Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Journal of Building Engineering 41 (2021) 102766

Contents lists available at ScienceDirect

Journal of Building Engineering

Nonlinear modeling of a damaged reinforced concrete building and design


improvement behavior
Leonardo M. Massone a, *, Eduardo Bedecarratz b, Fabián Rojas a, Mario Lafontaine b
a Department of Civil Engineering, University of Chile, Blanco Encalada, 2002, Santiago, Chile
b Rene Lagos Engineers, Magdalena 140 of, 2402, Santiago, Chile

ARTICLE INFO ABSTRACT

Keywords: Nonlinear modeling is performed for a reinforced concrete building that presented flexural damage during the
Nonlinear model 2010 Chile earthquake. The focus of the analysis is the behavior of a wall on the first floor that presented local-
Shear wall ized damage; the wall is T-shaped, has a web length of 8 m, and a flange length reaching 18 m. An unconfined
Building
concrete model and a steel model incorporating buckling were considered, accounting for the poor detailing of
Earthquake
the wall boundary element. The walls were modeled using nonlinear fiber elements. A rigid diaphragm was con-
Damage
sidered, and the slabs were incorporated only in areas where they could generate coupling effects in the walls,
being modeled as beams with plastic hinges at their ends. Nonlinear static and dynamic analyses were performed
on the building model, utilizing records for Santiago from the 2010 Mw 8.8 earthquake. For relatively low lateral
drift values (~0.003), the strain levels observed in the model for the wall suffering damage would lead to flexural
damage due to its large size, large amount of longitudinal steel reinforcement in the flange, and poor detailing of
the web boundary. Several other nonlinear models were created to compare the impact of the design improve-
ments on the overall performance of the building. From the results, it can be seen that the observed damage
would be avoided by implementing design solutions such as increased wall thickness, confinement, or decou-
pling of the main wall.

1. Introduction buildings present a coupling effect between the walls and the slab. This
is an important characteristic of Chilean buildings which is also com-
The most common structural systems used in mid-to high-rise resi- mon in buildings in some countries in Europe and Asia, providing rigid-
dential buildings in Chile to resist the lateral forces produced by earth- ity to the system, resulting in small lateral drift demands. In addition,
quakes are Reinforced Concrete (RC) Wall systems. These walls provide similar to other in Latin American countries, the design of RC walls in
the necessary stiffness and strength to withstand the loads produced by Chile is based on the ACI 318 building code requirements [3], including
strong ground motions accompanied by small lateral drift. Typically, some modifications to adapt the design to the local practice. Although
Chilean buildings are built with wall thicknesses around 20 cm, steel the design of RC walls is relatively straight forward when following cur-
ratio at the boundary of the wall (over the boundary area) of 5% on av- rent codes, including ACI 318, the behavior of RC walls under seismic
erage for a boundary length of about 10% of the wall length [1], and a loads is very complex because the walls behave differently depending
density of walls (wall area in one direction divided by the total area of on their configurations (wall size, cross-section geometry, wall height/
the floor) in both directions of 2~4% [2]; this is also typically the case length ratio, steel reinforcement, boundary detailing, and wall open-
for residential buildings with different numbers of stories. The struc- ings/discontinuities) and loading conditions, as was observed during
tural walls are located in the interior and perimeter of the building for the 2010 Chile earthquake [4–12].
the two principal directions of the building, since the walls are also The 2010 Mw 8.8 earthquake in Chile provided a unique opportu-
used to carry the vertical loads produced by the dead and live loads nity to study the behavior of RC walls in buildings that were designed to
over the buildings, generating a high level of axial load in the wall of updated building codes and subjected to strong ground motions that re-
tall buildings at the ground level during earthquakes. Also, the Chilean sulted in damage. The most common type of earthquake-related dam-

Corresponding author.
*

E-mail addresses: lmassone@ing.uchile.cl (L.M. Massone), ebedecarratz@renelagos.com (E. Bedecarratz), fabianrojas@uchile.cl (F. Rojas),
mlafontaine@renelagos.com (M. Lafontaine).

https://doi.org/10.1016/j.jobe.2021.102766
Received 29 October 2020; Received in revised form 13 May 2021; Accepted 19 May 2021
Available online 27 May 2021
2352-7102/© 2021 Elsevier Ltd. All rights reserved.
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

age to residential RC buildings (~20 stories) was generally observed at linear response by a factor that depends on the structural system, then
the ground level of the walls with reduced length and T-shaped cross- relating this to the curvature demand on the walls by expressions that
sections due to the high levels of axial load, flexion-compression ac- typically neglect the elastic displacement capacity; finally, the strains
tions, and no confinement in the boundary elements; this caused on the critical sections are obtained by a sectional analysis. In PBD,
spalling of the concrete cover, crushing of the concrete, and buckling however, both the global lateral roof displacements of the building and
and fracture (in some cases) of the steel bars in the wall boundaries, as the local strain demands (and therefore the potential damage) on the
reported in many studies [13–17]. To study this damage observed on shear walls are directly obtained using a nonlinear model that captures
RC walls during the 2010 earthquake, some numerical studies of such the inelastic response of the building for strong earthquakes. Therefore,
buildings using different models have been performed in recent years the PBD methodology is more appropriate for evaluating the impact of
[18–22]. These investigations were able to represent to some extent the the Chilean seismic code changes on the seismic performance.
damage observed in the buildings. One of the main PBD guidelines is the document published by the
In addition, the poor performance of some structures during the LATBSDC [24]; this was used in this study as a reference for the nonlin-
earthquake highlighted the need to improve the design of RC wall sys- ear building model (and not to seek its performance objectives). Models
tems in Chile. The changes in the design provisions have included es- of buildings using nonlinear elements to determine the failure mode ob-
tablishing a maximum level of the axial load over the RC walls to served in the 2010 Chile earthquake are studied in this work. In addi-
0.35f′cAg (f'cAg is the concrete wall section strength), requiring confine- tion to demonstrating the ability of commonly used nonlinear element
ment of the concrete of wall boundary elements if the compressive formulations to simulate the response of the building using static and
strain is larger than 0.003, and limiting the maximum compressive dynamic analysis, three alternative design solutions are analyzed to val-
strain in the concrete to 0.008 [23]. However, existing studies of dam- idate the improvement in the building response.
aged buildings have not evaluated the effect of incorporating these
changes in the analysis and have not utilized methodologies of Perfor- 2. Description of the building
mance-Based Design (PDB) to evaluate the design and behavior of these
buildings. 2.1. General description
Performance-Based Design is a design methodology that enables the
estimation of reliable building seismic performance. This is achieved by The focus of this study is a residential building located in the Ñuñoa
nonlinear modeling of the structural components as opposed to the typ- district of Santiago, Chile. Its structure incorporates RC walls, it has 23
ical linear models used in the prescriptive code design approaches. This total floors (2 of which are part of the basement), and has a total height
allows the generation of useful data for the seismic design of buildings of 62.8 m from the foundation level and 59.4 m from the ground level.
that are indirectly estimated in linear models, including the roof dis- It is founded on a dense gravel soil that, according to current Chilean
placements that the earthquake demands on the building, the curvature regulations, is classified as Type B or C soil (Vs30 ≥ 500 m/s for soil B
demand on the walls (particularly on the plastic hinge zones), and the and Vs30 ≥ 350 m/s for soil C) [25]. A nearby measurement of Vs30 for
actual concrete compressive and rebar tensile strains, which are vari- a different building provided an average value of 470 m/s, classifying
ables that can be directly correlated with structural damage [11,22]. the soil type as C. The floor plan of the building is typical of classic
This information allows the structural engineer to limit the damage ex- Chilean residential buildings—a high density of walls distributed
pected in the building (e.g., tensile/compressive strains), already de- throughout the floor to separate apartments and rooms, with a series of
signed under earthquake loads, by the control of the level and type of combined cross-sections, such as C-, T- and L-shapes, or even more com-
incursion of the nonlinear response and also to catch any undesired be- plex sections (Fig. 1). The second basement is set back with respect to
havior (e.g., fragile failure such as shear failure), which is only possible the first basement and the adjacent walls are staggered to reach the ap-
to evaluated under nonlinear-static and time-history analyses. An itera- propriate level (Fig. 2).
tive procedure might be required to comply with damage limits redefin- The thickness of the walls is 20 cm in most areas. The perimeter
ing the structural system or element dimensions. In prescriptive code beams have a height of 70 cm and a width of 20 cm. The slabs are
design, inelastic displacement values are estimated by multiplying the 16 cm thick in the typical floor, and 18 cm in the basement, with two

Fig. 1. First level floor plan – (a) entire floor, (b) complete section of analysis of wall axis 7.

2
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 2. First underground floor plan.

elevator openings of around 4 m by 2.7 m on each floor and stairs open- Table 1
ing of 3.3 m by 3.95 m on the typical floor and around 1.75 m by Design parameters.
3.95 m in the basement (Figs. 1 and 2). The height between floors is Parameter Description Direction Direction
2.5 m in the typical floor, 2.7 m in the first floor, 2.85 m in the first X Y
basement and 3.85 m in the second basement. The floor plan area of the
W [tonf] Seismic weight, defined as the sum of total dead 19,473
typical floor is 858 m2, with a wall area over floor area of 2.5% in both load and 25% of total live load.
directions. In addition, the building has a global slenderness ratio H/ Qe [tonf] Elastic base shear. 5250 4717
B = 59.4/22.05 = 2.7 (height over short plan width) measured from (27% W) (24% W)
the building basement. R* [−] Force modification factor according to NCh433. 8.5 8.9
Qe/R* Modified base shear. 617 532
[tonf] (3.1% (2.7%
2.2. Design W) W)
Qmin Minimum base shear according to NCh433. 974 (5% 974 (5%
The original design of the building was performed using a linear [tonf] W) W)
model and based on the application of the NCh433 [26] standard valid Qmax Maximum base shear according to NCh433. 2046 2046
[tonf] (10.5% (10.5%
before the year 2010. At that date, the design was oriented with a
W) W)
strength perspective, with little attention given to boundary wall detail- Qeff Effective base shear (controlled by minimum 974 (5% 974 (5%
ing. According to the design criteria, the building was based on a con- [tonf] shear). W) W)
crete compressive strength of 25 MPa for the first 14 levels, and 20 MPa R** [−] Effective force modification factor. 5.4 4.8
for the remaining levels. The yield stress of the reinforcing steel used Note: 50 modes were considered to achieve a total mass participation of 99.7%
was 420 MPa. A modal analysis procedure was performed for the force in Y direction, and 99.9% in X direction.
and moment determination. The design parameters summarized in
Table 1 include the elastic base shear (Qe), the force modification factor buildings in Chile were not required to have special boundary detailing
(R*), the minimum (Qmin) and maximum (Qmax) base shear, the effec- in walls due to the good performance observed in RC buildings during
tive shear (Qeff), the effective force modification factor (R**) and the the 1985 earthquake. The elevation geometry of the wall of axis 7 is
seismic weight. It can be seen that the shear force design corresponded quite regular, except for a setback at the second basement level.
to 5% of the seismic weight.
3. Nonlinear building model
2.3. Building damage
3.1. Element models in Perform 3D
Due to the 2010 earthquake, the building had serious flexural-axial
damage on a wall located on axis 7 of the first floor (Fig. 1). The wall The nonlinear building model is developed in Perform 3D [27] (Fig.
presents local damage with concrete cover spalling and buckling of its 5). In this work, the Shear Wall Element [27] is used for the modeling of
main longitudinal boundary and distributed vertical reinforcement due RC shear walls. This type of element is particularly useful for modeling
to the lack of proper detailing (Fig. 3). The damage zone propagates solid walls without irregular openings. The wall element is defined with
4 m inwards, measured from the tip of the wall web (Fig. 1 – ex- four nodes, each with six degrees of freedom (rotation and translation
treme + Y end of axis 7). This is the only severe damage to the build- in the three directions), and can be connected to other elements to gen-
ing. erate a three-dimensional wall system model. A nonlinear bending-
The reinforcement details of the damaged wall on the first floor are axial and a linear shear formulation in the longitudinal direction are in-
shown in Fig. 4. A large concentration of reinforcement can be seen in corporated in the element. In the transverse direction and out-of-plane
the wall flange, and the tip of the wall web does not present a confine- deformation of the element, its behavior is assumed to be linear. Bend-
ment detailing according to the current ACI 318 code [3], since older

3
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 3. Damage of the wall of axis 7.

Fig. 4. Reinforcement details of the wall of axis 7 on the first floor.

ing-axial behavior in the longitudinal axis (typically vertical) is charac- tural engineers take all the necessary design precautions and considera-
terized by fibers. The sectional behavior is obtained by integrating the tions for these irregularities to work correctly under earthquake ac-
individual behavior of each fiber (concrete and steel) assuming the tions, such as reviewing the force path with strut-and-tie models.
Bernoulli hypothesis, where the curvature along the element remains The beams are frame-type elements and are defined through two
constant. In addition, regarding the meshing of the elements, it is rec- nodes, each with six degrees of freedom. The beam consists of an elastic
ommended to reduce the height of the wall elements at the base, so that section that covers the span of the beam, plastic hinges concentrated at
the plastic deformation can be concentrated in a suitable element size. the ends of the section, and a rigid end zone. The plastic hinge has a
As recommended by FEMA 356 (2000) [28], the height of the element zero-length formulation and is defined through a moment-rotation
in the plastic zone should be the lowest of (a) half the length of the wall curve. For this work, a perfectly rigid-plastic response for the plastic
and (b) the story height. In this work, half of the story height was con- hinge was considered. A summary of the elements used is provided in
sidered for a better discretization, which is equivalent to 1.25 m as Table 2.
shown in Fig. 6. In general, there are no major discontinuities that need Nonlinear analysis can involve a large investment of computational
to be considered, except for the wall setback observed at the basement resources, often necessitating the analysis of simplified models without
of the building to accommodate car parking space. In addition, experi- a fine mesh or with irrelevant elements.
mental and analytical work [5,7,29] have shown that such a disconti- The main simplifications considered in the model are listed below:
nuity causes strain concentration at the discontinuity that is well cap-
tured by fiber-type elements, such as the one used here for walls. It is ● Non-refined meshing of shell elements: This reduces the analysis
important to mention that this type of discontinuity or any type of clas- time of the nonlinear model as there are fewer elements and it also
sification or penalties due to vertical or plan irregularities are not in- simplifies the fiber definition. The meshing of the wall of axis 7, in
cluded or assessed or prohibited in the Chilean design code, as it is done particular, is presented in Fig. 6. For the first four stories (from the
in other seismic design codes. It is common practice in Chile that struc- 2nd basement to the 2nd floor) a vertical mesh is proposed for a

4
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 5. Global view of the original model in Perform 3D – (a) general case, (b) first mode (torsional), (c) second mode (mostly translational in Y axis), (d) third mode
(mostly translational in X axis).

proper normal strain distribution, and the horizontal mesh is ● Rigid diaphragm constraints are assigned at each level. This
conditioned by the intersections with other elements. Similar assumption is reasonable, given the small slab openings due to the
criterion is applied to other walls, where in most cases, element staircase and elevators (4% of floor plan area) and the absence of
lengths vary from 1 m to 3 m. damage in the slab due to in-plane movement.
● Removal of slab elements: In Perform 3D, slab elements do not ● Fixed-base restrictions are applied.
have nonlinear properties, and incorporating linear slabs would
cause exaggerated and unrealistic coupling effects. As an The three fundamental periods of the linear model in Perform 3D
alternative, beam elements are modeled between walls (not are: 1.57 s, with a marked torsional response, 1.20 s, mostly transla-
crossing openings in slabs) to simulate coupling due to slabs. The tional in the most flexible direction (Y), and 1.11 s, also translational
modeled beams have a height equivalent to the thickness of the but in the X-direction, as shown in Table 3. The mass participation of
slab, and a width equivalent to 12 times the slab thickness [27]. the three modes is also reported in Table 3.
These elements are arranged to enable wall-slab coupling in the X
and Y directions. The distribution of these elements in the floor 3.2. Material models
plan can be seen in Fig. 7.
● No stiffness reduction factors are included, given that all relevant The concrete envelope curves were obtained from the models of
non-linearity is captured in beams (slabs) and wall elements in Saatcioglu and Razvi (1992) [31], considering in the confined case a
flexure. Gravity loads can reduce the slab flexural stiffness, but the transversal boundary detailing reinforcement equivalent to stirrups
effect of thin slabs in Chilean buildings is relatively small [4]. ϕ12 spaced at 100 mm, constituting a typical detailing, together with
● The over-strength of concrete and steel is considered. In this case, over-strength of the material. The resulting curves, incorporated into
the yield strength of steel is increased to 1.17fy and concrete the model, are shown in Fig. 8.
strength to 1.3f′c [24]. The steel curve is initially defined based on the monotonic model by
● For the implicit dynamic analysis, Rayleigh's damping is Menengotto and Pinto (1973) [32]. However, since there is evidence of
considered with a damping value of 2.5% on the first and third buckling of the longitudinal reinforcement, the loss of strength due to
vibration modes [24,30]. The building is loaded with the total dead this phenomenon is incorporated. As shown in Fig. 3, the buckled rein-
load plus the expected live load (25%) before the seismic action is forcing bars in the damaged wall are completely exposed due to con-
applied. crete spalling. Buckling occurs when the bars are not properly re-

5
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 7. Distribution of beams as a replacement for slab in the typical floor plan.

Table 3
Natural periods, vibration modes and participating mass ratios.
Modes Period Direction Participating Participating Participating
[s] Mass Ratio Mass Ratio Rotational Mass
in X in Y Ratio in Z

1 1.57 Torsional 0.04 0 0.58


Fig. 5(b)
2 1.20 Mostly 0 0.63 0
translational in
Y Fig. 5(c)
Fig. 6. Wall mesh on axis 7 in Perform 3D. 3 1.11 Mostly 0.59 0 0.02
translational in
Table 2 X Fig. 5(d)
Element properties.
Element Definition Non-Linear Behavior

Shear 4 nodes, Bending- axial


Wall 6 degrees behavior is obtained
Element of through fibers. Each
freedom wall element has
each several fibers
containing steel and
concrete area. The
sectional behavior is
obtained by
integrating the
behavior of each
individual fiber
following Bernoulli's
hypothesis.
Frame 2 nodes, Non-Linear behavior
Element 6 degrees is obtained by locating
of plastic hinges with
freedom zero length at the Fig. 8. Stress-strain curve for confined and unconfined concrete.
each ends of each frame
element. Plastic
hinges are defined
3.3. Base and improved-design models
through a Moment-
Rotation curve. To verify the actual performance of the building through a nonlinear
analysis, a so-called “original model” seeks to reproduce the actual con-
strained, causing their capacity to be reduced in compression. The ditions of the structure before the 2010 earthquake. However, compar-
cyclic response of a bare bar with an aspect ratio L/d = 20, similar to ing this first model with alternative structural design configurations is
the observed buckling length shown in Fig. 3, of a reinforcing steel of value because varying partial elements generates results which can
specimen obtained from an actual damaged building in Chile [33] is be used to improve the design. For the “original model”, all longitudinal
shown in Fig. 9. The reinforcing bars, with similar tensile strength reinforcement in walls with buckling behavior and concrete is consid-
(~490 MPa) to the bar considered in this work, reached slightly over ered unconfined. The “confined wall axis 7” model considers a formula-
200 MPa of strength in compression, which conditions the equilibrium tion in which the damaged wall was properly confined, such that the
of a wall section subjected to flexural moment and axial load, forcing an wall web thickness is increased from 20 cm to 30 cm and the boundary
increase in the compressed zone and an associated increase of the com- longitudinal reinforcement is restrained from buckling, together with a
pressive strains, accelerating the damage process. Thus, for the steel confined concrete associated to a special boundary detailing of the web
prone to buckle, a compressive strength of 200 MPa was considered, end. All other walls are considered not confined, but since prescriptive
which is slightly less than 50% of the yield strength of the non-buckling detailing is included in boundary elements with transversal reinforce-
steel (Fig. 10). Similar results of strength reductions in compression are ment spacing at 6db (db, longitudinal bar diameter), the reinforcement
obtained with a fiber model for bar buckling [34]. model is considered without buckling. In the “decoupled wall axis 7”

6
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

4. Analysis results

4.1. Original model

Firstly, it is important to understand the earthquake demand on


buildings in the city of Santiago. For this, the displacement spectra of
different records of the 2010 earthquake in Santiago are plotted in Fig.
11(a), considering a damping of 2.5% [24]. The spectral records are ori-
ented in the same direction as the studied building. For comparison pur-
poses, the current design code displacement spectra [25] are also in-
cluded for soil type B and C. The acceleration spectra are included in
Fig. 11(b) for the current [25] and previous design code [26] used for
the design of the studied building. It is worth noting that D.S. 61 (cur-
rent design code) soil type B [25] is identical to NCh 433 soil type 2
[26]. According to Godoy et al. [35], all stations indicated in the figure
are classified as soil type C (Vs30 ≥ 350 m/s), except for Santiago
Downtown which is classified as soil type B (Vs30 ≥ 500 m/s). It can be
seen that for the building cracked period, assuming a 30% increase
Fig. 9. Stress-strain response of a damaged bar with an aspect ratio of 20 [33]. from the uncracked period (1.2 s), giving a value of 1.6 s, for the trans-
lational mode (Y – direction with observed damage), the displacement
demand is between 10 and 22 cm. Considering a system of 23 stories,
rather than a single degree-of-freedom system, the expected lateral roof
displacement is about 30% larger than the spectral demand [36]; the
range translates into lateral roof displacements of 13 and 29 cm, or
drifts between 0.002 and 0.005. In the X-direction, the expected de-
mands for a cracked period of 1.4 s are smaller (~20 cm for the upper

Fig. 10. Stress-strain curve for buckling and non-buckling steel.

model, a section of the wall web is reduced to decrease the coupling be-
tween all wall segments. The removed section corresponds to a 3.3 m
long section that connects the web and flange of the wall, as shown in
Fig. 1 (in blue) and Fig. 6. The length of section (3.3 m) was selected to
effectively reduce the coupling of the wall cross-section, decreasing the
wall web by about 40%. In the model, it is assumed that special bound-
ary elements in walls are not required, such that all concrete is consid-
ered unconfined; similar to the “confined wall axis 7” case, however,
the reinforcement is considered without buckling, based on basic pre-
scriptive detailing. Finally, the “confined walls” model considers all
walls with special detailing, such that concrete is considered confined
at wall boundaries and the longitudinal reinforcement without buck-
ling. For comparison purposes, the wall thickness was not changed.
The models are all subjected to nonlinear static analysis (pushover);
the “original model” and “confined wall axis 7 model” are also sub-
jected to nonlinear dynamic analysis. All analyses consider P-Δ effects,
although the displacement demands are moderate. Despite the fact that
the principal natural vibration mode is marked torsional, the damage is
associated to flexure-axial behavior inducing large tensile-compressive
normal strains in wall boundaries, which are associated with transla-
tional modes. Thus, pushover analysis describing the translational
modes is considered to be adequate.

Fig. 11. Displacement and acceleration spectra – (a) displacement spectra, in-
cluding measured records in Santiago for the 2010 earthquake, (b) acceleration
spectra.

7
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

bound in Fig. 11(a)) and damage is less likely, which is consistent with building system, the deformation profile is plotted for the entire sec-
the building behavior. tion. It can be seen in Fig. 13 that the entire wall of axis 7 (the last 8 m)
Lagos et al. [37] showed that the vast majority of buildings that pre- is in compression. Moreover, the complete section works almost as a
sented damage in the 2010 earthquake were subjected to roof drifts single unit (following the Bernoulli hypothesis) for small lateral drift
close to 0.01, and buildings subjected to roof drifts lower than 0.005 values, with a neutral axis located in the flange of the wall of axis 7, and
generally maintained operational performance. Therefore, this case one end of the combined wall cross-section in tension (Fig. 1), while the
study would be outside the range for buildings with potential damage. tip of the wall of axis 7 withstands the largest compressive strains. For a
This leads to the question: why was the building damaged prematurely? roof drift of 0.003, the point of the wall of axis 7, on the first floor,
To attempt to answer this question, a nonlinear static analysis is car- reaches a concrete compressive strain of about -0.006 (Fig. 13). The
ried out for the “original model” case to evaluate the behavior of the compressive concrete strain exceeds the = −0.003 limit at over 3 m
structure for the expected lateral displacement demands. The nonlinear measured from the top of the wall. Thus, based on the model results, for
model, when subjected to a pushover in the +Y direction of the build- a roof displacement of about 18 cm (within the expected values in San-
ing, using its second modal form (translation in the Y direction) which tiago in the 2010 earthquake), compression damage along 3 m mea-
compresses the tip of the wall of axis 7 (as shown in Fig. 1), yields the sured from the top end is presented in the wall of axis 7. This result is
capacity curve of Fig. 12; the limit state of concrete compressive strain consistent with the actual damage along a length of 4 m measured from
of = -0.003 can be visualized in the figure for the point of the wall its tip that was observed in the building.
of axis 7 and other walls of the building. As shown in Fig. 12, for a lat- As shown in Fig. 14, the strain distribution along the height of the
eral roof drift of 0.0025 (equivalent to 15 cm of roof displacement), the building at the tip of the cross-section of the wall of axis 7 together with
wall of axis 7, on the first floor, is the first wall to reach the limit state of another six walls (depicted in Fig. 1) confirms that the damage due to
concrete compression strain of = −0.003. This result is consistent compressive strain is concentrated at the first floor, with maximum
with the expected roof drift demand between 0.002 and 0.005 (Fig. 11) damage observed on the wall of axis 7. The figure includes the strain
and with the actual performance of the building during the earthquake values at three drift levels, with the solid lines corresponding to the
that presented damage in the unconfined wall boundary of axis 7. strain at the +Y wall end and the dashed lines corresponding to the
To understand the cause of this premature failure, it is informative strains at the -Y wall end. In Fig. 14(a), the building is pushed in the +Y
to plot the deformation profile of the wall on the first floor with respect direction, whereas in Fig. 14(b), the building is pushed in the -Y direc-
to the Y-axis. Given that the wall of axis 7 is part of a relatively large tion. The damaged element of the wall of axis 7 has a height of 1.35 m,
structural system (see Fig. 1) that controls the global behavior of the with large compressions in the +Y end for the pushover of the building

Fig. 12. Normalized base shear versus lateral roof drift for the different models.

Fig. 13. Strain profile across the wall of axis 7 for the Original Model.

8
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 14. Strain profile along the tip of the wall of axis 7 for the Original Model – (a) push in the +Y direction, (b) push in the -Y direction.

in the same direction. In the actual building, the damage was concen- ● The “confined walls” model considers all walls with special
trated at a height close to 0.7 m. Therefore, having modeled the ele- boundary detailing (concrete is confined, and longitudinal
ments with a height of 1.35 m in the first levels, the plastic hinge would reinforcement without bucking).
distribute the damage over a larger height than was observed; thus, the
local strain values should be higher. The pushover results are shown in Fig. 12. It can be seen that, in all
alternative design cases, the change in the stiffness of the structure is in-
4.2. Improved-design models significant, except for the “decoupled wall axis 7” case; however, the
performance improves considerably. Other walls close to the wall of
As described previously, another three models are proposed to im- axis 7 (Fig. 1) reach a compressive strain of = −0.003 at a similar
prove the behavior of the building. The additional cases are: drift level as for the “original model” case, ranging from 0.0035 to
0.0045. Considering that damage in such walls was not observed in the
● The “confined wall axis 7” model increases the wall web to 30 cm actual building and the “original model” anticipates reaching that limit
and properly details the boundary (confined, without bar bucking). state at a slightly lower drift level, if the drift demand is maintained,
All other walls have concrete that is considered not confined, but relevant damage should not be observed for the alternative design solu-
longitudinal reinforcement is restrained to buckling. tions. On the other hand, the compressive strains in the wall of axis 7
● The “decoupled wall axis 7” model removes a section of the wall are drastically reduced, notably improving the performance of the
web. All walls have concrete that is considered not confined, but structure. The limit state of the compressive strain of = −0.003 is
longitudinal reinforcement is restrained to buckling. reached beyond a drift level of about 0.004 in all cases; however, given

9
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

the additional ductility in the walls, global strength degradation of the Two nonlinear dynamic analyses are carried out, using the records of
building is observed beyond that point. In the “confined wall axis 7” Maipú (representing the upper bound in terms of displacement de-
case, the degradation of the overall building occurs at about 0.004 drift mands and with a soil site condition of Vs30 ~ 450 m/s, consistent with
given the inability of other walls to sustain larger drift levels due to the expected soil type C in the building, and located at around 16 km
their limited ductility, which is larger than the value observed for the from the building site) and Santiago Centro (representing the lower
“original” model (0.0025). The “decoupled wall axis 7” case with the bound in terms of displacement demands with a soil site condition of
modification of the wall configuration produces a ductile response, Vs30 > 500 m/s, and located at around 5 km from the building site).
where the = −0.003 strain level is not observed for wall axis 7, and Each set of records were incorporated simultaneously into the model in
the overall degradation of the building is reached above a 0.008 drift two orthogonal and consistent directions with the orientation of the ac-
level. Even though the response seems better than that of the other solu- celerometers and the orientation of the building.
tions, the stiffness is largely reduced, which would induce a larger dis- In Figs. 16 and 17, the general results of the dynamic analyses are
placement demand in the building system. plotted, including the acceleration signal records, the roof drift, and the
The strain profiles associated with the study cases are plotted in Fig. variation in the concrete strain at the tip of the wall of axis 7; the com-
15. An improvement in the strain profile can be observed in terms of the pressive strain is also plotted at the tip of other walls (depicted in Fig. 1)
maximum compressive strains for the “confined wall axis 7” model (Fig. in the same direction (+Y) at the first floor, as shown in Fig. 1 in red. It
15(a)). For the “decoupled wall axis 7” model (Fig. 15(b)), a different can be seen in Fig. 16(c) that the roof drift for the Santiago Centro
and convenient behavior is observed, with largely reduced compressive record reaches a maximum value of 0.002. On the other hand, in Fig. 16
strains in wall axis 7 at consistent drift levels (drift demands might be (d), the wall of axis 7 experiences the highest levels of compressive
larger). On the other hand, by confining all the walls of the building strain; however, the difference with respect to neighboring walls is min-
(Fig. 15(c)), the overall performance of the structure also improves; the imal. Similar results for the Maipú records are presented for Fig. 17. In
improvement is less, however, than with the “confined wall axis 7” case Fig. 17(c), it is observed that the roof drift for the Maipú record reaches
in which the wall web thickness is increased, reducing the axial stress in a maximum value of 0.0035, which is larger than with the Santiago
the wall and further reducing the compressive strain in the wall bound- Centro record, but smaller than the expected upper limit defined with
ary. the spectral displacements. The differences might be associated with
the effect of nonlinear incursions of the building. On the other hand, in
4.3. Nonlinear dynamic analysis Fig. 17(d), it can be seen that the wall of axis 7 reaches the highest lev-
els of compressive strain again; for this case, contrary to the case of San-
The previous results have shown that the static nonlinear analysis tiago Centro, the difference with respect to surrounding walls is signifi-
reproduced, with high fidelity, the observed damage presented in the cant, due to the formation of the plastic hinge in the wall of axis 7. In
study building. It was also employed to evaluate the impact of design Fig. 17(e), the same results as in Fig. 17(d) can be seen, but for the “con-
changes on the building capacity curve. fined wall axis 7” model; the strain values for the wall of axis 7 are
To complement the analysis, the “original model” and the “confined smaller compared to those obtained for the “original model”, highlight-
wall axis 7” model are subjected to a time-history dynamic analysis. ing the relevance of the wall special detailing. The strain values for the

Fig. 15. Strain profile across the wall of axis 7 for - (a) “confined wall axis 7”, (b) “decoupled wall axis 7”, and (c) “confined walls”.

10
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 16. Time history analysis for Santiago Center record - (a) acceleration in the X-direction, (b) acceleration in the Y-direction, (c) lateral roof drift (Y-direction),
(d) strain at wall boundary for the original model, and (e) strain at wall boundary for the “confined wall axis 7” model.

other six walls are similar for both models. It is important to note that Another common performance acceptance criterion is the maximum
for both models the roof drift demands are similar (Fig. 17(c)). The dif- inter-story drift, which is shown in Fig. 20 for the +Y direction for both
ferences observed in Fig. 17 between both models are not apparent in records, considering the absolute inter-story drift. It can be seen that
Fig. 16 given the smaller demands. the inter-story drift generated by the Maipú record exceeds the 0.005
In Fig. 18, the strain profiles associated with the points in time that drift limit established by LATBSDC [24] for buildings with fragile non-
generate the largest compression and tension strain values in the tip of structural elements by about 10%.
the wall of axis 7 are presented. Both profiles are consistent with the ex- In summary, the results obtained through the nonlinear dynamic
pected strain magnitudes obtained from the nonlinear static analysis analysis for the records used show consistent performance between the
(pushover) presented previously. Thus, for this case, the static nonlin- nonlinear static analysis and the observed damage.
ear analysis can be used to obtain displacement demands and strain dis-
tributions as an approximation of the dynamic behavior of the struc- 5. Conclusions
ture. Furthermore, the Maipú record better presents compressive strain
levels (<-0.003) consistent with the observed damage, which is also In this work, a reinforced concrete building is studied, located in the
consistent with the expected soil type classification for the building. Ñuñoa district in Santiago, Chile, which presented damage due to flex-
The concrete compression for the instant of maximum strain at the ure and axial actions in one of its walls on the first floor (axis 7). None
tip of the wall of axis 7 and other six walls along the height of the build- of the walls of the building were designed with special boundary detail-
ing is shown in Fig. 19. The figure includes results for both records, ing.
with the solid lines corresponding to the strain at the +Y wall cross- The building is modeled based on the expected properties of the
section end and the dashed lines corresponding to the strains at the -Y building, and a series of alternative design solutions were analyzed. The
wall cross-section end. It can be seen that the level of compression at main conclusions of the study are listed below:
the tip of the wall of axis 7 varies drastically from one record to an-
other, and presents the largest value among all wall elements for the • In the city of Santiago, Chile, according to Fig. 11, spectral
Maipú record. For the Maipú record, a plastic hinge in the wall of axis 7 displacement demands of 10–22 cm could have been obtained for
is formed that reaches levels of compressive strains about 6 times the main translational period of the building, leading to
higher than the Santiago Centro record, reaching values larger than - corresponding roof drifts of 0.002–0.005 for a 23-story building.
0.007. As mentioned previously, large compressive strains in wall The records from Santiago Centro and Maipú represent the lower
boundary elements with poor detailing result in concrete degradation and upper bound, respectively, and, therefore, were selected for the
(unconfined) and bar degradation (buckling). The decrease in strength dynamic analysis.
requires an increase of the compressed area of the wall to maintain ax- • According to the result of the nonlinear static analysis of the
ial equilibrium, which in turn shifts the neutral axis away from the com- “original model” shown in Fig. 12, it can be concluded that the
pressive side, increasing the compressive strains and accelerating the model adequately predicts the damage type and location present in
damage progress. the building for the expected displacement demand. The first wall to
exceed the concrete compressive strain limit (unconfined) of

11
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 17. Time history analysis for Maipú record - (a) acceleration in the X-direction, (b) acceleration in the Y-direction, (c) lateral roof drift (Y-direction), (d)
strain at wall boundary for the original model, and (e) strain at wall boundary for the “confined wall axis 7” model.

Fig. 18. Time history strain envelope across the wall of axis 7.

= −0.003 is the wall of axis 7 on the first floor; this was the wall • Increasing the thickness of the wall of axis 7–30 cm and providing a
and the location of the observed damage. Moreover, the fibers of the special detailing at the wall web end is shown to be a viable solution
section within a length of 3 m measured from the tip of the wall of to significantly improve the performance of the wall; however, other
axis 7 exceed the compressive strain limit, which is also consistent walls require attention as they can also fail prematurely for small
with the actual progress of wall damage. Other walls reach the incremental roof drift demands.
compressive strain limit at a larger drift level. • Decoupling the long wall of axis 7 presents an improvement in the
• The results of the nonlinear dynamic analysis for the “original local and global behavior of the building; however, this causes the
model” also predict the damage observed in the building. They are building to become more flexible, which might increase the non-
also consistent in terms of the roof displacement demands and strain structural damage and the displacement demand.
levels (for all walls) compared to the nonlinear static analysis • Providing special boundary detailing to all the walls of the building
(pushover). considerably improves performance; however, the wall of axis 7

12
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

Fig. 19. Time history strain envelope along the tip of the wall of axis 7 – (a) Santiago record, (b) Maipu record.

Acknowledgments

The National Agency for Research and Development (ANID) for the
Fondecyt Regular 2020 N° 1200023 has partially funded this work.

References

[1] C. Estay, Características de muros de hormigón armado diseñados en Chile (in


Spanish) Civil engineering thesis University of Chile, 2008.
[2] L.M. Massone, P. Bonelli, R. Lagos, C. Lüders, J. Moehle, J.W. Wallace, Seismic
design and construction practices for RC structural wall buildings, Earthq. Spectra
28 (S1) (2012) S245–S256.
[3] ACI Committee 318, Building Code Requirements for Structural Concrete and
Commentary (318R-19), American Concrete Institute, Farmington Hills, MI, 2019.
[4] A. Lemnitzer, L.M. Massone, D.A. Skolnik, J.C. De la Llera, J.W. Wallace, Aftershock
response of RC buildings in Santiago, Chile, succeeding the magnitude 8.8 maule
earthquake, Eng. Struct. 76 (2014) 324–338.
[5] L.M. Massone, F.R. Rojas, M.G. Ahumada, Analytical study of the response of
reinforced concrete walls with discontinuities of flag-wall type, Struct. Concr. 18 (6)
(2017) 962–973.
[6] L.M. Massone, G. Muñoz, F. Rojas, Experimental and numerical cyclic response of
RC walls with openings, Eng. Struct. 178 (2019) 318–330.
[7] L.M. Massone, S. Díaz, I. Manríquez, F. Rojas, R. Herrera, Experimental cyclic
Fig. 20. Inter-story drift for maximum lateral displacement instant for the time response of RC walls with setback discontinuities, Eng. Struct. 178 (2019) 410–422.
[8] F.J. Jimenez, L.M. Massone, Experimental seismic shear forces in RC cantilever
history analysis in the Y direction.
walls with irregularities, Bull. Earthq. Eng. 16 (10) (2018) 4735–4760.
[9] L.M. Massone, E. Bass, Dynamic shear amplification of reinforced concrete coupled
reaches the compressive strain limit of = −0.003 nearly at the walls, Eng. Struct. 220 (2020) 110867.
same time as other walls. [10] F. Rojas, L.M. Massone, A. Hernández, Analytical study of the sectional behavior
and the effective width of T-shaped reinforced concrete walls, Eng. Struct. 237
(2021) 112110.
CRediT authorship contribution statement [11] L.M. Massone, J.I. Alfaro, Displacement and curvature estimation for the design of
reinforced concrete slender walls, Struct. Des. Tall Special Build. 25 (16) (2016)
823–841.
Leonardo M. Massone: Formal analysis, Supervision, Writing – [12] T.S. Welt, L.M. Massone, J.M. LaFave, D.E. Lehman, S.L. McCabe, P. Polanco,
original draft. Eduardo Bedecarratz: Formal analysis, Created the Confinement behavior of rectangular reinforced concrete prisms simulating wall
model that correctly captures the response. Did all the runs for the boundary elements, J. Struct. Eng. ASCE 143 (4) (2017) 04016204.
[13] F. Naeim, M. Lew, L.D. Carpenter, N.F. Youssef, F. Rojas, G.R. Saragoni, M.S.
analysis and provided the preliminary comparison of results. Fabián
Adaros, Performance of tall buildings in Santiago, Chile during the 27 February
Rojas: Provided feedback on the model formulation. Mario La- 2010 offshore Maule, Chile earthquake, Struct. Des. Tall Special Build. 20 (2011)
fontaine: Help with the construction of the nonlinear model and pro- 1–16.
[14] L.D. Carpenter, F. Naeim, M. Lew, N.F. Youssef, F. Rojas, G.R. Saragoni, M.S.
vided information on the damaged structure.
Adaros, Performance of tall buildings in Viña del Mar in the 27 February 2010
offshore Maule, Chile earthquake, Struct. Des. Tall Special Build. 20 (2011) 17–36.
Declaration of competing interest [15] F. Rojas, F. Naeim, M. Lew, L.D. Carpenter, N.F. Youssef, G.R. Saragoni, M.S.
Adaros, Performance of tall buildings in Concepción during the 27 February 2010
moment magnitude 8.8 offshore Maule, Chile earthquake, Struct. Des. Tall Special
The authors declare that they have no known competing financial Build. 20 (2011) 37–64.
interests or personal relationships that could have appeared to influ- [16] J. Moehle, J.W. Wallace, J. Maffei, C. Sempere, A. Celestino, J.J. Besa, et al.,
ence the work reported in this paper. February 27, 2010 Chile Earthquake Reconnaissance Team Investigation:
Reinforced Concrete Buildings, Reconnaissance Report; EERI, 2010.
[17] J.W. Wallace, L.M. Massone, P. Bonelli, J. Dragovich, R. Lagos, C. Lüders, J. Moehle,
Damage and implications for seismic design of RC structural wall buildings, Earthq.
Spectra 28 (1) (2012) 281–299.

13
L.M. Massone et al. Journal of Building Engineering 41 (2021) 102766

[18] Z. Tuna, J.W. Wallace, Collapse assessment of the alto rio building in the 2010 Chile [28] FEMA 356, Prestandard and Commentary for the Seismic Rehabilitation of
earthquake, Earthq. Spectra 31 (3) (2015) 1397–1425. Buildings, Federal Emergency Management Agency, Washington, D.C, 2000.
[19] C. Alarcon, M.A. Hube, R. Jünemann, et al., Characteristics and displacement [29] L.M. Massone, I. Manríquez, S. Díaz, F. Rojas, R. Herrera, Understanding the cyclic
capacity of reinforced concrete walls in damaged buildings during 2010 Chile response of RC walls with setback discontinuities through a finite element model
earthquake, Bull. Earthq. Eng. 13 (2015) 1119–1139. and a strut-and-tie model, Bull. Earthq. Eng. 17 (12) (2019) 6547–6563.
[20] R. Jünemann, J.C. de la Llera, M.A. Hube, J.A. Vásquez, M.F. Chacón, Study of the [30] N. Satake, K. Suda, T. Arakawa, A. Sasaki, Y. Tamura, Damping evaluation using
damage of reinforced concrete shear walls during the 2010 Chile earthquake, full-scale data of buildings in Japan, J. Struct. Eng. 129 (2003) 470–477.
Earthq. Eng. Struct. Dynam. 45 (2016) 1621–1641. [31] M. Saatcioglu, S. Razvi, Strength and ductility of confined concrete, J. Struct. Eng.
[21] P. Zhang, J.I. Restrepo, J.P. Conte, J. Ou, Nonlinear finite element modeling and 118 (6) (1992) 1590–1607.
response analysis of the collapsed Alto Rio building in the 2010 Chile Maule [32] M. Menegotto, E. Pinto, Method of analysis of cyclically loaded RC plane frames
earthquake, Struct. Des. Tall Special Build. 26 (16) (2017) e1364. including changes in geometry and non-elastic behavior of elements under normal
[22] L.M. Massone, I. Cáceres, Damage correlation in flexural walls with a displacement force and bending, Proceedings, IABSE Symposium, Lisbon, Portugal, 1973.
approach method for boundary detailing, J. Build. Eng. 32 (2020) 101807. [33] L.M. Massone, P.A. Herrera, Experimental study of the residual fatigue life of
[23] L.M. Massone, Fundamental principles of the reinforced concrete design code reinforcement bars damaged by an earthquake, Mater. Struct. 52 (3) (2019) 61.
changes in Chile following the Mw 8.8 earthquake in 2010, Eng. Struct. 56 (2013) [34] L.M. Massone, D. Moroder, Buckling modeling of reinforcing bars with
1335–1345. imperfections, Eng. Struct. 31 (3) (2009) 758–767.
[24] LATBSDC, An Alternative Procedure for Seismic Analysis and Design of Tall [35] C. Godoy, L. González, E. Sáez, Importancia de la velocidad de onda de corte y del
Buildings Located in the Los Angeles Region, Los Angeles Tall Buildings Structural período predominante para la evaluación de la respuesta de sitio en Santiago (in
Design Council, CA, 2020. Spanish), Obras y Proyectos 17 (2015) 61–67.
[25] MINVU, D.S. N 61, Reglamento que fija el diseño sísmico de edificios y deroga D.S N [36] ASCE/SEI 41-13, Seismic Evaluation and Retrofit of Existing Buildings, American
117, Chilean Ministry of Housing and Urbanism, Diario Official, 2010 (in Spanish) Society of Civil Engineers, 2013.
(Accessed 13 December 2011). [37] R. Lagos, M. Kupfer, J. Lindenberg, P. Bonelli, R. Saragoni, T. Guendelman, L.
[26] NCh433.Of, Diseño sísmico de edificios (in Spanish). INN, 1996, Instituto Nacional Massone, R. Boroschek, F. Yañez, Seismic performance of high-rise concrete
de Normalización, Santiago, Chile, 1996. buildings in Chile, Int. J. High-Rise Build. 1 (3) (2012) 181–194.
[27] Computers and Structures Inc, PERFORM: COMPONENTS and ELEMENTS. For
Perform-3D and Perform-Collapse. Berkeley, CA, fourth ed., 2006.

14

You might also like