Vous êtes sur la page 1sur 19

Damage and Implications for Seismic

Design of RC Structural Wall Buildings


John W. Wallace,
a)
M.EERI, Leonardo M. Massone,
b)
Patricio Bonelli,
c)
Jeff Dragovich,
d)
M.EERI, Ren Lagos,
e)
Carl Lders,
f)
and Jack Moehle,
g)
M.EERI
In 1996, Chile adopted NCh433.Of96, which includes seismic design
approaches similar to those used in ASCE 7-10 (2010) and a concrete code
based on ACI 318-95 (1995). Since reinforced concrete buildings are the
predominant form of construction in Chile for buildings over four stories, the
27 February 2010 earthquake provides an excellent opportunity to assess the per-
formance of reinforced concrete buildings designed using modern codes similar
to those used in the United States. A description of observed damage is provided
and correlated with a number of factors, including relatively high levels of wall
axial load, the lack of well-detailed wall boundaries, and the common usage of
flanged walls. Based on a detailed assessment of these issues, potential updates to
U.S. codes and recommendations are suggested related to design and detailing of
special reinforced concrete shear walls. [DOI: 10.1193/1.4000047]
INTRODUCTION
The area impacted by the M
W
8.8 2010 earthquake is the most densely populated region
of Chile and includes the cities of Via del Mar, Santiago and Concepcin. Between 1985
and 2009, a total of 1,939 construction permits were issued for reinforced concrete buildings
with nine or more floors (CChC 2010). Damage observed following the earthquake was
generally concentrated in newer and taller buildings, with one complete collapse (Alto
Ro building), several partial collapses, and about 40 severely damaged buildings that
were either repaired, or in rare cases, demolished. Severely damaged reinforced concrete
buildings correspond to about 2% (40/1,939) of the newer building stock with nine or
more floors in south-central Chile.
As summarized by Massone et al. (2012), typical buildings in Chile include a large num-
ber of structural walls compared with U.S. construction, such that the ratio of total wall area
Earthquake Spectra, Volume 28, No. S1, pages S281S299, June 2012; 2012, Earthquake Engineering Research Institute
a)
Professor, University of California, Los Angeles, Department of Civil & Environmental Engineering, 5731
Boelter Hall, Los Angeles, California, 90095-1593
b)
Assistant professor, University of Chile, Department of Civil Engineering, Blanco Encalada 2002, Santiago,
Chile
c)
Universidad Federico Santa Mara, Departamento de Obras Civiles, Av. Espaa 1680, Valparaso, Chile
d)
Research Structural Engineer, National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg,
Maryland, 20899
e)
Ren Lagos & Asociados, Magdalena 140, Santiago, Chile
f)
Professor Emeritus, Catholic University of Chile, Department of Civil & Geotechnical Engineering, Av. Vicua
Mackenna 4860, Santiago, Chile
g)
Professor, University of California, Berkeley, Department of Civil & Environmental Engineering, 775 Davis
Hall, Berkeley, California, 94720-1710
S281
to floor plan area is on average roughly 3% in each principal direction of a building. Walls in
each principal direction are often connected to form T-shaped or L-shaped cross sections.
Buildings constructed since 1996 tend to use a common layout, with longitudinal walls on
each side of a central corridor that form a spine and perpendicular (transverse) walls, or ribs,
spaced at roughly 5m along the central corridor. Wall thickness of 15 cm to 20 cm is com-
mon. Seismic design approaches used in Chile are similar to those used in the United States,
e.g., modal response spectrum analysis; however, code design spectra in the 1996 Chile code
were strongly influenced by the M7.8 March 3, 1985 earthquake that occurred along the
subduction zone at the interface of the South American and Nazca plates.
Design requirements for RC shear wall buildings, the predominant form of construction
in Chile for buildings over four stories, were updated via reference from NCh433.Of96 to
ACI 318-95, with only minor exceptions. An important exception, based on the good per-
formance of roughly 400 RC wall buildings in Via del Mar in the 1985 earthquake, is that
ACI 318-95 requirements for special transverse reinforcement at wall boundaries to confine
the concrete and restrain rebar bucking were eliminated.
Given the large number of modern RC buildings that exist in the impacted region, the
27 February 2010 M
W
8.8 Chile earthquake offers a excellent opportunity to assess the per-
formance of reinforced concrete buildings designed using modern codes similar to those used
in the United States. In the following, we discuss damage observations and identify issues
that may not be adequately addressed or may need to be updated in U.S. codes and recom-
mendations as well as in other countries that experience strong ground shaking.
OBSERVED DAMAGE
Widespread and significant damage to RC structural walls (Figure 1), particularly at or
near the ground level, was observed in buildings in Santiago, Via del Mar, and Concepcin,
the major population centers in Chile, following the 2010 earthquake. In general, crushing
and spalling of concrete and buckling of vertical reinforcement were observed, often over the
entire wall length. Typically, the damage was concentrated over a short height equal to one to
three times the wall thickness, apparently because buckling of vertical bars led to damage
concentration. Closer inspection of wall boundary regions (Figure 2) revealed relatively large
Figure 1. Typical wall damage in the 2010 Chile earthquake.
S282 WALLACE ET AL.
spacing of hoops (20 cm) and horizontal web reinforcement (20 cm), as well as 90-degree
hooks (Massone et al. 2012). Because walls are thin, typically 15 cm to 20 cm thick, any
spalling of cover concrete (about 2 cm on each side) results in a 20% or 27% reduction in the
wall thickness. Once cover concrete spalls, the 90-degree hooks used on transverse reinforce-
ment at wall boundaries become ineffective (they open). The large spacing of transverse
reinforcement and the opening of 90-degree hooks after concrete spalling likely contributed
to buckling of vertical reinforcement.
Large cyclic strain demands on reinforcement, combined with relatively large axial stress,
may have produced abrupt buckling in some walls without the need for significant or any
concrete cover spalling. Buckling of longitudinal reinforcement in walls also was combined
in some cases with fracture of reinforcement, again, likely due to large cyclic strain demands
and the long duration motions. Lateral instability (buckling), primarily at web boundaries of
T- or L-shaped wall cross sections, was observed (Figure 3); prior to the Chile and
Figure 2. Boundary wall reinforcement: (a) Fracture, (b) buckling, and (c) detailing.
Figure 3. (a) Wall lateral instability and (b) wall discontinuities/instability.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S283
New Zealand earthquakes, this lateral instability failure had been primarily observed in
laboratory tests (e.g., Thomsen and Wallace 2004). Detailed surveys conducted as part of
ATC-94 (2011) indicate that lateral wall instability was not likely driven by prior yielding
in tension (as had originally been suspected based on past research, e.g., Corley et al. 1981;
Paulay and Priestley, 1993; Chai and Elayer, 1999) but instead was the result of lateral
instability of previously crushed, thin wall boundary zones. This failure mode has not
been adequately studied.
Damage to thin walls was commonly observed in first floor or the first basement level,
either at the base of the wall or near the top of the wall (Figure 1). The location of damage
might have been influenced by concrete quality at the top of walls (e.g., over- consolidation
producing a weak layer at the top of the wall); however, in many cases the location of damage
was likely due to stress/strain concentrations from abrupt changes in the wall length to
accommodate parking spaces, either at ground or basement levels (Figure 3b). Damage
to vertical and horizontal wall segments, created by window and door openings on building
facades, was frequently observed (Figure 4a). Damage to these wall segments, typically with
aspect ratio close to one, generally consisted of wide diagonal cracks (wider than 1cm in
many cases) and concrete spalling. Similar damage was observed in coupling beams
(Figure 4b), as well as in wall segments across corridors, either near the roof level or
near basement levels (Figure 5, also see Figure 9). In some cases, coupling beams and
walls were not placed in the same plane (Figure 4b), creating eccentric connections that
led to damage. Significant concrete spalling and large diagonal cracks (width of several cen-
timeters) also were observed at discontinuities, such as shown in Figure 5, and within floor
diaphragms, where struts formed to redistribute forces from damaged walls to adjacent walls.
In some older buildings (pre-1990), damage was observed in lightly reinforced (both
longitudinal and transverse reinforcement, by U.S. standards) lintel beams along interior
Figure 4. (a) Pier and spandrel failure. (b) Coupling beam spandrel failure.
S284 WALLACE ET AL.
corridor walls (Figure 6a), with beam damage sometimes extending over the entire building
height (in all stories). In other cases, poor anchorage of beam reinforcement within walls
was noted (Figure 6b). In newer construction, lintel beams over doorways were replaced
with nonstructural material; therefore, only slab coupling exists along and across corridor
walls. Damage to corridor slabs, which usually span 2 m to 3 m and are 15 cm to 20 cm
thick, was observed in numerous buildings (Figure 7a), as the slabs are subjected to large
rotation demands due to wall deformations. However, in some cases, severe slab damage
was observed to be due to, or exacerbated by, large vertical deformations imposed on the
slab due to wall shortening at lower levels due to concrete crushing/spalling and rebar
buckling. In some taller buildings, large (cap) beams were used to couple walls at or
near the roof level. Large shear demands on cap beams, or similar cap beams near
the building base where walls were extended across the corridor (Figure 5), tended to pro-
duce significant diagonal cracking and concrete spalling. It is noted that use of diagonal
reinforcement in wall piers, coupling beams, and cap beams, is not common in Chile.
In typical Chilean buildings, shear walls are used to divide apartment units; within units,
partition walls are common. Severe damage was observed to these interior partition walls in
some buildings (Figure 7b). Evidence of foundation rotation (soil heaving) was observed at
Figure 5. Damage at various discontinuities.
Figure 6. (a) Lintel beam failure and (b) wall-beam connection damage.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S285
one building in Concepcin and one building in San Pedro with mat foundations near grade;
however, foundation performance was generally good for taller buildings.
Although severe damage was observed in taller buildings in Concepcin, only one mod-
ern RC building collapsed, the 15-story Alto Ro building constructed in 2007. The building
included features similar to those in other damaged buildings, such as T-shaped wall sec-
tions and vertical irregularities (setbacks) in transverse walls and discontinuous perimeter
walls just below the top of the first floor along axis I to accommodate parking (Figure 8), as
well as discontinuities below the first floor where transverse walls were extended across the
central corridor at axes 5, 8, 11, and 13 (e.g, Figures 8b, 9a, 9b). Damage at transverse wall
elevations at Axes 8 and 13 are provided in Figure 9a, 9b (IDIEM 2010), which indicate
concrete spalling and reinforcement buckling at discontinuities. A schematic of a possible
collapse scenario is shown in Figure 9c. Based on observed wall damage in a large number
of buildings, as well as the wall configuration for the Alto Ro building, it appears quite
likely that flexural compression failure (concrete crushing, rebar buckling) occurred at
the location where the transverse wall lengths were reduced on the side of the building
with parking (at the top of the first story, indicated as 1 on Figure 9c). Damage may
have initiated at axes with T-walls (8, 13, and 20), or alternatively at the discontinuity
in the L-shaped to rectangular transverse wall (11, 17, 24) at the top of the first story.
In either case, once crushing/buckling initiated in one of these walls, axial load would
be redistributed, making the other wall more susceptible to failure. High shear stresses
in the first story corridor wall below the open corridor above are likely to produce damage
at this location (Fig, 9a,b), indicted as 2 on Figure 9c. Once these transverse walls were
damaged at the perimeter and at the central corridor, it is plausible that splice failure
occurred along the remaining portion of the wall, indicated as 3 on Figure 9c. Splice
failures of web vertical bars and boundary vertical bars at the intersection of Axis 8
Figure 7. (a) Corridor slab damage and (b) nonstructural elements failure.
S286 WALLACE ET AL.
and Axis A were observed (Figure 10a), whereas smaller (10 mm) diameter vertical bars in
the short perimeter wall segments at Axis 8 along Axis A fractured (Figure 10a). It is
unclear whether splice failure and fracture of vertical bars led to overturning, or were a
result of overturning (Figure 10b) due to damage observed between axes D and I (Figure 9).
It also is noted that response spectra for a ground motions recorded about 1 km away from
Figure 8. (a) Alto Ro typical floor plan (above first level) and (b) Alto Ro first floor plan.
Figure 9. Alto Ro partial elevations with observed damage: (a) Axis 8, (b) Axis 13, (c) collapse
scenario.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S287
the building (Massone et al. 2012) indicate generally larger spectral demands near the Alto
Ro building, likely due to the impact of soil. More detailed studies are needed to assess this
and other possible collapse scenarios.
DAMAGE ASSESSMENT
Damage observed to RC buildings in Chile raises a number of issues, e.g., what role did
drift demand, axial stress, wall configuration, lap splices, and the lack of well detailed regions
at the wall boundaries play in the degree of damage observed? An initial assessment of these
issues as well as potential code approaches that could be used to address these issues are
presented.
DRIFT DEMANDS
Linear and nonlinear displacement spectra for various ground motions recorded in the
1985 and 2010 earthquakes are presented in Figure 11a, along with simple bilinear spectra for
motions recorded at various locations (simplified SIII) and in downtown Concepcin about
1 km from the Alto Rio building on soft soils (simplified SIV). The simplified (bilinear)
spectra are used to estimate building drift demands, which are plotted in Figure 11b for build-
ings greater than 5 stories using fundamental building periods of T
1
N20

2
p
, where
N20 represents the fundamental period for low-amplitude vibrations (Wood et al. 1987,
Midorikawa 1990) and

2
p
accounts for concrete cracking for larger amplitude vibrations.
Roof displacement is estimated from the spectra by multiplying the spectral displacement
(of the single-degree-of-freedom oscillator) by 1.25 for five-story buildings, 1.5 for buildings
greater than ten stories, and by using linear interpolation between five and ten stories (similar
to the approach used in ASCE-41). A story height of 2.75 m is assumed, and spectral dis-
placements for both SIII and SIV simplified spectra are multiplied by 1.25 to estimate spectra
for 2% damping, which provide a better estimate of inelastic drift (Shibata and Sozen 1976).
Figure 10. (a) Wall damage at ground line, Axes A and 8; (b) overall view of collapsed building.
S288 WALLACE ET AL.
The resulting roof drift ratios range from about 0.8% drift for five-story buildings up to a peak
value of about 1.0% for buildings between 10 and 30 stories for the simplified SIV spectrum.
The drift at failure for compression-controlled walls is estimated to identify building
heights with vulnerable walls for both SIII and SIV simplified spectra as:
u
h
w
1140

max
h
w
Nh
s
, where the constant 11/40 is based on a linear increasing distribution
of lateral forces over the height of the wall (Wallace and Moehle 1992),
max
is estimated
equal to 0.004l
w
for compression-controlled walls, l
w
is the wall length, h
s
is the story
height, and N is the number of stories. In general, compression-controlled walls in 10- to
15-story buildings (SIII) and 10- to 20-story (SIV) buildings would be vulnerable to com-
pression failure. Walls in buildings below ten stories typically have axial load P 0.10A
g
f
0
c
and thus are unlikely to be compression-controlled.
THE ROLE OF AXIAL STRESS
Neither the Chilean (NCh433.Of96; INN 1996) nor the ACI code (ACI 318-08 2008) put
a limit on the level of axial stress allowed for gravity load or combined gravity and lateral
loads, although a limit of P
u
< 0.35P
0
was incorporated into UBC-94. As noted by Massone
et al. (2012), median axial wall stress has increased from about 0.1A
g
f
0
c
for pre-1965 con-
struction and 0.2A
g
f
0
c
for post-1980 construction. For taller buildings (15 to 25 stories), thin-
ner walls, and walls with larger tributary areas, axial load ratios of 0.3A
g
f
0
c
to 0.4A
g
f
0
c
are
possible (Massone et al. 2012).
To assess the impact of axial stress on wall deformation capacity, moment-curvature
relations (Figure 12a) were calculated for a typical wall web in a 12-story building in
Santiago. The wall cross section is 7 m 0.15 m with 8-25 mm diameter bars at each
wall boundary and 8 mm vertical web bars at 20 cm spacing. Two levels of axial stress
are considered, 0.2 and 0.3A
g
f
0
c
; ratios of cl
w
, or neutral axis depth to wall length, also
are plotted (Figure 12b). The relations plotted reveal negative slope (likely to produce
damage concentration) beyond the yield curvature with eventual concrete crushing and
Figure 11. (a) Displacement spectra and (b) estimated drift demands.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S289
relatively low curvature capacity (it is noted that the buckling of vertical bars is not consid-
ered). The plot of cl
w
indicates that this ratio is never less than approximately 0.45 for
P 0.3A
g
f
0
c
. It is noted that ACI 318-99, and subsequent editions, require Special Boundary
Elements for cl
w
> 0.24 (Wallace and Orakcal 2002).
ACI 318-08 (2008) requires that transverse reinforcement (A
sh
) at wall boundaries satisfy
Equation 21-5 A
sh
0.09sb
c
f
0
c
f
yt
, where s is vertical spacing of transverse reinforcement,
b
c
is the dimension of the confined core, f
0
c
is the concrete compressive strength, and f
yt
is
the yield stress of the transverse reinforcement. ACI 318-08 Equation 21-4, A
sh
0.3sb
c
f
0
c
f
yt
A
g
A
ch
1, where A
g
A
ch
is the ratio of the gross area to the confined core at
the wall boundary, is based on equating the pre- and post-spalling axial load for columns
using a simple model that accounts for the stress increase due to confinement of the column
core. This equation does not need to be satisfied at wall boundaries, although it was required
prior to ACI 318-99. For thin walls, the ratio of concrete cover to wall thickness is large, often
in the range of 0.2 to 0.3. In such cases, spalling of concrete cover results in substantial loss of
axial load capacity at a wall boundary, possibly overstressing wall boundary vertical rein-
forcement (and some web vertical reinforcement), resulting in abrupt strength loss due
to buckling of reinforcement. These observations suggest studies are needed to consider
whether Equation 21-4 should be reinstated, or whether relatively thin walls require even
more stringent detailing to adequately confine the core concrete and to restrain buckling
of reinforcement (i.e., the transverse reinforcement required by ACI 318-08 21-4 might
not be adequate). Alternative means to address this issue might be to limit the ratio of
cover to wall thickness or to specify a minimum wall thickness.
Wall lateral stability failures were observed for walls with apparent high axial stress
(Figure 3a, 3b) suggesting that it might be prudent to incorporate a minimum wall thickness
as a function of the unsupported wall height, e.g., t
w
h
s:
, where h
s
= unsupported wall
(story) height. The value used for might depend on the level of axial stress, neutral axis
depth, or expected maximum extreme fiber compressive strain; commonly suggested values
for range from 1/10 (Moehle et al. 2011) to 1/16 (UBC 1997).
Figure 12. (a) Moment - curvature relations and (b) neutral axis depths.
S290 WALLACE ET AL.
THE ROLE OF WALL CONFIGURATION
Common floor plans used for buildings in Chile, with long corridor walls in one direc-
tion and perpendicular walls in the transverse direction (sometimes referred to as a backbone
and rib pattern, or a fishbone pattern), results in buildings where lateral force resistance
is provided by walls with T- and L-shaped cross sections. Prior research on walls with
T-shaped cross-sections has revealed that these walls behave substantially different than
rectangular walls with symmetrically placed longitudinal (vertical) reinforcement (Wallace
1994, 1996; Thomsen and Wallace 2004). For a wall with a T-shaped cross section, the web
boundary is subjected to both large tensile and compressive strains (e.g., 0.025 in tension
and 0.01 in compression for TW2 at 1.5% lateral drift; Orakcal and Wallace 2006); there-
fore, this region typically must be well confined to avoid concrete crushing and reinforce-
ment buckling.
Load versus displacement results are presented for similarly detailed web boundaries for
walls TW1 and RW2 (Figure 13) tested by Thomsen and Wallace (2004). Significant loss in
lateral-load capacity was observed for wall TW1 at a little greater than 1% lateral drift when
all eight vertical boundary bars and some vertical web bars buckled. In contrast, wall TW2
(Figure 13, photo) reached lateral drift ratios in excess of 2.5% before lateral strength degra-
dation initiated. For wall TW2, strength loss resulted due to lateral instability (buckling) of
the well-confined web boundary, after concrete cover spalled. Lateral instability failures and
fracture of wall web boundary vertical reinforcement (Figure 14) suggest that the lack of
transverse reinforcement at web boundaries of T- and L-shaped walls could have played a
Figure 13. Load-displacement relations for TW1 and RW2.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S291
significant role in the level of damage observed to walls in Chile and existing U.S. code
requirements should be carefully reviewed.
DISPLACEMENT-BASED DESIGN PROVISIONS AND MODEL ASSUMPTIONS
Sections 21.6.2 of ACI 318-99 and 21.9.6.2 of ACI 318-08 include provisions that are
derived from a displacement-based approach (Moehle 1992, Wallace and Orakcal 2002).
In the model used to develop ACI 318-99 provisions, the design displacement
u
is related
to local plastic hinge rotation (
p
) and extreme fiber compressive strain (
cu
) as:
EQ-TARGET;temp:intralink-;e1;50;296
p


u
h
w

u


cu
c

l
p

l
w
2

(1a)
EQ-TARGET;temp:intralink-;e2;50;246
cu
2

u
h
w

c
l
w

(1b)
If the compressive strain exceeds a limiting value, taken as 0.003 in ACI 318-08, then
special boundary elements are required. In ACI 318-08, this approach is modified to define a
limiting neutral axis depth (instead of a limiting concrete compressive strain) as:
EQ-TARGET;temp:intralink-;e3;50;166 c
ACI;limit

0.003l
w
2
u
h
w


l
w
667
u
h
w

l
w
600
u
h
w

(2)
In this approach, it is obvious that the result is sensitive to the values used for design
displacement
u
and plastic hinge length l
p
, where it is assumed that yielding spreads out over
Figure 14. Rebar fracture at web boundaries of T-shaped walls in Concepcin: (a) Centro Mayor
building and (b) Alto Ro Building @ 13-A (Figure 8).
S292 WALLACE ET AL.
a height (plastic hinge length) of l
p
l
w
2. Despite fairly low drift demands (Figure 11),
significant wall damage was observed (Figure 1) and it was evident that inelastic deforma-
tions did not spread out in poorly detailed, highly compressed walls commonly used in Chile;
therefore, it is important to reassess the ACI (and the ASCE 7) provisions in light of these
observations.
The approach presented by Wallace and Orackal (2002) is modified here to assess the
potential impact of variation of the plastic hinge length on the need for Special Boundary
Elements (SBEs), that is, the regions at wall edges where closely-spaced transverse reinfor-
cement is needed to provide nonlinear deformation capacity by ensuring a stable compression
zone (and thus, adequate spread of plasticity over the wall height). Given that the ACI 318-08
relation assumes a specified spread of plasticity l
p
l
w
2, the impact of concentrating
damage over shorter height was investigated by modifying the relationship presented by
Wallace and Orakcal (2002) to use plastic hinge length equal to a multiple of the wall thick-
ness l
p
t
w
:
EQ-TARGET;temp:intralink-;e4;71;472

u
h
w

cu

t
w
l
w
l
w
c

2
t
w
h
w


y
l
w

11
40
h
w
l
w

t
w
l
w


2
2
t
w
h
w
t
w
l
w

(3)
Where t
w
is the wall thickness, c is the neutral axis depth, h
w
is the wall height, l
w
is the
wall length, and
y
is the yield curvature of the section. In this study, the yield curvature
is estimated as
y

sy
l
w
c; alternatively, yield curvature can be estimated as
1.5 to 2.0
sy
l
w
, where
sy
is the tensile reinforcement yield strain. The constant 11/40 is
based on a linear increasing distribution of lateral forces over the height of the wall (Wallace
and Moehle, 1992). For this preliminary study, wall aspect ratio h
w
l
w
is set to 10 (for a 20
story building; h
w
l
w
50m5m) and the ratio of l
w
t
w
is set to 25 for Chilean buildings
(5m0.2m). Concrete compressive strain is set to 0.003 (the value that defines when SBEs
are required by ACI 318-08). Results are presented in Figure 15a for a plastic hinge length of
l
p
t
w
, with set equal to 2, 6, and 12. Results for the ACI 318-08 model also are shown; it
is noted that the ACI model results are different than those produced with Equation 3 because
the ACI model neglects elastic deformations. For l
p
12t
w
, if the drift ratio is about 0.015,
Figure 15. Impact of plastic hinge length on SBE variable (a) cl
w
, (b) h
w
l
w
.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S293
the neutral axis must exceed 0.15l
w
before SBE are required. However, for the same neutral
axis depth, 0.15l
w
, if inelastic deformations are concentrated over l
p
2t
w
, a drift ratio of
only about 0.008 can be tolerated before SBEs are required. In the limit, if approaches zero
(a compression-controlled wall, where the extreme fiber concrete compressive stress reaches
0.003 prior to yield of tensile reinforcement at the opposite wall boundary, i.e., no spread of
plasticity), then only the elastic drift is reached, assuming rebar buckling does not occur prior
to reaching the yield displacement. For equal to zero and yield curvature of 2.0
sy
l
w
, the
resulting elastic drift is:
u
h
w
0.01h
w
l
w
0.0005N, for h
w
N2.5 m and l
w
5 m.
Elastic drift ratios are plotted on Figure 11b and suggest that compression controlled walls in
buildings between roughly 10 and 20 stories are more susceptible to damage. This result
is consistent with observed damage. Walls in buildings less than ten stories are likely to
have lower levels of axial stress < 0.10A
g
f
0
c
and are less likely to be compression-
controlled, which is consistent with the performance of buildings in Via del Mar in the
1985 earthquake.
Variation of the wall aspect ratio h
w
l
w
is shown in Figure 15b for walls where
plasticity spreads out over the ACI assumed plastic hinge length of l
p
l
w
2. Results
for l
p
2t
w
(not shown) are similar to that for well-detailed walls with large c/l
w
values
(large axial load). Figure 15b includes aspect ratios ranging from 5 to 20, or 10 to 40 stories,
respectively. Results indicate that buildings over 30 stories are less prone to damage even for
relatively large cl
w
values, for example, due to large axial load and/or large flexural com-
pression force, since drift levels greater than about 2% are required in most cases to require
SBEs and anticipated drift demands are generally less than 1.0% (Figure 11b). Figure 15b
also shows that well-detailed walls with small ratios of cl
w
values can easily achieve
1% roof drift ratios. Poorly-detailed walls or walls with large axial load (large cl
w
values)
for relatively short buildings (e.g., ten stories or less) require lateral drift ratios of about
0.5% to 1% before SBEs are required per ACI 318; however, displacement demands for
these shorter, stiffer buildings (fundamental period of approximately 0.5 sec) are only
about one-half of that needed to require SBEs. These observations again help explain
the higher concentration of damage in buildings with poorly detailed walls in the range
of 15 to 20 stories.
As noted previously, the design displacement for the displacement-based design
approach for shear walls in ACI 318 is obtained using ASCE 7 provisions as:
EQ-TARGET;temp:intralink-;e5;50;247
u
ACI
x
ASCE C
d

e
I (4)
where
e
is the elastic displacement for cracked section properties reduced by the response
modification coefficient R (equal to 5 and 6 for bearing wall systems and building frame
systems, respectively), C
d
is the deflection amplification factor (equal to 5 for both systems),
and I is the Importance Factor. In current U.S. codes, the intent is to provide 90% confidence
of non-collapse for MCE shaking. In contrast, the current ACI confinement trigger is based
on 50% confidence of not exceeding the concrete crushing limit in the Design Basis Earth-
quake, which is much lower shaking intensity than the MCE.
It is necessary to adjust ACI 318-08 Equation 21-8 to be more consistent with the
building code performance intent. Three factors need to be considered (neglecting the
S294 WALLACE ET AL.
role of C
d
): (1) MCE exceeds DBE; (2) there is dispersion about the median response; (3)
damping is likely to be lower than the 5% value assumed in the ACI provisions (e.g., on the
order of 2% to 3%; ATC-72 2010). To address these issues, the displacement value used in
the denominator of Equation 21-8 in ACI 318-11 should be increased by a factor of approxi-
mately 1.5 to adjust to MCE level shaking and to consider dispersion, and by approximately
1.2 to 1.3 to account for potential lower damping ratios; therefore, for Equation 21-8, either a
multiplier of two should be applied to the ASCE 7-05 displacement, or the coefficient of 600
in the denominator should be approximately doubled to 1,200.
BUCKLING AND FRACTURE OF REINFORCEMENT
Poorly detailed and/or compression-controlled walls, that is, walls that lack closely
spaced transverse reinforcement to sustain a stable compression zone and ensure spread
of plasticity by confining core concrete and suppressing rebar buckling, exhibited poor beha-
vior (Figure 2). The longitudinal boundary bars, typically of 18 mm to 25 mm diameter in
Chilean buildings, are typically enclosed within 8 mmdiameter horizontal web reinforcement
spaced at s 20 cm on center with 90 hooks at the wall boundary (see Figure 2c); therefore,
sd
b
ratios are typically in the range of 8 to 11. Such large sd
b
ratios are likely to result
in buckling of vertical reinforcement following even modest tensile strain excursions, i.e.,
around 0.01 (Rodriguez et al. 1999). Damage appeared to initiate at the wall boundary (and
extend over a height of approximately two to three wall thicknesses), and then propagate
towards the interior of the wall, or towards the wall flange in the case of T-shaped cross
section walls.
Reinforcement at wall boundaries is subjected to large variations in tension and compres-
sive strains when subjected to earthquake (reversed cyclic) loading. Compressive strains are
large, especially for large axial stress and for web boundaries of walls with T-shaped cross
sections (Figure 16), potentially leading to compression failure. Alternatively, if reinforce-
ment is subjected to large tensile strain demands (exceeding yield), cracks open, and upon
reverse loading, all compression must be resisted by reinforcement, potentially leading
to buckling failure. As noted previously, due to larger variation in the tension and compres-
sive strain demands (Figure 16), web boundary longitudinal reinforcement for walls with
T-shaped cross sections are much more susceptible to buckling than are boundary longitu-
dinal reinforcement in walls with rectangular cross section (for the same axial stress ratio;
e.g., see Figure 13, Thomsen and Wallace 2004).
Given the typical wall configurations used in Chile (corridor and transverse walls), as
well as the lack of closely spaced transverse reinforcement at wall boundaries, it is not sur-
prising that damage at web boundaries of walls with T-shaped cross sections was common
(Figure 16). However, the quantity of transverse reinforcement required by ACI 318-08 does
not ensure that the post-spalling axial strength of the core concrete is sufficient to sustain the
axial load demand. In addition, relatively few tests have been conducted on T-shaped walls,
or walls with flanges, and in the tests where ductile behavior was observed, e.g., TW2
(Figure 13), the sd
b
ratio was substantially less than the limiting value (3.33d
b
versus
6d
b
). The findings suggest that more stringent detailing may be needed to ensure ductile
behavior at T-shaped walls.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S295
CONCLUSIONS
The M
W
8.8 earthquake that struck Chile on 27 February 2010 provides a excellent
opportunity to study the performance of reinforced concrete buildings designed using a
modern seismic code provisions and concrete design based on ACI 318-95. Based on recon-
naissance efforts and subsequent preliminary studies, a number of issues have been identified
to help us understand the observed damage, particularly the damage to shear walls, and to
identify areas where changes to ACI 318-08 may be warranted. Based on this preliminary
work, the following observations are noted:

Pre-1985 buildings in Chile typically performed well because of the large ratios of
A
W
A
f
(stiff buildings), relatively low wall axial stress (typically < 0.10A
g
f
0
c
), and
relatively low displacement demands.

Post-1985 buildings, and particularly buildings constructed since 2000, shear walls
tend to have larger axial stress ratios and larger roof drift ratios, particularly on soft
soils (SIV). For compression-controlled walls, where the concrete compressive
strain reaches 0.003 prior to yield of tension reinforcement, simplified estimates
of drift demands and drift capacities indicated that compression-controlled walls
in 10-15 and 10-20 story buildings were susceptible to failures for SIII and SIV
soils, respectively. These findings are generally consistent with observed damage.

Post-2000 buildings in Chile commonly include central corridor walls and multiple
transverse walls, creating T-shaped walls, which are susceptible to web boundary
damage due to the large reinforcement tensile strains accompanied by large concrete
Figure 16. Web boundary strain demands for T-shaped wall.
S296 WALLACE ET AL.
compressive strains that develop at web boundary. For poorly-detailed web bound-
aries, sudden loss of lateral strength due to buckling of vertical boundary and web
bars has been observed in tests and likely was a significant factor in the degree of
damage observed following the 2010 earthquake (along with axial stress).

The displacement-based design approach used in ACI 318-08 assumes a plastic


hinge length of l
p
0.5l
w
to determine the need for special boundary elements;
however, damage was concentrate over a much shorter distance, typically over a
height equal to two to three times the wall thickness. Sufficient transverse reinforce-
ment must be provided to ensure a stable compression zone at wall boundaries to
achieve the spread of plasticity assumed.

In current U.S. codes the intent is to provide 90% confidence of non-collapse for
MCE shaking. In contrast, the current ACI 318-08 Equation 21-8 confinement trig-
ger is based on 50% confidence of not exceeding the concrete crushing limit in the
Design Basis Earthquake (which is much lower shaking intensity than the MCE and
is based on an assumed 5% damping ratio). To address these factors, the displace-
ment used in Equation 21-8 should be increased; a factor of two is suggested.

Limits on story height to wall thickness h


s
t
w
, such as the limit of 16 used in the
1997 Uniform Building Code, should be considered to reduce the likelihood of wall
lateral instability failures. Use of a lower ratio might be appropriate for web bound-
aries of flanged walls, where large cyclic tension and compressive strains occur.
ACKNOWLEDGEMENTS
The authors would like to individuals from three Chile universities who contributed gen-
erously to the EERI team reconnaissance include:

From the Federico Santa Maria Technical University: Carlos Aguirre and Arturo
Milln

From the Pontificia Universidad Catlica de Chile: Professors Juan Carlos De La


Llera, and Matias Hube, with Vicente Arizta, Juan Jose Besa, Alvaro Carboni,
Claudio Frings, Juan Pablo Herranz, Cristbal Moena, Rodrigo Oviedo, Victor
Sandoval, Nicols Santa Cruz, Csar Seplveda, Benjamin Westenenk

From the Universidad de Chile: Professors Mara Ofelia Moroni, and Rodolfo
Saragoni
All of the members listed provided tremendous assistance, especially the students from
Pontificia Universidad Catlica de Chile and Federico Santa Maria Technical University.
Travel funds were provided by the EERI Learning From Earthquakes program (NSF
CMMI-0758529) and by NEEScomm (NSF CMMI-0927178). Opinions, findings, conclu-
sions, and recommendations in this paper are those of the authors, and do not necessarily
represent those of the sponsor or others mentioned.
REFERENCES
American Concrete Institute (ACI 318-99), 1999. Building Code Requirements for Structural
Concrete (ACI 318-99) and Commentary (318R-99), Farmington Hills, MI, 391 pp.
American Concrete Institute (ACI 318-95), 1995. Building Code Requirements for Structural
Concrete (ACI 318-95) and Commentary (318R-95), Farmington Hills, MI, 391 pp.
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S297
American Concrete Institute (ACI 318-08), 2008. Building Code Requirements for Structural
Concrete (ACI 318-08) and Commentary (318R-08), Farmington Hills, MI, 465 pp.
American Society of Civil Engineers (ASCE 7-10), 2010. ASCE/SEI 7-10: Minimum Design
Loads for Buildings and Other Structures, Reston, VA, 650 pp.
Applied Technology Council (ATC 72-1), 2010. Modeling and Acceptance Criteria for Seismic
Design and Analysis of Tall Buildings, Report No. PEER/ATC-72-1, Applied Technology
Council, Redwood City, CA, 242 pp.
Brown, J., and Kunnath, S. K., 2004. Low-cycle fatigue failure of reinforcing steel bars, ACI
Materials Journal 101, 457466.
Caldern, J. A., 2007. Update on Structural System Characteristics used in RC Building Con-
struction in Chile, Civil Engineering Thesis, University of Chile, 76 pp. (in Spanish).
Comit Inmobiliario (CChC), 2010. Communication based on INE data, Instituto Nacional de
Estadsticas (National Institute of Statistics), http://www.ine.cl/.
Centro de Investigacin, Desarrllo e Innovacin de Estructuras y Materiales (IDIEM), 2010. Peri-
taje Estructural Edificio Alto Ro, Ciudad de Concepcin, Informe Final, Descripcin de
Caida y Factores Asociados al Colapso, Revisin 1, Informe No 644.424-00, University
of Chile (in Spanish).
Instituto Nacional de Normalizacin (INN), 2008. NCh430.Of2008: Reinforced Concrete
Design and Calculation Requirements, Santiago, Chile, 17 pp. (in Spanish).
Instituto Nacional de Normalizacin (INN), 1996. NCh433.Of96: Earthquake Resistant Design of
Buildings, Santiago, Chile, 43 pp. (in Spanish).
Massone, L. M, Bonelli, P., Lagos, R., Lder, C., Moehle, J., and Wallace, J. W., 2012. Seismic
design and construction practices for reinforced concrete structural wall buildings, Earthquake
Spectra 28, this issue.
Midorikawa, S., 1990. Ambient Vibration Tests of Buildings in Santiago and Via del Mar, DIE
No.90-1, Departamento de Ingenieria Estructural, Pontificia Universidad Catlica de Chile,
169 pp.
Moehle, J. P., Displacement-based design of RC structures subjected to earthquakes, Earthquake
Spectra 8, 403428.
Rodriguez, M. E., Botero, J. C., and Villa, J., 1999. Cyclic stress-strain behavior of reinforcing
steel including effect of Bbuckling, J. Struct. Eng., ASCE, 125, 605612.
Shibata, A., and Sozen, M. A., 1976. Substitute structure method for seismic design in R/C,
Journal of the Structural Division, ASCE, 102, 118.
Thomsen, J. H., and Wallace, J. W., 2004. Displacement-based Design of slender RC structural
walls experimental verification, J. Struct. Eng., ASCE, 130, 618630.
Uniform Building Code (UBC-97), 1997. Uniform Building Code, International Conference of
Building Officials, Whittier, California, 492 pp.
Wallace, J. W., 1995. Seismic design of RC shear walls; Part I: New code format, J. Struct. Eng.,
ASCE, 121, 7587.
Wallace, J. W., 1996. Evaluation of UBC-94 provisions for seismic design of RC structural walls,
Earthquake Spectra 12, 327348.
Wallace, J. W., and Moehle, J. P., 1989. The 3 March 1985 Chile Earthquake: Structural
Requirements for Bearing Wall Buildings, EERC Report No. UCB/EERC-89/5, Earthquake
Engineering Research Center, University of California, Berkeley.
S298 WALLACE ET AL.
Wallace, J. W., and Moehle, J. P., 1992. Ductility and detailing requirements of bearing wall
buildings, J. Struct. Eng., ASCE, 118, 16251644.
Wallace, J. W., and Moehle, J. P., 1993. An evaluation of ductility and detailing requirements
of bearing wall buildings using data from the 3 March 1985 Chile earthquake, Earthquake
Spectra 9, 137156.
Wallace, J. W., and Orakcal, K., 2002. ACI 318-99 provisions for seismic design of structural
walls, ACI Structural Journal 99, 499508.
Wallace, J. W., and Thomsen, J. H., 1995. Seismic design of RC shear walls; Part II: Applica-
tions, J. Struct. Eng., ASCE, 121, 88101.
Wood, S. L., Wight, J. K., and Moehle, J. P., 1987. The 1985 Chile Earthquake, Observations on
Earthquake Resistant Construction in Via del Mar, Civil Engineering Studies, Structural
Research Series No. 532, University of Illinois, Urbana, 176 pp.
(Received 26 July 2011; accepted 14 April 2012)
DAMAGE AND IMPLICATIONS FOR SEISMIC DESIGNOF RC STRUCTURAL WALL BUILDINGS S299

Vous aimerez peut-être aussi