinge length agrees
well with the average experimental plastic hinge length of
243 mm. The experimental plastic hinge length is simply defined
as the distance between the bottom and top edges of the spalled
region of the walls at failure (Fig. 16). The resulting displace-
ments and rotations are summarized in Tables 7 (Dflex) and 8
(hflex), respectively. The estimated ultimate displacements for
the three walls are on average 61% of the experimental displace-
ments (Dexpu ), and the ultimate rotations are on average 72% of the
experimental rotations (hexp u ), where the ultimate rotations at the
edge of the walls were obtained from the displacement transduc-
ers 7 and 8 in Fig 7. These large differences are attributed mostly
to shear deformation and rigid body rotation of the wall speci-
mens induced by reinforcement slip of the vertical bars. To pre-
dict the displacement correctly, a plastic hinge length of
Lp = 5.9, 4.9, and 5.2tw has to be used for walls W1, W2, and
W3, respectively.
The ultimate displacements and rotations are also estimated
using a four-component model (Dfc, hfc) that was developed for this
comparison an includes flexure, shear, slip, and the base rotations
of the wall specimens (Table 7 and 8). The flexure component
(Dflex, hflex) is estimated from the plastic hinge model described pre-
viously using lp = 2.5tw. The shear component, which contributes
 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23 21

Drift (%) Table 6

Comparison between nominal and experimental lateral strength.
−3 −2 −1 0 1 2 3
200
Wall Vn (kN) Vi (kN) V exp
max V n =V exp
max
V i =V max exp
Horizontal load (kN)

W1 248 111 144.3 1.72 0.77

100
W2 248 126 166.0 1.49 0.76
W3 248 133 185.6 1.34 0.72
0

−100 (a) W1

Strain profile
−200
−52.5 −35 −17.5 0 17.5 35 52.5
Displacement (mm)

Unconfined concrete strain

Drift (%)
Confined concrete strain
−3 −2 −1 0 1 2 3
200
Steel strain
Horizontal load (kN)

100

700 mm
0

Neutral axis depth

−100 (b) W2

−200
−52.5 −35 −17.5 0 17.5 35 52.5

Displacement (mm)

Drift (%)
−3 −2 −1 0 1 2 3
200
Confined concrete block
Horizontal load (kN)

100 Unconfined concrete block

Location of steel bars
Location for concrete strain
0

−100 (c) W3 Fig. 14. Fiber model. The height of the fibers shown in this diagram is larger than
the 1 mm fiber height used in the analysis.

−200
−52.5 −35 −17.5 0 17.5 35 52.5
Displacement (mm)
slip component (Dslip, hslip) is estimated using the equation pro-
posed by Sezen and Setzler [21]. The slip of the bars in the tensile
Reinforcement yielding
Cover concrete spalling side generates a rigid body rotation of the walls that induces lateral
Concrete crushing failure displacement. Finally, the contribution due to rigid body rotation
of the concrete base was measured in the test (Drot, hrot). The result-
Fig. 13. Load–displacement relationships show that the lateral strength of the walls ing displacements and rotations are on average, 73% and 84% of the
(peak loads from the tests) is strongly influenced by the ALR. The 144.3 kN strength
experimental displacements and rotations, respectively. It is con-
from wall W1 increases to 166 kN (15% increase) and 185.6 kN (28.6% increase) in
walls W2 and W3, respectively. cluded that a nonlinear shear deformation needs to be considered
in the four component model to provide a better estimation of the
ultimate displacement.

Table 5
Summary of test results.

Result W1 W2 W3
Yield load (kN) 96.6 73.4 83.6
Peak load (kN) 144.3 166.0 185.6
Failure load (kN) 81.3 153.5 146.5
Drift at yielding (%) 0.6 0.3 0.3
Drift at peak load (%) 1.8 1.8 1.2
Drift at failure (%) 2.7 1.8 1.5

only to the displacement (Dshear), is estimated with a simplified

elastic approach as Dshear = 1.2V exp exp
max hw =GAg where V max is the peak Fig. 15. Ultimate lateral displacement and rotation estimation using the plastic
lateral load and G = 0.4Ec = 13,080 MPa is the shear modulus. The hinge approach.
22 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23

(a) W1 (b) W2 (c) W3

Fig. 16. Damage was concentrated at the concrete base of the wall specimens; the average plastic hinge lengths measured along the wall length for W1, W2 and W3 were
220, 240, and 270 mm, respectively.

Table 7
Comparison between the analytical estimation and experimental values of the ultimate lateral displacement.

Wall Dflex (mm) Dshear (mm) Dslip (mm) Drot (mm) Dfc (mm) Dexp
u (mm) Dflex =Dexp
u Dfc =Dexp
u

W1 25.7 0.3 3.1 1.7 30.8 48.0 0.54 0.64

W2 20.1 0.4 1.6 1.6 23.7 31.3 0.64 0.76
W3 17.0 0.4 2.0 1.6 21.0 26.5 0.64 0.79

Table 8
Comparison between the analytical estimation and experimental values of the ultimate rotation.

Wall hflex (rad) hslip (rad) hrot (rad) hfc (rad) hexp
u (rad) hflex =hexp
u hfc =hexp
u

W1 0.0172 0.0018 0.0010 0.0200 0.0276 0.62 0.72

W2 0.0138 0.0009 0.0009 0.0156 0.0179 0.77 0.87
W3 0.0121 0.0011 0.0009 0.0166 0.0141 0.78 0.91

5. Conclusions
The earthquake that struck southern and central Chile in 2010
provides valuable information about building performance.
Despite of the global assessment of acceptable performance of RC
buildings, the observed brittle damage induced by the earthquake
motivates conducting research with the purpose of improving the
seismic design provisions of RC buildings. Wall characteristics
obtained from a survey performed on real buildings led us to
observe an average M/Vlw ratio of 2.02, low wall thicknesses, and
lack of boundary confinement. These factors suggest that RC walls
experienced brittle damage controlled by flexural-compressive
interaction. The mean ALR of critical walls in five damaged build-
ings was 0.18 for service gravity forces. This ALR increases signifi-
cantly if seismic forces are included.
The flexural-compressive failure mode observed in RC walls
during 2010 Chile earthquake was reproduced experimentally in
the performed tests. Concrete crushing, vertical reinforcement
buckling, horizontal reinforcement opening at wall boundaries,
and out-of-plane wall buckling observed in the tests was similar
to the damage observed in walls after the earthquake. The failure
mode of wall W3, with ALR of 0.35, was brittle and extremely sud-
den; experimental concrete crushing in compression occurred at
1.5% drift, immediately after spalling of the concrete cover. Wall
W2, with ALR of 0.25, showed the same type of failure mode but
less brittle; failure occurred at 1.8% drift, three cycles after concrete
spalling. For wall W1, with ALR of 0.15, the behavior was more
ductile than that for the other two walls and failure occurred at
2.7% drift. It is concluded that the displacement capacity of the wall
was reduced almost by half when the ALR was increased from 0.15
to 0.35. It is concluded that the ALR limit of 0.35 adopted in current
Chilean code is not adequate for slender walls with non-seismic
boundary detailing.
The average measured plastic hinge length for the three walls
was 243 mm which agrees well with the result obtained by Takah-
ashi et al. [20]. The plastic hinge approach with a plastic hinge
length of Lp = 2.5tw estimates on average only 61% of the lateral dis-
placement. A plastic hinge length of Lp = 5.9, 4.9, and 5.2tw has to
be considered to estimate the ultimate displacement adequately
for walls W1, W2, and W3, respectively.
Some of the buildings which did not exhibit damage in 2010
may be susceptible to brittle behavior in a future earthquake.
These structures should be analyzed to determine if retrofit is
required to sustain several inelastic deformation cycles in future
events. The use of energy dissipation devices in such buildings
may be a good alternative for retrofitting since they could control
the seismic axial loads of walls and could limit the flexural-com-
pressive failure modes exhibited by the tested walls and the walls
 C. Alarcon et al. / Engineering Structures 73 (2014) 13–23
in actual buildings. Additionally, for future research it is recom-
mended to test thicker RC walls to validate the minimum thickness
of 300 mm required for walls with boundary confinement in the
current Chilean Code.
Acknowledgments
The authors are grateful to many people and institutions which
made this work possible. This research has been funded by the
Chilean Fondo Nacional de Ciencia y Tecnología, FONDECYT through
Grant #1110377 and FONDAP through Grant #15110017. The
authors would also like to thank the professors Carl Lüders and
Bozidar Stojadinovic for their support during the tests, and the
students R. Jünemann, A. Marihuén, R. Manieu, F. Quitral, J. Reme-
sar, J. Rendic, M. Ochagavía, F. Riquelme, M. Saavedra, C. Barrueto
and A. Gutiérrez for their contribution in the project. The authors
are also thankful to the engineers and technicians from DICTUC
S.A and the Laboratory of the Structural and Geotechnical
Engineering Department of Pontificia Universidad Católica de
Chile.
References
[1] EERI. The Mw 8.8 Chile Earthquake of February 27, 2010. EERI Earthquake
Special Report 2010.
[2] Jünemann R, Hube M, de la Llera JC, Kausel E. Characteristics of reinforced
concrete shear wall buildings damaged during 2010 Chile Earthquake.
Proceedings of 15th World Conference on Earthquake Engineering, Lisbon,
Portugal, 2012, Paper N°2265.
[3] Massone L, Bonelli P, Lagos R, Lüders C, Moehle J, Wallace J. Seismic design and
construction practices for RC structural wall buildings. Earthquake Spectra
2012;28(S1):S245–56.
[4] Wallace J, Massone L, Bonelli P, Dragovich J, Lagos R, Lüders C, Moehle J.
Damage and implications for seismic design of RC structural wall buildings.
Earthquake Spectra 2010;28(S1):S281–99.
23
[5] Westenenk B, de la Llera JC, Jünemann R, Hube M, Besa JJ, Lüders C, et al.
Analysis and interpretation of the seismic response of RC buildings in
Concepción during the February 27, 2010, Chile Earthquake. Bull Earthquake
Eng 2012;11(1):69–91.
[6] Wood SL. Performance of reinforced concrete buildings during the 1985 Chile
Earthquake: implications for the design of structural walls. Earthquake Spectra
1991;7(4):607–38.
[7] Hidalgo PA, Ledezma CA, Jordán RM. Seismic behavior of squat reinforced
concrete shear walls. Earthquake Spectra 2002;18(2):187–208.
[8] American Concrete Institute (ACI). Building Code Requirements for Structural
Concrete and Commentary (ACI 318–11); 2011.
[9] Zhang Y, Wang Z. Seismic behavior of reinforced concrete shear walls
subjected to high axial loading. ACI Struct J 2000;97(5):739–50.
[10] Su R, Wong S. Seismic behavior of slender reinforced concrete shear walls
under high axial load ratio. Eng Struct 2007;29:1957–65.
[11] Alarcon C. Influence of axial load in the sesmic behavior of reinforced concrete
walls with nonseismic detailing. Master of Science Thesis. Pontificia Universad
Católica de Chile; 2013.
[12] Westenenk B, de la Llera J, Besa J, Jünemann R, Moehle J, Lüders C, et al.
Response of reinforced concrete buildings in Concepción during the Maule
Earthquake. Earthquake Sprectra 2012;28(S1):S257–80.
[13] Computers & Structures, Inc. ETABS v.9.7.1 2011. Berkeley, California, USA.
[14] Instituto Nacional de Normalización (INN). Earthquake resistant design of
buildings, NCh433 Of.1996 Mod 2009. Santiago, Chile; 2009 [in Spanish].
[15] DS 60 MINVU (DS 60). Reinforced concrete design code, replacing D.S N 118,
2010. Chilean Ministry of Housing and Urbanism, Diario Oficial; 13 December
2011 [in Spanish].
[16] Instituto Nacional de Normalización (INN). Structural design – general
dispositions and combinations of loads, NCh3171 Of.2010. Santiago, Chile;
2010 [in Spanish].
[17] MathWorks Inc. MATLAB version R2011b.
[18] Karthik M, Mander J. Stress-block parameters for unconfined and confined
concrete based on a unified stress-strain model. J Struct Eng 2011;137(2):
270–3.
[19] Mander J, Priestley M, Park R. Theoretical stress–strain model for confined
concrete. J Struct Eng 1988;114(8):1804–26.
[20] Takahashi S, Yoshida K, Ichinose T, Sanada Y, Matsumoto K, Fukuyama H, et al.
Flexural drift capacity of reinforced concrete wall with limited confinement.
ACI Struct J 2013;110(1):95–104.
[21] Sezen H, Setzler E. Reinforcement slip in reinforced concrete columns. ACI
Struct J 2008;105(3):280–9.