Professional Documents
Culture Documents
By
CAROLINA MAGNA-VERDUGO
B.S. (Pontificia Universidad Católica de Chile) 2006
M.S. (Pontificia Universidad Católica de Chile) 2006
DISSERTATION
DOCTOR OF PHILOSOPHY
in
in the
of the
UNIVERSITY OF CALIFORNIA
DAVIS
Approved:
_____________________________________
Sashi K. Kunnath, Chair
_____________________________________
John E. Bolander
_____________________________________
Rob Y.H. Chai
Committee in Charge
2014
i
UMI Number: 3685260
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.
UMI 3685260
Published by ProQuest LLC (2015). Copyright in the Dissertation held by the Author.
Microform Edition © ProQuest LLC.
All rights reserved. This work is protected against
unauthorized copying under Title 17, United States Code
ProQuest LLC.
789 East Eisenhower Parkway
P.O. Box 1346
Ann Arbor, MI 48106 - 1346
Abstract
The seismic behavior of reinforced concrete shear walls in Chile generally demonstrated
adequate performance as seen from the response of most Chilean multistory shear-wall
reinforced concrete buildings after the 1985 and 2010 earthquakes. While shear walls are widely
used in building construction, models for simulating the nonlinear response of shear walls have
seen limited advances. The need to accurately model shear wall behavior is becoming more
important as increased confidence in the seismic behavior of RC walls has led to more relaxed
requirements for reinforcement and confinement of typical wall configurations. For example,
walls built prior to the 1985 earthquake in Chile were based more on traditional practice and
were not necessarily enforced by seismic codes. The majority of post-1985 construction was
based on ACI-318 guidelines with relaxed provisions for the confinement of the walls. The
Chile resulted in different behavior and observed damage compared to the pre-1985 walls.
In this study, typical Chilean walls are modeled and analyzed so to evaluate the
walls during the 2010 Chilean earthquake. Due to its simplicity and lower computational cost, a
multi-spring macro-model is used for the simulation. Since the modeling of an entire building
modeled using OpenSEES. The calibration of the 2D section model is performed through
ii
comparison of the dynamic properties with a 3D elastic model of the entire building developed in
behavior of selected structural walls. The research also provides guidelines for calibrating the
spring properties of the macro-element as a function of the slenderness ratio of the wall, and the
wall web reinforcement ratio. Findings from the study will be useful in advancing nonlinear
iii
Acknowledgments
I would like to express my very great appreciation and my sincere gratitude to my
academic supervisor, Professor Sashi K. Kunnath, for his constant guidance and unconditional
support during my years in Davis, and the course of this investigation. I am truly thankful for the
trust he placed in me from the beginning, for always taking the time to advice me, and for
Additionally, I would like to offer my special thanks to Professor John Bolander and
Professor Rob Chai for their comments and suggestions as a member of my dissertation
committee, as well as for their insightful observations on the writing of this dissertation.
I also would like to thank the people I worked with these years: Zhiyu Zong, Amin
Ahmadi, José Abell, and Nima Tafazzoli. Thank you for all the help, and for making my work
I wish to thank all my friends who made my life in the United States amazing pleasing
and gratifying. Especially to Nima, and to the Abell-Crempien, and Toro-Labbé families, thank
you for all the memorable evenings, lunches, and family times. I miss you a lot!!
Finally, I would like to thank my family for their support and encouragement throughout
my studies. My husband, Patricio, made these years an amazing journey, will stay with me
forever. Thank you my love for your friendship, encouragement, assistance and unconditional
iv
love. To my son, Maximiliano, for changing my life completely and for giving me the final
strength to finish this work. Thank you for existing and thank you for letting me be your mom.
Additionally I would like to send my gratitude to my mother, Elena. Thank you for always being
there when I have needed you, for all the right words at the right time, for all your
v
Table of Contents
ABSTRACT.. .................................................................................................................... II
ACKNOWLEDGMENTS .............................................................................................. IV
EARTHQUAKES................................................................................................................. 2
vi
CHAPTER 2 REINFORCED CONCRETE WALLS IN CHILE:
EARTHQUAKES ........................................................................................................... 12
CHILE............................................................................................................................. 13
2.4 DAMAGE IN WALLS OBSERVED DURING THE PAST SEVERE EARTHQUAKES ......... 22
MODELS….. ................................................................................................................... 35
vii
3.2.3 Simulation Results ......................................................................................... 65
viii
CHAPTER 6 CONCLUSIONS ................................................................................ 136
EARTHQUAKES............................................................................................................. 137
6.2.2 Evaluation of the Chilean Codes and Design Tradition on Seismic Behavior
6.4 ASSESSMENT OF BUILDINGS IN CHILE WITH DAMAGED WALLS AFTER THE 2010
ix
List of Figures
Figure 1.1: Typical reinforced concrete wall cross section............................................................................................ 3
Figure 2.1: Portion of building plans, all dimensions in centimeters, φ denotes the diameter of the rebar in
millimeters. a) Building built before 1985, b) Building built after 2010 ..................................................................... 15
Figure 2.3: Typical RC shear-wall boundary zone. a) Unconfined, b) Partially confined. Modified from (NIST,
Figure 2.4: Configurations of the use of RC walls in Chile. a) L and T shape walls, b) "Flag" walls, c) Single walls
...................................................................................................................................................................................... 26
Figure 2.5: Typical damage observed in RC walls. a), b) Damage due to in-plane behavior, c) Damage due to out-of-
Figure 2.6: Detailing of wall boundaries. a) Damage wall with inadequate transverse reinforcement, b) U and L
shape confinement, c) Plan section of typical detailing of wall boundary in damage walls........................................ 29
Figure 2.7: Damage in "Flag" walls. Photo by Patricio Bonelli .................................................................................. 30
Figure 2.8: Significant damage observed in walls after 2010 earthquake. .................................................................. 33
Figure 3.1: Macroscopic models and typical shear wall model. a) EBM, b) MVLEM (Modified from (Orakcal et al.,
2004)), c) Panel model (Modified from (Massone, et al., 2006)) ................................................................................ 37
Figure 3.5: Wall cross-section (Modified from (Orakcal et al., 2004)) ....................................................................... 47
Figure 3.6: Load-displacement response for different aspect ratios ............................................................................ 48
Figure 3.7: Cross-section of Type 1 walls: a) pre-1985, b) post-1985 (Figure not to scale) ....................................... 53
x
Figure 3.8: Cross-section of Type 2 walls: a) pre-1985, b) post-1985 (Figure not to scale) ....................................... 54
Figure 3.10: Comparison of the proposed model and the ACI equation for the estimation of the shear strength: a)
Figure 3.11: Comparison of the proposed model, ACI, and Krolocki model of shear strength estimations for pre-
Figure 3.12: Averages values for the threshold of shear strength value for pre-1985 walls: a) Intermediate walls, b)
Figure 3.13: Load-displacement response for AR walls of 2.5: a) Type 1, b) Type 2................................................. 66
Figure 3.14: Load-displacement response for AR walls of 1.75: a) Type 1, b) Type 2............................................... 67
Figure 3.15: Load-displacement response for AR walls of 1.5: a) Type 1, b) Type 2................................................. 68
Figure 3.16: Load-displacement response for AR walls of 1.2: a) Type 1, b) Type 2................................................. 69
Figure 3.17: Load-displacement response for AR walls of 2.5: a) Type 1, b) Type 2................................................. 70
Figure 3.18: Axial load effects in the load-displacement response for AR walls of 2.5: a) Type 1, b) Type 2 .......... 74
Figure 3.19: Axial load effects in the load-displacement response for AR walls of 1.75: a) Type 1, b) Type 2 ........ 75
Figure 3.20: Axial load effects in the load-displacement response for AR walls of 1.5: a) Type 1, b) Type 2 .......... 76
Figure 3.21: Axial load effects in the load-displacement response for AR walls of 1.2: a) Type 1, b) Type 2 .......... 77
Figure 3.22: Axial load effects in the load-displacement response for AR walls of 0.3: a) Type 1, b) Type 2 .......... 78
Figure 4.1: Finite element model of RC wall. a) Concrete, b) Reinforcing steel ........................................................ 82
Figure 4.3: Results of pushover simulations for three mesh sizes. .............................................................................. 85
Figure 4.4: Results showing distribution of z-direction (vertical) stresses: ................................................................. 86
Figure 4.5: Pushover curves for Type 1, Pre-1985 wall panels: .................................................................................. 92
Figure 4.6: Pushover curves for Type 2, Pre-1985 wall panels: .................................................................................. 93
Figure 4.7: Pushover curves for Type 1, Post-1985 wall panels: ................................................................................ 95
Figure 4.8: Pushover curves for Type 2, Post-1985 wall panels: ................................................................................ 96
xi
Figure 4.10: Comparison of proposed shear model versus FE simulation: pre-1985 Type 1 wall panels: a) AR=0.3,
axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3, axial load ratio = ; d)
Figure 4.11: Comparison of proposed shear model versus FE simulation: pre-1985 Type 2 wall panels: a) AR=0.3,
axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3, axial load ratio = ; d)
Figure 4.12: Comparison of proposed shear model versus FE simulation: post-1985 Type 1 wall panels: a) AR=0.3,
axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3, axial load ratio = ; d)
Figure 4.13: Comparison of proposed shear model versus FE simulation: post-1985 Type 1 wall panels: a) AR=0.3,
axial load ratio = ; b) AR=0.3, axial load ratio = ; c) AR=0.3, axial load ratio = ; d)
Figure 5.3: Convergence study for SAP2000 models: (a) Wall A; (b) Wall B ......................................................... 113
Figure 5.5: Frame selected for detailed modeling: a) Elevation view, b) SAP2000 model detail ............................. 116
Figure 5.7: Damage in RC walls for Building 2 (provided by Prof. Jorge Carvallo) ................................................ 118
Figure 5.9: Frame selected for detailed modeling: a) Elevation view, b) SAP2000 model detail ............................. 121
Figure 5.12: Bending moment variation along the height of 2nd floor wall, Building 1 ............................................ 127
Figure 5.13: Axial load distributions of the axial springs along the wall length: a) positive - direction, b) negative
xii
Figure 5.14: Comparison between the response of the 2nd floor wall and the shear capacity models for positive -
direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear model, Model B = Shear model
at flexural limit state, Model C = Shear model at critical state .................................................................................. 129
Figure 5.15: Comparison between the response of the 2nd floor wall and the shear capacity models for negative -
direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear model, Model B = Shear model
at flexural limit state, Model C = Shear model at critical state .................................................................................. 129
Figure 5.17: Pushover curve for full frame, Building 2: ............................................................................................ 132
Figure 5.18: Bending moment variation along the height of 1st floor wall, Building 2 ............................................. 133
Figure 5.19: Axial load variations in the axial springs along the wall length:........................................................... 134
Figure 5.20: Comparison between the response of the 1st floor wall and the shear capacity models for positive -
direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear model, Model B = Shear model
at flexural limit state, Model C = Shear model at critical state .................................................................................. 135
Figure 5.21: Comparison between the response of the 1st floor wall and the shear capacity models for negative -
direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear model, Model B = Shear model
at flexural limit state, Model C = Shear model at critical state .................................................................................. 135
xiii
List of Tables
Table 2.1: Evolution of Chilean seismic code, modified from (Hidalgo & Riddell, 2001) ......................................... 16
Table 2.2: Description of important damage in walls after 1985 earthquake .............................................................. 24
Table 2.3: Description of important damage in RC walls after the 2010 earthquake (B=Basement, F=Floor) .......... 31
Table 3.7: RC walls models for axial load study ......................................................................................................... 72
Table 4.1: Characteristics of typical squat and intermediate Chilean walls ................................................................ 88
Table 4.2: Statistical basis for wall configurations considered in study ...................................................................... 88
Table 4.4: Cross-sectional data for Type 1 pre-1985 walls ......................................................................................... 89
Table 4.5: Cross-sectional data for Type 2 pre-1985 walls ......................................................................................... 90
Table 4.6: Cross-sectional data for Type 1 post-1985 walls ........................................................................................ 90
Table 4.7: Cross-sectional data for Type 2 post-1985 walls ........................................................................................ 90
Table 4.9: Mean and standard deviation values for estimated variables.................................................................... 101
Table 5.1: Description of the gathered data during the field trip ............................................................................... 110
Table 5.2: Masses assigned to calibrate the frame model, Building 1 ....................................................................... 124
xiv
Table 5.3: Wall parameters, Building 1 ..................................................................................................................... 125
Table 5.4: Masses assigned to calibrate the frame model, Building 2 ....................................................................... 131
Table A.1: Database used for the estimation of the shear strength ............................................................................ 154
xv
Chapter 1
Introduction
1.1 Background
collapse under severe earthquakes, is not always achieved as is apparent from the structural
behavior observed in past severe earthquakes around the world. Therefore, it is important to
predict the inelastic seismic behavior of structures and ensure suitable performance levels before
regions since their seismic behavior has been demonstrated to be adequate during past severe
earthquakes, both from serviceability as well as a safety standpoint. Consequently, their use has
response of Chilean multistory reinforced concrete shear-wall buildings after the severe
earthquakes of March 1985 and February 2010, which has been shown to be controlled, in most
1
1.2 Characteristics of Chilean Buildings and their Behavior in Past
Severe Earthquakes
Chile is considered as one of the countries with pronounced seismic activity. This fact
has imprinted a strong seismic design tradition among Chilean engineers. Historically, the
Chilean seismic building design philosophy has been based on reinforced concrete structural
walls with high levels of redundancy to resist the seismic loads to which the buildings will be
subjected during their lifetime. These stiff and redundant structural systems have performed well
in past earthquakes (in March 1985 and February 2010) probably due to the amount of structural
walls per plan considered, which tends to be greater than seen in other seismic prone countries.
For buildings constructed before 1985, the percentage of wall area with respect to the full plan
area per floor was closer to 3% (Wood, 1991). Surprisingly, this ratio did not change much
through the years considering buildings constructed after 1985 but before 2010 (Lagos & Kupfer,
2012).
However, the layouts of the buildings did change over time. Buildings constructed after
1985 evolved, using the same wall resistant system with the same wall area per floor, to taller
buildings with higher axial loads and thinner walls (Massone et al., 2012), (Lagos & Kupfer,
2012). Currently, Chilean buildings typically consist of a larger amount of vertical and lateral
load resistant elements per floor with smaller structural elements cross-sections and lighter
reinforcement than the ones observed in pre-1985 buildings. The reduction in reinforcement with
no special ductile detailing of walls was based on past earthquake experiences where the Chilean
buildings demonstrated good performance. Characteristics of buildings built before 1985 were
used as a basis for the Chilean seismic provisions that governed the building seismic design until
2
the 2010 earthquake. Pre-1985 buildings consisted of typical reinforced concrete walls with
unconfined cross-sections, larger reinforcement ratios in the wall boundary elements, and larger
thicknesses. In the years following the 1985 event, wall configurations started to change,
evolving to thinner walls with lighter reinforcement, particularly in the boundary elements. A
comparison of the typical wall cross-sections constructed before and after 1985 can be observed
in Figure 1.1.
(a) (b)
As mentioned before, the typical construction material for mid-rise and high-rise Chilean
buildings is reinforced concrete. Concrete strength varies typically from 20 MPa to 30 MPa, and
the reinforcing steel usually has a yield strength of 420 MPa and an ultimate strength of 630
MPa. Commonly, Chilean buildings show regular structural element configurations in plan.
Previous studies (Guzmán, 1998) classified the Chilean buildings built between 1964 and
1998 into four main categories. A total of 225 buildings were inspected from which 77%
presented reinforced concrete walls as the lateral resisting system, and 18.7% reinforced concrete
moment-resistant frames with a wall core, typically used for stairs or elevator. The four
3
• Type 1: Reinforced concrete shear-wall buildings. A wall system is used as the
structural resisting system for vertical and lateral loads. Walls are continuous along
the building height, although some of them are not presented in the first floor. The
distribution of stiffnesses per plan is regular with symmetry about one or both axis.
The amount of walls per floor area usually ranges between 2%-3%. These types of
• Type 2: Reinforced concrete moment-resistant frames buildings with a wall core. The
frames are distributed regularly in the plan area from the wall core and they meet the
strong column weak beam criteria. Mass and/or stiffness irregularities along the
height, like setbacks, are common features of such buildings. According to the study,
this structural resisting system is used primarily for office buildings due to their
• Type 3: Wall buildings with moment-resistant frames in the building perimeter. This
type of building consists in a structural system based on a central wall core and a
moment-resistant frame system in the plan perimeter. The main difference with Type
2 is the connection between walls and frames, which comprises of thicker slabs
• Type 4: Buildings that cannot be classified in the previous categories. For example:
purely moment-frame buildings, steel buildings with reinforced concrete core walls,
among others.
More recent studies (Lagos, et al., 2012), (NIST, 2012) have classified the buildings in
two main categories according their use: residential and office buildings. Residential buildings
consist of reinforced concrete structural walls in the majority of the plan partitions, such as
4
corridors, separation of different apartment units, stair and elevator cores, and some partitions
within the same unit. Structural walls vary from 15 to 30 cm thick with two curtains of steel
reinforcement in both directions (vertical and horizontal). Almost no coupling beams are present
in post-1985 buildings and the floor is composed of a 13-18 cm thick flat slab with spans
between 5 and 8 m.
Office buildings present a similar layout as residential buildings, but in office buildings
structural walls exhibit reduced lengths and wider thicknesses. According to the study mentioned
above (Guzmán, 1998), their structural resistance system could be classified as Type 3,
reinforced concrete wall core with moment-resistance frames at the building perimeter. Usually
these buildings have post-tensioned flat slabs per floor with spans varying from 8 to 10 meters,
From a general perspective, reinforced concrete (RC) shear wall buildings in Chile have
behavior has been associated with the design and construction tradition rooted in Chilean
engineering practice. The type and amount of damage observed in reinforced concrete buildings
following the 2010 earthquake varies considerably when compared to observed damage after the
1985 event. This has been attributed to the spread of computational tools with which tighter
designs could be achieved and facilitated the deviation from traditional practice. Additionally,
the economic growth of the country has led to an explosive demand for residential and offices
spaces in highly urbanized areas, such as Santiago, Viña del Mar, and Concepción. This demand
may explain the increase of the height of buildings along with the number of the buildings
5
Focusing on structural walls, the general damage observed after the two earthquakes can
be classified mainly in two categories: flexural-compression damage and shear damage. In some
cases a combination between these two damage modes has been documented. In general, during
both earthquakes the damage consisted of horizontal and diagonal cracking, spalling, crushing,
and bar buckling in the reinforced concrete shear-walls. After the 1985 seismic event, more shear
related damage could be observed than after the 2010 event. In 2010 more damage in the first
story and first basement, related to high compression and flexural behavior was observed.
It is worth mentioning that the good performance of the RC buildings during the 1985
earthquake led to a relaxation of the Chilean seismic code. This change, along with tighter
designs, may be one of the reasons why damage due to flexural-compression failure increased in
RC walls during the 2010 earthquake. Despite this, the behavior of the Chilean buildings during
the 2010 event is still considered very successful. More details on the observed damage observed
While shear walls are widely used in building construction, models for simulating their
nonlinear response have seen limited advances. The need to accurately model shear wall
walls has led to more relaxed requirements for reinforcement and confinement of typical wall
configurations. As noted earlier, this was demonstrated when the type of damage to walls after
the 1985 earthquake differed greatly from the damage observed in the 2010 earthquake, having
as one of its possible causes the relaxation of the confinement required in shear wall after the
1985 earthquake (past practice was based on the Chilean design tradition until that time).
6
Shear wall modeling has evolved through three fundamental methods: approaches
derived from beam-column type models (in which flexure is the dominant mode of response),
multi-spring macro-models, and finite element models. While shear effects can be incorporated
by aggregating an inelastic shear spring in series with the flexural beam-column element, true
behavior along a finite length. Multi-spring macro-models consists of a set of springs distributed
in a configuration that allows a better representation of the strain distribution across the section
of the wall as well as the migration of the neutral axis under lateral cyclic loading. For these two
parameters is a critical step to obtain reasonable results from the simulations. Recently,
panel behavior, facilitating shear–flexure interaction, but the current model only works well for
slender and some intermediate walls with aspect ratios (ratio between height and length of the
Extensive work has been done in the development of a macroscopic model capable of
assessing with reasonable accuracy the behavior of reinforced concrete shear walls subjected to
lateral loads. The simplest form of a multi-spring macro-model was presented by Kabeyasawa
(Kabeyasawa et al., 1983), in which the fluctuation of the neutral axis of the wall cross-section
was accounted for the first time in a macroscopic model. This model idealized a wall member as
three vertical line elements with infinitely rigid beams at the top and bottom floor levels. The
two outside truss elements represented the axial stiffnesses of the boundary columns, while the
central element was a one-component model consisting of vertical, horizontal and rotational
7
springs concentrated at the base with different stiffnesses. Subsequent studies added more
features to the model; elements such a rigid vertical beam, horizontal shear spring, among others,
were included to account for the axial and flexural stiffnesses of the central panel (Vulcano,
1992), incorporation of membrane behavior to the horizontal shear spring (Colotti, 1993), efforts
in improving the material constitutive models to be used by the multi-spring model, and
calibration of model parameters for slender walls (Orakcal et al., 2004), and updating the multi-
spring macro-model to a fiber model with membrane behavior to account for the flexure-shear
interaction in squat walls. (Massone et al. (2006, 2009)). More details about the multi-spring
Notwithstanding the advances noted above, the multi-spring macro model currently does
not work suitably for every type of wall; only slender walls can be modeled accurately using this
approach, and the improvements made for walls with smaller shear span ratios are not
The need for precise and simple techniques to model reinforced concrete walls has made
simple approach to obtain accurate and reliable results when reinforced concrete walls are
modeled as part of a building could be of greater help in seismic engineering practice. Micro-
models, such as the finite element approach, for reinforced concrete buildings requires
significant computational effort when large-scale structures are simulated, as well as in-depth
knowledge of 3D material and computational modeling to achieve accurate results. These two
8
issues make finite element modeling a prohibitive approach for most of the common engineering
projects. Considering this, the focus of this dissertation is the formulation of recommendations
for simple and efficient use of the multi-spring macro-model for reinforced concrete walls to be
applied in building modeling. The main objectives of this study are the following:
understanding of the different observed failure modes following the 1985 and
2010 earthquakes.
Multiple Vertical Line Element model (MVLEM) and the Panel model in order to
accomplish the broader objective listed above, the models will be evaluated by
loading.
representing the wall webs of different types of walls are modeled using the
components and systems. From the response of these panels, models for
representing the shear behavior of walls will be developed. This shear model is
will serve as the input for wall macromodels that can be simulated in programs
and squat walls, based on their geometric properties within a 2D framework. The
9
behavior of reinforced concrete walls buildings will be evaluated through
computer models using structural plans and damage reports obtained during a
survey conducted by the author in 2010 after the February 27 earthquake in Chile.
This dissertation is organized into 5 chapters. Chapter 2 presents a review of past and
current practice of reinforced concrete wall design and construction in Chile. The review is based
on aspects of overall design philosophy, Chilean engineering practice, and the damage observed
after the past two major Chilean earthquakes. Important features of the seismic codes in use
specifically the equivalent beam model, the MVLEM, and the panel model (Massone et al.,
2006), and the ability these models to accurately predict the response of single wall tests.
Additionally, the MVLEM is used to model a set of hypothetical cantilever walls subjected to
monotonic loadings. The selected wall configurations were based on a statistical analysis of
existing walls found in Chile before and after the 1985 seismic event. Since the new shear model
will be developed in the following chapter, in this part of the study, a simple shear model is used
Chapter 4 outlines the finite element modeling performed in LS-DYNA on a large set of
wall panels. The resulting parametric study (considering a range of geometric and section
properties) provides the basis for the development of the shear model to be used as part of the
MVLEM for simulating the response of typical walls damaged during the 2010 earthquake.
10
Chapter 5 presents the nonlinear simulation of damaged walls in existing buildings. Two
post-1985 reinforced concrete buildings in Chile are first modeled in SAP2000 (Wilson &
Habibullah, 1995) to obtain their elastic dynamic properties. Selected two-dimensional wall
segments are extracted from the building and mass and stiffness properties of the walls are
calibrated to match the 3D model. Nonlinear simulation models of the walls are developed in
OpenSEES using the macro-model recommendations developed in the previous chapter. The
response and failure mode obtained from the OpenSEES nonlinear macro-model is compared
with the damage reported after the 2010 earthquake in order to assess the effectiveness of the
Finally, in Chapter 6 the main conclusions from this research study are summarized along
11
Chapter 2
design practice. Some of these changes have emerged from updates to the code provisions and
others due to general adaptation of new tradition among the practice. As in other earthquake-
prone regions, the use of reinforced concrete shear-wall buildings has been one of the most
important seismic resistant systems used in Chile, due to their seismic behavior has been
demonstrated to be adequate during past severe earthquakes. This type of system is still used
today, but it has experienced important changes in its application (Magna-Verdugo & Hidalgo,
2010).
Among the structural elements in a building, the one that has seen the most changes in
their design criteria are reinforced concrete (RC) shear walls. Between the last two severe
earthquakes experienced in Chile, in March 3th, 1985 and in February 27th, 2010, several
differences in the layout and behavior of these elements can be observed (Massone et al., 2012).
Examples of this are the decrease in wall thickness, the increase of axial load applied to these
members due to an increment in the total height of the building, among others. These
12
modifications may have affected the general type of damage observed in RC walls after the 2010
earthquake, which was not typical of the damage detected after the 1985 earthquake.
This chapter presents a summary of the types of damage observed in reinforced concrete
shear walls after the 2010 and 1985 earthquakes in Chile, and a description of the differences in
design criteria between the walls designed before and after 1985.
Tradition in Chile
The seismic building codes in Chile were designed to provide buildings with high
stiffness and strength, and with low or moderate ductility. This approach has been very
successful during the past severe earthquakes, such as the ones experienced in Chile in 1985 and
2010.
The seismic provisions that can be found in the seismic building codes guide the
structural design towards having a high lateral resistant system with the purpose of avoiding the
collapse of the structures during severe earthquakes. This requirement establishes the survival
(collapse prevention) limit inside the code. According to what can be seen in the Chilean seismic
codes of 1972 and 1996 (INN, 1972), (INN, 1996), the seismic design provisions included in the
1996 revision maintained the same the level of design forces that were prescribed in the 1972
version of the code for reinforced concrete shear wall buildings. On the other hand, according to
what was observed during the past earthquakes, especially until 2003, the buildings in Chile are
13
capable of being subjected to moderate earthquakes without having any damage, which suggests
Summarizing this design philosophy, the Chilean seismic code has been developed to
• Frequent seismic events are defined as earthquakes that occur several times during the
building lifetime (50 to 70 years). For this type of events, the buildings must be designed
• During severe seismic events, the buildings may suffer damage or may even need to be
demolished after the quake, but the collapse should be avoided at all events.
Even though the code is oriented to provide high lateral resistance to the structural
system, the Chilean design tradition of residential buildings has been essential to the structural
characteristics present in Chilean buildings, especially for buildings built before 1985. Until that
year, the structural layout with structural walls was almost exclusively used as the lateral load
resisting system. Additionally the design provided larger thickness of the walls to make sure that
they will not need to be reinforced for shear behavior. Since 1985, this practice has evolved to
provide the building with a mixed system composed of a walled core plus a frame system
surrounding it. Moreover, the thickness of the walls has experienced a reduction during the
years. Therefore, it is necessary to clarify that the structural characteristics present in Chilean
buildings until 1985 were not forced by the provisions of the code, but were mainly used because
of the existing tradition among architects and engineers. Figure 2.1 shows a comparison between
two portions of plans of different buildings, one constructed before 1985, and the other before
14
2010. This figure is an example of the differences that can be observed on wall thickness and
(a)
(b)
Figure 2.1: Portion of building plans, all dimensions in centimeters, φ denotes the diameter of the
rebar in millimeters. a) Building built before 1985, b) Building built after 2010
The Chilean seismic code has generally followed the standards used in USA, Japan and
New Zealand. Particularly in Chile, the earthquakes in May of 1960 in the south of the country,
and March of 1985 in the central part of the country have been the origin of many research
topics. These investigations have helped the Chilean practice to adapt to the seismic design
15
advances of more developed countries. Table 2.1 summarizes the historical evolution of the
Table 2.1: Evolution of Chilean seismic code, modified from (Hidalgo & Riddell, 2001)
Year Event
1928 Talca earthquake. Chilean government sent a bill to regulate the seismic design of the constructions and the approval
process of building permits.
1935 The Ordenanza General de Construcciones y Urbanización (General Ordinance of Constructions and Urbanizations)
comes into effect. In this ordinance, rules for the construction and the design of structures were established. It also
limited the use of certain materials and banned the used of some types of construction widely used until that time. This
ordinance initiated modern seismic engineering in Chile. This code was applied provisionally in 1930 and ruled
construction in Chile until 1939, when the Ordinance suffered a major revision after the 1939 Chillan earthquake (Wood
et al., 1987).
1939 The new version of Ordenanza General de Construcciones y Urbanización established building height limits, did not
allow unreinforced masonry construction, and rigorous provisions were specified for adobe construction. It included a
seismic base shear dependent on the period of the structure, which was uniformly distributed along the height of the
building.
1949 The next revision to the ordinance contained less severe seismic requirements, and it was made after the general practice
noted that the provision of 1939 led to highly costly buildings. In this code, height limitations were eliminated for
reinforced concrete and structural steel buildings (Wood et al. 1987). Additionally, some modifications to the seismic
base shear and lateral distributed forces were introduced.
1959 A committee was established in INDITECNOR (Instituto Nacional de Investigaciones Techno1ógicas y Normalización,
National Institute of Technologic Research and Normalization). The purpose of this committee was to develop a code
especially designed for seismic design, which incorporated the knowledge gathered until that time from Chile and
abroad. The project of this new code was paused after the great earthquake of 1960 in the southern part of Chile.
1960 Valdivia Earthquake. This earthquake struck the southern part of Chile. Magnitude 9.5. The most powerful earthquake
ever recorded in the history of the world.
1963 The first Chilean Conference in Earthquake Engineering was held.
1966 A provisional version of the modern seismic Chilean code, called NCh 433, was approved to replace the chapters of the
ordinance, which referred to the buildings seismic design. It included a distribution of the seismic forces along the
height that follows the first modal shape. Additionally it incorporated dynamic analysis of buildings through a design
spectrum.
1969 The 4th World Conference of Earthquake Engineering was held in Santiago, Chile.
1972 The first version of the Chilean seismic code for seismic design of buildings (NCh433.Of72) was formalized and
became a law on April 1974. In this code, the reinforced concrete design was described in the codes NCh 429.Of57
(“Reinforced Concrete: First Part”) and NCh 430.Of61 (“Reinforced Concrete: Second Part”), which are based on the
1952 and 1959 versions of the German DIN 1045 standard (Riddell et al., 1987).
1985 1985 earthquake. This earthquake struck the central part of Chile. During the 1985 earthquake, the firsts records of a
severe earthquake were obtained in the country.
1996 The second version of the Chilean seismic code for seismic design of buildings (NCh433.Of96) was formalized. This
version included the design philosophy described in the previous section, and its provisions for reinforced concrete shear
walls buildings were based on the successful behavior that this type of buildings had during the 1985 earthquake. In this
code, the reinforced concrete design was extracted from the ACI 318-95, following the Chapter 21 provisions for
seismic design, but excluded requirements included in 21.6.6.1 and 21.6.6.4 regarding the confinement of the boundary
elements in shear walls. This modification was based in the performance of the buildings in 1985.
2003 The first version of the Chilean seismic code for seismic design of industrial structures (NCh2369.Of2003) was
formalized.
2003 The first version of the Chilean seismic code for seismic design of seismic isolated buildings (NCh2745.Of2003) was
formalized.
2009 Modifications were made to the previous Chilean seismic code. Among several changes, regarding the reinforced
concrete structures, the NCh433.Of96 modified in 2009 updated the reinforced concrete provisions to the use of the
ACI318-05 code.
2010 Maule earthquake. This earthquake struck the south-central part of Chile. Magnitude 8.8.
16
It is important to highlight that the evolution of the seismic code in Chile has been mainly
related to the practice and observed behavior after severe earthquakes. The Chilean engineers
have put their efforts in trying to replicate into the code the good behavior that the Chilean
buildings have had during the past years. Nowadays, even though the 1996 code (including the
modifications of 2009) provides a similar level of lateral resistance when compared to the 1972
code, Chilean engineering practice has pushed to a tighter building design, since because of the
code and the use of computational tools have allowed certain relaxations in the design that do not
affect the overall lateral resistance of the building. For example, unlike the 1972 code, the 1996
code does not impose any lower limit to the wall thickness, allowing engineers to use reduced
thicknesses as long as the overall lateral resistance meets the requirements of the code.
Therefore, the differences in the behaviors observed after the 1985 and 2010 earthquakes might
mention that the seismic regulations until 2010 correspond to the code NCh433.Of 96 (INN,
1996) with modifications of 2009 (INN, 2009), but since this standard was in effect for a short
time before February 2010, its impact in the final behavior of the structures during the 2010
This was the first code especially developed as a guideline for the design of buildings.
Among the provisions of this code was the classification of buildings according to occupancy,
importance and structural characteristics. Also, it included static and dynamic methods for
determining lateral forces. The static analysis was allowed only for buildings up to 15 stories or
45 m in height with regular distributions of mass or stiffness along the height, and the total base
shear was not permitted to be less than 0.12 times its weight for one-story buildings (Wood et al.,
17
1987). This base shear is distributed into the various resisting elements according to their
stiffnesses. With respect to the dynamic analysis, the code allowed the use of a spectrum that was
computed as the product of the seismic coefficient, the occupancy factor of the building, and a
factor for the type of structural system. The minimum base shear was also limited, and a modal
The Chilean code of 1972 did not include provisions for the seismic design or detailing
requirements to ensure ductility of the structural system. As mentioned before, the design
guidelines were based on the German DIN 1045 standard (DIN, 1953), which did not specify any
seismic design requirements. Since the code did not contain any seismic provisions, Chilean
engineers often used the Appendix A of the ACI318-83 (ACI318-83, 1983) code to design frame
buildings. However, for the majority of the buildings, usually shear wall reinforced concrete
buildings, almost no ductility provisions were used. In general, Chilean engineers at that time did
not follow the ACI completely; they usually relied on other sources and on their own experience
Regarding to wall detailing, this code worked together with the Chilean code
NCh430.Of61 (INN, 1961), which specified that the thickness of a wall must be more than 1/25
times the distance between lateral supports measured vertically or horizontally, and should not be
less than 20 cm. The minimum thickness may be reduced to 15 cm for the top 6 meters of a
building. With respect to the shear reinforcement, double wire meshes in the plane of the wall
was required with a reinforcement ratio not less than 0.2%, and the maximum bar spacing was 30
cm. There were no limits on the amount of flexural reinforcement placed at the edge of walls,
and there were no specifications for confinement of the boundary elements. Figure 2.2 shows a
18
It is important to mention that in general practice, Chilean engineers usually avoided
reinforcing the walls for shear, because the procedure was complicated. Instead, they preferred to
provide more walls distributed in the plan area with greater thicknesses in order to limit the shear
demands over each wall to be lower than the admissible limits of the code.
The 1972 Chilean code for the seismic design of buildings was improved after the 1985
earthquake. The code tried to replicate the successful behavior of buildings during that
earthquake. The level of lateral resistance required by this code was similar to one required by
the 1972 code, but provisions for the seismic design of structural elements were included through
the implementation of the ACI 318-95 (ACI318-95, 1995) provisions for the design of reinforced
In this code it was established implicitly the design philosophy described in the previous
section: For a moderate seismic event, the behavior of the building should be kept within the
linear-elastic range and no structural damage was allowed. For a strong earthquake, a building
was allowed to enter into the inelastic range but the residual deformations must be minimal and
minor cracks were accepted. Overall the building must be reparable. When a severe earthquake
strikes the buildings, full entry into the inelastic range was allowed, the residual deformations
19
and cracks could be significant, and the building might end up being in an irreparable state, but
structural collapse was not permitted. The current code, the modified version of 2009 with the
adjustments made after the 2010 earthquake, still includes this design philosophy.
The NCh433.Of 96 comprised provisions that can be found in several modern seismic
codes around the world. The code included the occupancy factor of the building, classification of
the structural system of the building (frames, walls, mixed, etc.), type of foundation soil
demands seen in past earthquakes, minimum and maximum base shear, limitations on seismic
deformations and spacing between buildings, among others. Particularly, the structural system
factor was calibrated according to the experience after the 1985 earthquake, and it reflected the
preference of Chilean engineers to continue the use of shear walls as a primary structural system
It is important to mention that this version of the code no longer used the NCh 429.Of57
(INN, 1957) and NCh 430.Of61 (INN, 1961) codes for the design of reinforced concrete
members, instead the requirements of ACI 318-95 (ACI318-95, 1995) was selected as a model
document for the new Chilean reinforced concrete code. This new version of the Chilean
reinforced concrete code was not formalized until 2008. Therefore, until that year, Chilean
engineers followed the ACI318-95 as a guideline for the design of reinforced concrete members
with special focus on the seismic provisions included in Chapter 21. With respect to shear-wall
buildings, only one exception to the ACI code was made, the requirements for the boundary
elements detailed from 21.6.6.1 to 21.6.6.4 was not mandatory. This change in the vision of the
reinforced concrete design code may have led to a decrease of thickness of the walls and the light
20
boundary reinforcement that could be observed after the 2010 earthquake in building constructed
after 1985.
As mentioned in the previous chapter, typical pre-1985 buildings have a ratio of wall
direction (Wood et al., 1987), (Riddell et al., 1987), (Sozen, 1989), (Wallace & Moehle, 1993),
(Massone et al., 2012). The thicknesses of walls varied between 20 to 30 cm, and they contained
larger amount of reinforcing steel in the boundary elements, and double curtains of reinforcing
steel in the wall web. An important feature that should be noted about the buildings constructed
before 1985 is their shear design. Most of the buildings were designed for moderate average
shear stresses. The amount of walls provided for the lateral resisting system per floor in the two
principal directions was intended to ensure that the shear stress in each wall did not exceed 6.0
kgf/cm2. This value was known as the “admissible strength” for reinforced concrete, according to
the design codes used at that time. The design tradition (it was not a code provision) rooted
among the engineers of that time resulted in the fact that the majority of the walls designed
before 1985 did not require any special shear reinforcement, and led to building designs with a
large amount of walls with higher thicknesses per floor. Horizontal bars typically terminate with
Post-1985 buildings, especially those built closer to 2010 include structural walls with
thicknesses between 15 and 30 cm, and two curtain of reinforcing steel. The ratio of boundary
elements reinforcement decreased with respect to pre-1985 walls, and the web reinforcement
21
ratio was observed to increase. As in pre-1985 walls, 90° hooks around the boundary
reinforcement could be observed in the horizontal bars distributed along the wall height.
In some post-1985 walls partial confinement of boundary elements of the walls was also
present. This partial confinement was built using ties with alternating 90° and 135° hooks on
every other vertical bar. Buildings incorporating this type of detailing did not experience
flexural-compression damage that unconfined walls experienced during the 2010 earthquake
(NIST, 2012). Figure 2.3 shows the typical shear-wall boundary zone found after 2010
earthquake.
(a)
(b)
Figure 2.3: Typical RC shear-wall boundary zone. a) Unconfined, b) Partially confined. Modified
from (NIST, 2012)
Earthquakes
Several articles written after the 1985 earthquake highlight the outstanding performance
of reinforced concrete buildings (Cruz et al., 1988), (Wood, 1991), (Cassis & Bonelli, 1992),
22
(Wallace & Moehle, 1993). The epicenter of this earthquake was located near the city of Viña
del Mar, where a large number of mid to high-rise buildings were constructed. The damage
observed in the reinforced concrete walls varies greatly according to the structural system of the
building, year of construction, code used for design, etc. Due to all these variations, there is not a
simple way to categorize the damage observed in RC walls after the 1985 earthquake. In Table
2.2 a summary of typical damage observed in buildings during the 1985 earthquake is illustrated
according to the information found in the literature (Wood et al., 1987), (Riddell et al., 1987),
(Hidalgo et al., 2002). This table only shows a small selection of the inventories written after the
earthquake.
According to the documentation found after this seismic event, from approximately 165
buildings with available data, 5 were classified as having experienced severe damage, 8
experienced moderate damage, and 21 buildings showed only light damage. On the other hand,
no structural damage was observed in 131 buildings (Wood, 1991). When the height of the
building was analyzed, it can be found that 60% of the moderately and severely damage
buildings are concentrated in buildings with 12 to 15 stories. Additionally, when the density of
the walls is inspected (ratio of the wall area to plan area in each direction), it was observed that
buildings with a high wall density experienced less damage (Wood, 1991).
In some of the damaged buildings an unusual structural layout can be observed. This
particular configuration was used to give sea view to the residents. Floor plans of several
buildings built before 1985 can be found elsewhere (Riddell et al., 1987). It is important to note
that not all severely damaged buildings had irregular plans, but in some cases this unusual
distribution of stiffness per floor did lead to severe damage during the 1985 event.
23
Table 2.2: Description of important damage in walls after 1985 earthquake
Although it is difficult to categorize the behavior of buildings and RC walls during the
1985 earthquake, it can be generally stated that the response of Chilean multistory shear-wall,
reinforced concrete buildings was usually controlled by flexure behavior with mild development
Another interesting fact about the design provisions adopted after the 1985 earthquake
was the relaxation of the boundary confinement requirements for shear walls. Until 1985 the use
of the ACI provisions was not widespread in practice, and in some cases non-earthquake based
24
codes were used. Due to the traditional structural system used until that time (rigid, reinforced
concrete shear wall systems with large areas of walls per floor) behaved well during this
earthquake, a relaxation with respect to the ACI 318 requirement about the need for confinement
of boundary elements of the walls was included in the Chilean seismic code, but no limits were
imposed to the amount of walls per floor, or to the minimum wall thickness required, or the
amount of axial load to which the wall should be subjected. After the 1985 earthquake, some
studies highlighted the need of special confinement reinforcement when the walls are
significantly stressed (Wyllie, et al., 1986), but no requirements were made in the new Chilean
seismic code released in 1996. This relaxation might be one of the reasons for the type of failure
In general, the behavior of the buildings during the 2010 earthquake can be qualified as
successful. Of the nearly 10,000 buildings over 3 stories designed from 1985 and 2010, only 40
suffered severe damage and 4 collapsed (1 complete collapse, and 3 partial collapse). Of these
10,000 buildings, 20% have 9 or more stories, and 3% 20 or more up to 52 stories. No collapse
was reported in high-rise buildings above 20 stories (Lagos & Kupfer, 2012). These facts
demonstrate that although damage has been recorded in some buildings, in general, their
It is important to note that the primary earthquake resistant system for buildings in Chile
common to see different configurations of the RC walls. Some of these configurations are shown
in Figure 2.4.
25
(a) (b) (c)
Figure 2.4: Configurations of the use of RC walls in Chile. a) L and T shape walls, b) "Flag" walls,
c) Single walls
characteristics based on what was observed after the 1985 earthquake. This unexpected behavior
refers specifically to the type of failure that was common among RC shear walls. After what was
observed during the 1985 earthquake, some researchers expected more shear damage than
flexural damage in the RC walls. However, the most common type of failure in RC walls
observed after the 2010 earthquake was mainly dominated by a mixture between flexural and
compression behavior, as illustrated in Figure 2.5 (b) and (c). Figure 2.5 illustrates two types of
During the 2010 earthquake, most of the damage was concentrated in the types of walls
illustrated in Figure 2.4 (a) and (b). The damage observed, considering this type of behavior,
(Figure 2.5 (a) and (b)) can be characterized with: crushing and spalling of concrete at wall
boundaries, buckling of the vertical reinforcement at boundary regions with horizontal extension
into the wall web, some shear failures into the wall length (Lagos & Kupfer, 2012), overall wall
26
buckling, and damage caused by discontinuities of the walls and coupling from beams, slabs,
(a)
(b)
(c)
Figure 2.5: Typical damage observed in RC walls. a), b) Damage due to in-plane behavior, c)
Damage due to out-of-plane behavior. Photo c) by Matías Hube
27
According to some studies (Telleen et al, 2012a), the mechanism of failure of RC walls
could had been developed following two possible processes, which can lead to different
approaches to prevent future failures. The first process can be described as a failure mechanism
initiated by concrete spalling due to compression strain demands. On the other hand, the second
process explains the failure through the buckling of the longitudinal reinforcement bars causing
the spalling of the concrete – this buckling is due to flexural tension followed by compression in
the bars. Also, it is proposed that both processes assume initiation of the damage in the wall
boundaries, with propagation of the damage into the wall length as a result of subsequent cycles
after the boundary elements have lost their capacity. It is important to notice that this study also
points out that well-detailed walls would also have experienced damage if the failure was first
initiated by spalling. Nevertheless, the common flexure-compression failure during the 2010
earthquake in Chile was present mostly in poorly detailed walls (when the transverse
Following this assertion, one of the major problems that the Chilean shear walls have is
the insufficient amount of transverse reinforcement in the boundary elements. This insufficiency
was facilitated by the changes in the code after the 1985 earthquake, where the ACI 318
provisions for the confinements and boundary elements in walls were neglected. After the 2010
earthquake, the type of damage in RC shear walls showed insufficient amount of transverse
reinforcement, and confinement in the boundary elements for the axial loads to which they were
subjected. The boundary elements usually exhibited L or U shape hoops and the bars used for the
reinforcement have a small diameter. These types of walls can be observed in Figure 2.6, where
the typical confinement provided and inadequate transverse reinforcement are shown.
28
(a) (b) (c)
Figure 2.6: Detailing of wall boundaries. a) Damage wall with inadequate transverse reinforcement,
b) U and L shape confinement, c) Plan section of typical detailing of wall boundary in damage
walls.
Another problem observed in RC walls was the lack of transverse reinforcement, and this
problem was observed in “Flag” walls (Figure 2.4 (b)). The major issue with these walls has to
do with the wall length discontinuity from floor to floor. This discontinuity was provided by
design, usually to accommodate parking spaces in the first floor or first basement. The length of
these types of walls typically is reduced from the second floor to the first floor (or first floor to
the first basement), as shown in Figure 2.7, causing concrete crushing failure at the end of the
wall, at the ground level. After successive load cycles during the earthquake, the vertical load
To summarize the type of damage observed during the 2010 earthquake, Wallace and
Moehle (Wallace & Moehle, 2012) describe that the level of axial stress, which probably led to a
compression failure, influenced the brittle failure of the wall boundaries. Additionally, the larger
displacements demands than expected, the use of unsymmetrical wall cross section, and the lack
of closely spaced transverse reinforcement at wall boundaries are the other causes that can be
29
Figure 2.7: Damage in "Flag" walls. Photo by Patricio Bonelli
This type of damage may have developed as a consequence of the in-plane damage. As
illustrated in Figure 2.5 (c), it consists on buckling of a portion of the section of the wall out of
its principal plane. Typically it is concentrated on the end regions of the wall, where the
compression and tension strains are larger due to in-plane wall flexure. This type of failure
probably started with spalling of the concrete cover, which leaves a thin core with longitudinal
reinforcement that tends to buckle out of the wall plane displacing with them the rest of the wall
(Wallace & Moehle, 2012). Not only the thickness and clear height of the wall can be important
for this type of behavior, but also tension strains, which causes concrete cracking, becomes
important, since upon load reversal during an earthquake, vertical bars must carry the vertical
Table 2.3 summarizes some buildings with the type of damage listed above. The names
of the buildings were omitted as required by the companies that provided the data. Some of the
30
Table 2.3: Description of important damage in RC walls after the 2010 earthquake (B=Basement,
F=Floor)
Typical wall
Year N° of
Building Structural layout in plan thickness Damage detected in walls
Built stories
(cm)
The principal damage was located in
17-35 cm the first and second basements. The
Almost rectangular. One
from 1B to most damaged walls were oriented in
side of the building has a
3F, 17-25 the short direction and at the interior of
curved shape. One
cm from 4F the building. The major damage was
direction much larger
to 14F, and caused by high compression and
than the other one, this
17 cm upper flexural demands. Also, some of the
1 2007 20+4B causes the resistant
stories. walls presented confinement in L or U
structure to have small
Typical: 17- shape. Spalling of the concrete and the
walls in one direction,
20 cm in Bs buckling of the longitudinal bars were
and longer walls in the
and 1F, 17 also observed. Some walls experience
direction perpendicular
cm, upper out-of-plane behavior, causing an
to it.
floors. inclination of the building in 0.19
degrees.
Rectangular with walls
The major damage could be observed
distributed
in the first floor, where shear failure,
symmetrically. The walls
compression, and flexural failure were
were almost evenly
present. In the first floor some walls
arranged in the plan, 20 cm in all
2 2005 14+1B experienced total loss of stiffness due
with higher floors
to the total loss of concrete. Also
concentration at the
damage due to shear and flexural
center, where the
behavior can be observed. No out-of-
hallway or the elevators
plane behavior was reported in walls.
are usually located.
Local damage of one wall in the first
Typical: 15 basement. Initial buckling of the
Rectangular with walls
cm. In the longitudinal bars and problems with the
distributed
first confinement could be observed at the
symmetrically in the two
3 2005 11+1B basement the top of this wall. This wall can be
directions. Higher
exterior classified as a "flag" wall, but it has a
concentration of walls in
walls had 20 beam at the top, which causes
the center.
cm. separation between this wall and the
corresponding wall in the first floor.
Almost rectangular
Spread important damage in walls from
configuration in plan.
the first basement to the third floor.
The layout is symmetric 15-25 cm in
Still the damage was visible until the
in one direction, but 1B to 13F,
fifth floor. The damage included
asymmetric in the 15-20 cm in
spalling of the concrete, buckling and
perpendicular direction. the upper
4 2003 19+1B fracture to the longitudinal bars, and
The amount of walls is floors.
shear damage. Some walls in the first
similar in both Typical: 20
floor experience out-of-plane behavior.
directions. The plan is cm in all the
It could be observed flexural plus
reduced in both floors.
compression behavior, and
directions in the upper
confinements with U and L shape.
floors.
31
Table 2.3 (continued): Description of important damage in RC walls after the 2010 earthquake
(B=Basement, F=Floor)
Typical wall
Year N° of
Building Structural layout in plan thickness Damage detected in walls
Built stories
(cm)
32
(a)
(b)
(c)
33
(d)
(e)
(f)
Figure 2.8 (continued): Significant damage observed in walls after 2010 earthquake.
a) Building 1 (Photo by Matías Hube), b) Building 2, c) Building 3, d) Building 4, e) Building 5
(Photo by Jorge Carvallo), f) Building 6
34
Chapter 3
proposed in this study this chapter is divided in two main sections. In the first one a review of
three different well-known macro-models found in the literature is carried out. This part of the
study focuses on the ability of simple and advanced macro-models to simulate the nonlinear
response of RC structural walls and to assess the advantages and disadvantages of different
models. To achieve this objective, a comparative study of isolated walls utilizing three existing
modeling approaches is performed: (i) the Equivalent Beam Model (EBM), (ii) the Multiple
Vertical Line Element model (MVLEM), and (iii) a shear panel element model (Massone et al.,
2006). The comparison is based on the ability of the models to reproduce experimentally
recorded response of available wall tests in the literature. The simulation study is then extended
to walls with different aspect ratios to compare the performance of the models and to highlight
the differences in the modeling approaches. The shear walls are modeled only in the 2D plane,
In the second part of this chapter, the MVLEM is selected to perform simulations of
single isolated RC walls in order to evaluate the effects of the changes in the building codes in
Chile and in Chilean practice mentioned in the previous chapter. These simulations include
35
typical RC walls that can be found in typical buildings before and after the 1985 earthquakes.
Their behavior is assessed to obtain their capacity curves (base shear vs. lateral displacement).
The walls were classified according to their aspect ratio and axial load levels so as to additionally
observe the effects of these critical variables on the lateral force-displacement responses.
Three different macroscopic models are used to represent the nonlinear behavior of three
shear walls: the Equivalent Beam Model (EBM), the Panel model developed recently (Massone
et al., 2006), and the Multiple Vertical Line Element Model (MVLEM). Only models for isolated
The EBM represents the shear wall by a line element at the centroidal axis of the wall
(Figure 3.1 (a)). This model consists of a flexural elastic member attached to nonlinear rotational
springs at each end of the wall to account for the inelastic behavior of the critical regions.
Modifications to this model were introduced by adding plastic hinges at the ends of each beam
elements and the incorporation of inelastic shear deformations effects (Takayanagi et al., 1979).
In accordance with the modeling approach, one or more elements along the height of the wall
can be used. The major disadvantage of this model relates to the fact that all the deformations
occur at the centroidal axis. This assumption disregards the migration of the neutral axis of the
wall cross-section during loading and unloading causing effects such as rocking of the wall, and
interaction with the frame surrounding the wall to be improperly considered. The three walls
modeled with this macro-model use eight elements stacked along the wall height. The plastic
hinge length is computed using well-known recommendations (Paulay & Priestley, 1992).
36
5
Rigid Beam
6 4
Plastic Hinge
(1-c)h
kH
h
Linear Elastic
Element ch
k1 k2 kn
Plastic Hinge 2
1 Rigid Beam
3
m
m
. .
. .
. .
2
2
1 1
(a) 5 (b)
RC Wall 6 4 εy
εx
h
ch
γ xy
2
3 1
m
.
.
.
2
(c)
Figure 3.1: Macroscopic models and typical shear wall model. a) EBM, b) MVLEM (Modified from
(Orakcal et al., 2004)), c) Panel model (Modified from (Massone, et al., 2006))
The MVLEM (Figure 3.1 (b)) was originally proposed by Kabeyasawa (Kabeyasawa et
al., 1983) and improved by other researchers ( (Vulcano, 1992), (Colotti, 1993), (Orakcal et al.,
2004)). The main characteristics of this model are the ability to capture important features such
37
as shifting of the neutral axis; it also offers the flexibility to incorporate various material
hysteretic models, confinement, nonlinear shear behavior, and the effect of a fluctuating axial
force on strength and stiffness. The inelastic axial and flexural responses of the wall are
represented by a number of vertical-parallel uniaxial elements with infinitely rigid beams at the
top and bottom of the wall element. Additionally, the inelastic shear response is simulated by a
single horizontal spring. A rigid element of length ch characterizes the deformation of the wall
member under different distributions of curvatures. It is important to note that one of the
investigations conducted on the model parameters demonstrated that the number of elements
stacked on the top of each other along the height of the wall (m), and the number of vertical
elements within each wall element (n) have a small effect in the global response (Orakcal et al.,
2004). Similarly, the influence of the center of rotation parameter c can be diminished by
stacking more elements along the wall height (Fischinger et al., 1992). Currently, the MVLEM is
not capable of simulating the interaction between shear and flexural behaviors. Hence, the model
can suitably simulate compression and flexural failure only when the shear component is not
important. One of the major problems of this model is the difficulty to estimate the shear
properties when experimental results are not available. The three walls considered in the present
study are modeled with eight elements along the wall height, with a c value of 0.4 as
recommended by previous studies (Vulcano et al., 1988). The number of vertical springs along
the length of the wall varies between 8 and 10 for the different walls.
Finally, the model developed more recently (Massone et al., 2006), henceforth referred to
as the Panel Model (Figure 3.1 (c)), is an analytical model that couples wall flexural and shear
responses, retaining most of the features of the MVLEM. The model replaces the vertical
uniaxial element of the MVLEM by a combination of a shear spring with a uniaxial vertical
38
spring. In this way, a reinforced concrete panel behavior is incorporated into a two-dimensional
macro-model. Each combination of springs (fibers) act as an RC panel element subjected to in-
plane uniform normal and shear stresses, working as a membrane element. Therefore, the fiber is
no longer a uniaxial element. In this model each fiber is treated as a biaxial element,
incorporating flexure-shear interaction at the fiber level. The constitutive panel behavior can be
represented by membrane models, such as the modified compression field theory (MCFT)
(Vecchio & Collins, 1986), or the rotating-angle softened truss model (RA-STM), (Pang & Hsu,
1995). The reinforcing steel of the panel is still represented by a uniaxial constitutive stress-
strain model applied in the directions of the reinforcing bars. On the other hand, the concrete
panel is simulated by a uniaxial constitutive stress-strain model along its principal directions. It
is assumed that the principal stress and strain have the same direction. The three walls modeled
in the present study are represented by eight elements along the wall height, and the number of
vertical fibers along the length of the wall varies between 8 and 10 for the different walls.
The three shear walls selected for the comparison study in this paper are modeled using
the OpenSEES platform and following the descriptions of the macro-models described above. In
all the simulations, concrete is modeled using a uniaxial constitutive model with tensile strength,
nonlinear tension stiffening, and compressive behavior based on the Thorenfeldt curve
(Concrete06 model in OpenSEES). Reinforcing steel bars are modeled using a modified
When the EBM is used for the simulations, the wall elements are represented using the
Beam with Hinges Element with fiber sections based on patch and layer components (patch for
concrete sections, layers for reinforcing steel). The shear behavior is imposed by adding a shear
39
spring in series with the flexural behavior through the Section Aggregator command with a
uniaxial material model (PINCHING4), which incorporates degradation and pinching behavior
through a multi-linear force-deformation curve. The envelope of the force–deformation curve for
the uniaxial shear material is derived based on the envelope of the experimental data when the
wall is subjected to a cyclic load. For the MVLEM, the vertical elements are modeled using truss
elements with fiber sections based on patch and layer components similar to the case of the EBM
element. The horizontal shear springs are represented by zero-length elements. Rigid elements
are modeled through multi-point constraints (rigidLink option in OpenSEES). As in the case of
EBM, the horizontal shear spring is modeled using a uniaxial material model through a multi-
linear force-deformation curve. The envelope of the force–deformation curve for the shear spring
The panel model for all the walls is constructed by using the Flexure-Shear Interaction
element considers distributed-plasticity, and includes interaction between flexural and shear
components. As with the MVLEM, the section of the wall is defined as a fiber section, but based
on fiber components instead of patch and layer components. This change was introduced to
ensure a proper use of the panel elements, considering smeared properties of the concrete and
steel.
The analytical models described in the previous section are used to simulate the response
of three model walls tested by different researchers. Specimen details for each of these tests are
summarized in Table 3.1. All the tests described are single wall 2D tests with specimens
subjected to a cyclic load. The tests were selected from the literature to include an example of a
40
slender, an intermediate, and a squat wall. The classification of these three types of wall is made
according to their shear-span ratio (M/Vlw), as described in Table 3.1. Slender walls have a shear-
span ratio greater than 2, squat walls have a shear-span ratio lower than 1, and intermediate walls
The specimen selected to represent the slender walls is the RW2 (Thomsen & Wallace,
1995). The specimen was a rectangular cantilever wall with an axial load of approximately
0.07Agfc’. The wall dimensions were 3.66 m in height, and 0.102 m thick, with a web length of
1.22 m. The concrete compressive strength (fc’) was 27.5MPa, and the longitudinal and web
reinforcement used were Grade 60 (fy = 414 MPa) bars. Cyclic lateral displacements were
applied at the top of the wall using a hydraulic actuator. Displacements, loads, and strains at
critical locations of the wall specimen were measured during the test. The specimen had shear-
The intermediate wall selected was the specimen SW5 (Pilakoutas & Elnashai, 1995). As
shown in Table 3.1, the shear-span ratio for this specimen is 2.0. The specimen was a rectangular
cantilever wall with a height of 1.2 m, wall thickness of 0.06 m, and a web length of 0.6 m. The
41
concrete compressive strength was 36.9 MPa. The reinforcement bars of the boundary elements
had a yield stress fy = 500 MPa, and the longitudinal and web reinforcement bars used had fy =
550 MPa. Cyclic lateral displacements were applied at the top of the wall. The specimen was free
with respect to an in-plane horizontal movement, but was restrained to move vertically and out of
To investigate squat walls, the specimen N11 is used (Hidalgo et al., 2002). General
specimen information for this wall is illustrated in Table 3.1. The shear-span ratio for this wall
was 0.50. The specimen was tested by fixing its base and avoiding rotations at the top. A lateral
load was applied at specimen mid-height, generating a linear bending moment distribution with
opposite signs and equal magnitude at the wall ends. The specimen was a rectangular wall
without any axial load applied. The dimensions were 1.4 m in height, 0.1 m thick, with a web
length of 1.4 m. The concrete compressive strength was 16.3 MPa, and the longitudinal and web
The RW2 specimen is modeled using the three macro-models described. It is important to
note that based on its shear-span ratio, this wall is controlled primarily by a flexural response.
Therefore, a linear elastic force-deformation behavior is imposed on the horizontal shear spring
in the MVLEM and the EBM cases. The analysis is performed by applying an increasing
monotonic load at the top of the wall model, and recording the displacement at the top along with
the predicted base shear. Figure 3.2 compares the measured and predicted lateral load versus top
displacement responses for this specimen, according to the three macro-model representations.
The three macro-models capture reasonably well the strength of the wall and the global measured
42
crack range) can be observed in all three models. The MVLEM shows better agreement than the
Panel model or the EBM with the measured response. Nonetheless, the MVLEM response is
sensitive to the force-deformation behavior adopted for the horizontal shear spring. The sudden
drop in the residual strength of the EBM and the Panel model could be attributed to numerical
issues presented when problems to find equilibrium after a severe damage of the section is
common to observe.
160
140
120
Load (kN)
100
80
60
Test
40 MVLEM
20 Panel
EBM
0
0 20 40 60 80 100 120
Displacement (mm)
In the case of the intermediate wall, its response should comprise a combination of its
to the horizontal shear spring in the MVLEM and the EBM cases. This multi-linear curve is
derived using the data obtained from the test of the specimen, separating the shear and flexural
components using relative stiffnesses. Figure 3.3 compares the measured and predicted lateral
43
load versus top displacement responses for this specimen. As illustrated, only the MVLEM is
capable of predicting the strength of the wall, and the global measured response. However, in
this case, the MVLEM response might vary greatly if the force-deformation behavior specified
for the horizontal shear spring is not accurate. Additionally, it can be observed that the EBM is
capable of predicting the shear strength; however the maximum load value is reached at lower
displacement values when compared to the test results. On the other hand, the Panel model
underestimates the strength of the specimen and does not follow the test plateau. The sensitivity
of the MVLEM to the horizontal shear spring behavior indicates that it is possible to model this
type of wall with the MVLEM only if the shear behavior can be predicted reasonably when
100
80
Test
Load (kN)
MVLEM
60
Panel
EBM
40
20
0
0 5 10 15 20 25
Displacement (mm)
44
Since the N11 specimen is a squat wall (shear-span ratio less than 1), its response is
is imposed, following the data obtained from the wall test, to the horizontal shear spring for the
MVLEM and the EBM. The analysis is performed by applying an increasing monotonic load at
the mid-height of the wall model aided by two rigid elements, which transmit the load in the
form of a load plus a moment at the top of the wall. The displacement at the top along with the
predicted base shear is recorded. Figure 3.4 compares the measured and predicted lateral load
versus top displacement responses for this specimen. In this case, only the MVLEM is capable of
predicting reasonably well the strength of the wall and the global measured response. Although
the three models are capable of estimating the shear strength, the Panel model and the EBM
predict that the shear strength is reached at lower displacement values and the response decays
rapidly compared to the observed results. This indicates the limited capacity of these two models
when predicting the response of squat walls. Nonetheless, as it was noticed before, the MVLEM
response might vary greatly if the force-deformation behavior adopted for the horizontal shear
spring is not accurate. This indicates that great effort must be devoted to the estimation of the
shear behavior component when experimental data is not available. This is a major issue if the
45
Hidalgo et al. N11 (2002)
300
250
150
100
Test
MVLEM
50
Panel
EBM
0
0 5 10 15 20
Displacement (mm)
In order to investigate the ability of each macro-model to predict the response of walls
with different aspect ratios, the height of the specimen RW2 was varied, but its cross section and
transverse reinforcement was preserved, as shown in Figure 3.5. The heights chosen are: 4.88 m,
1.83 m, and 0.61 m, which produce aspect ratios of 4.0, 1.5, and 0.5, respectively. Since these
walls are cantilever walls, their shear-span ratios are the same as their aspect ratios. This gives
one wall for each classification, namely, one slender, one intermediate, and one squat wall.
46
19 mm 3 @ 51 mm 153 mm 3 @ 191 mm 153 mm 3 @ 51 mm 19 mm
19 mm
102 mm 64 mm
19 mm
8 - #3 bars #2 bars (db=6.35 mm) Hoops (db=4.76 mm)
(db=9.53 mm) @ 191 mm @ 76 mm
1219 mm
The walls described are modeled using OpenSEES platform, following the same
procedure described previously for the EBM, the MVLEM and the Panel model. The only
feature that differs from the previous modeling is the process for specifying the force-
deformation behavior of the horizontal shear spring, for the MVLEM and the EBM cases. In the
Panel model there is no need to input directly any special shear properties. Only the detailing of
the transverse reinforcement ratio was adjusted in order to preserve the same ratio for the three
walls.
Since no experimental data is available for the aspect ratio study, the force-deformation
behavior of the shear spring is computed using the Softened Membrane Model (SMM)
developed by Hsu and Zhu (Hsu & Zhu, 2002). Details of the calculation of the model
parameters are described in Appendix B. The SMM allows the development of a multi-linear
shear stress versus shear strain curve of a RC panel, which is extended to a force-deformation
curve for the three walls described. It is important to note that instead of the SMM, the Panel
model in OpenSEES includes the Rotating Angle Softened Truss Model (Pang & Hsu, 1995),
with some modifications in the concrete stress-strain model (Massone et al., 2006).
47
EBM responses
800
700
AR 4.0
AR 1.5
600 AR 0.5
500
Load (kN)
400
300
200
100
0
0 20 40 60 80 100 120 140 160
Displacement (mm)
(a)
MVLEM responses
500
450
400
350
300
Load (kN)
AR 4.0
250 AR 1.5
200
AR 0.5
150
100
50
0
0 20 40 60 80 100 120 140 160
Displacement (mm)
(b)
48
Panel model responses
300
AR 4.0
250
AR 1.5
AR 0.5
200
Load (kN)
150
100
50
0
0 10 20 30 40 50 60 70
Displacement (mm)
(c)
Figure 3.6 shows the response of the three walls with different aspect ratios. While the
predicted response of all models for the slender wall is similar, there is considerable variation in
the prediction of the response of the intermediate and squat walls. Figure 3.6 (a) illustrates the
response using EBM for the different aspect ratios. The estimates of strength are consistent with
expected behavior but the post-yield response of the squat wall shows considerable strain
hardening which is unlikely for a shear dominant response. Figure 3.6 (b) displays the responses
of the different walls using MVLEM. The predicted behavior of the slender and intermediate
wall is consistent with expected behavior. The ductility of the squat wall is obviously controlled
by the shear spring which is not adequately captured by the response; however this may be
attributed to the limitation of the SMM used to predict the shear response of the wall. Finally,
the responses obtained with the Panel model is shown in Figure 3.6 (c). The response of the
49
slender wall is similar to those predicted by EBM and MVLEM suggesting that all three models
are capable of predicting the response of flexural walls. The estimates of strength for both the
intermediate and squat walls are lower than those of the other models. Further, the significant
decay in the response after reaching the peak strength appears to be overly conservative based on
constructed before 1985 and before 2010 was carried out to analyze the possible difference in
their responses and failure modes. Twenty walls were modeled to assess their behavior under the
same load conditions, but considering their properties according to the Chilean design practice
before the 1985 and 2010 earthquakes. The motivation for this study lies in the fact that the
buildings, and especially RC walls, constructed after 1985 no longer present the same
characteristics and performance than those built before that year. This might have implied more
damage in RC walls during the 2010 earthquake, since the code provisions do not require the
The walls selected for this study were chosen based on typical dimensions and material
properties of buildings constructed before and after 1985. Four cross-sections, two pre-1985 and
two post-1985, were created based on RC wall characteristics of several building constructed
50
before 1985, and after 1985, but before 2010, respectively. The database used to obtain these
characteristics considered 57 walls from pre-1985 buildings and 55 walls from post-1985
buildings, and included damage and undamaged walls located between the first basement and the
second floor. Averages of the cross-sectional properties were computed with the available data,
from which two types of walls were chosen according to their lengths (lw). Based on this
selection criterion (lw), 33 walls were used for computing the properties of pre-1985 walls and 40
walls for post-1985 walls. Table 3.2 and Table 3.3 show statistical data of the properties for the
types of walls selected for representing each construction type. The tables present values of the
ratio between the length and the thickness of the walls (lw/tw), reinforcement ratios of the
boundary elements (ρs), and horizontal and vertical reinforcement ratios of the wall web (ρwh,
ρwv). The value of ρs refers to reinforcement area of the boundary elements calculated over the
ρs ρw
lw (cm) lw/tw
(%) (%)
Average 282 12 0.66 0.37
Type 1 Min 200 7 0.16 0.17
(lw=200-355cm) Max 355 18 2.29 1.83
Std 53 3 0.68 0.38
Average 667 24 0.97 0.24
Type 2 Min 580 16 0.15 0.16
(lw=580-1010cm) Max 1010 34 2.25 0.37
Std 138 6 0.71 0.06
For the purpose of this study, two types of walls, corresponding to each era (pre and post
1985), were modeled: Type 1 for a wall length equal to 300 cm, and Type 2 for lw equal to 750
cm. With the wall length established, values for thicknesses and heights were computed:
thicknesses according to the average values of lw/tw presented in Table 3.2 and Table 3.3, and
51
heights (hw) following the different aspect ratios to be investigated. The cross-sections of the two
types of walls were modeled using three different heights, which represents three different aspect
ratios (AR=hw/lw) for each era. According to their aspect ratio the walls were classified as slender
(hw/lw ≥ 2), intermediate (2 ≥ hw/lw ≥ 1), and squat (hw/lw ≤1) walls. Table 3.4 llustrates the
characteristics of these walls, Figure 3.7 and Figure 3.8 show the cross-sections modeled. In all
cases, the compressive strength of concrete is 25 MPa and the yield strength of the reinforcement
is 420 MPa.
lw ρs ρwh ρwv
lw/tw
(cm) (%) (%) (%)
Average 293 14 0.51 0.30 0.49
Type 1 Min 200 10 0.11 0.20 0.20
(lw=200-370cm) Max 370 19 2.21 2.54 2.37
Std 57 3 0.52 0.08 0.50
Average 635 32 0.58 0.39 0.37
Type 2 Min 470 22 0.13 0.25 0.20
(lw=470-1034cm) Max 1034 52 1.86 0.79 1.33
Std 154 8 0.43 0.20 0.27
52
Table 3.4: Properties of walls modeled
225 mm 225 mm
47.5 mm 187.5 mm 13 @ 160 mm 187.5 mm 47.5 mm
47.5 mm
250 mm 10mm @ 160mm 155 mm
47.5 mm
9 - 25 mm 10mm @ 160mm 9 - 25 mm
3000 mm
(a)
67 mm 67 mm
46 mm 187 mm 15 @ 160 mm 187 mm 46 mm
46 mm
200 mm 10mm @ 160mm 108 mm
46 mm
4 - 22 mm 10mm @ 250mm 4 - 22 mm
3000 mm
(b)
Figure 3.7: Cross-section of Type 1 walls: a) pre-1985, b) post-1985 (Figure not to scale)
53
54 mm 664 mm 282 mm 22 @ 250 mm 282 mm 664 mm 54 mm
54 mm
300 mm 10mm @ 250mm 192 mm
54 mm
27 - 28 mm 10mm @ 250mm 27 - 28 mm
7500 mm
(a)
134 mm 134 mm
46 mm 210 mm 42 @ 160 mm 210 mm 46 mm
46 mm
250 mm 10mm @ 160mm 158 mm
46 mm
9 - 22 mm 10mm @ 160mm 9 - 22 mm
7500 mm
(b)
Figure 3.8: Cross-section of Type 2 walls: a) pre-1985, b) post-1985 (Figure not to scale)
The models are identified with an alphanumeric notation, where the first four digits
indicate the ‘era’ that the wall represents (1985 for pre-1985 walls, 2010 for post-1985 walls),
the next digit represents the wall type (Type 1 or Type 2), and the last letter, the aspect ratio
inspected (a, b, c, d, and e for 0.3, 1.2, 1.5, 1.75, and 2.5, respectively). In all the cases, the same
steel yield strength was used for the boundary element and the wall web (fy). Table 3.4 shows the
walls reinforcement ratios for the boundary elements (ρs) and the web (ρwv and ρwh) of the walls.
Symmetric sections were considered as it can be observed in Figure 3.7 and Figure 3.8.
The simulations were performed considering single cantilever walls with axial load
assigned according to the typical axial load for the period in which the walls were built. Based on
the ratios given elsewhere (Massone et al., 2012), the assigned axial load to pre-1985 and post-
1985 walls were 0.10Agf’c, and 0.25Agf’c, respectively. Additionally, a study of the axial load
effect on the lateral strength of the walls is included. The details of these simulations are
described in the following sections. The twenty wall models were loaded laterally at the top with
a monotonic load until 5% drift was reached. It is important to mention that for these simulations
54
the aspect ratio (AR) of each wall matches its shear-span ratio (M/Vlw), since each wall was
The twenty shear walls selected for this study were modeled in OpenSEES using the
macro-model known as Multiple Vertical Line Element Model (MVLEM) with a c value (center
of rotation) of 0.4, as was chosen in the previous macro-model study. More detailed information
about this macro-model and its characteristics can be found elsewhere (Magna-Verdugo &
Kunnath, 2012). Figure 3.1 (b) illustrates the macro-model used for the simulations.
As it was described before, the main characteristic of the MVLEM is its ability to capture
important features such as shifting of the neutral axis, incorporation of various material
hysteretic models, confinement, nonlinear shear behavior, and the effect of a fluctuating axial
force on strength and stiffness. The inelastic axial and flexural responses of the wall are
represented by n vertical-parallel uniaxial elements with infinitely rigid beams at the top and
bottom. The inelastic shear response is simulated by a single horizontal spring. The RC wall is
modeled by stacking m number of elements (Figure 3.1 (b)) on the top of each other along the
In all the simulations, a uniaxial constitutive model with tensile strength, nonlinear
tension stiffening, and compressive behavior based on the Thorenfeldt curve (Concrete06 model
in OpenSEES) was used for modeling the concrete, and a modified Menegotto–Pinto model
(Steel02 model in OpenSEES) was used to model the steel. Following the MVLEM definition,
the vertical elements were modeled using truss elements with fiber sections based on patch and
layer components (patch for concrete sections, layers for reinforcing steel). The horizontal shear
55
springs were represented by zero-length elements, and rigid elements through multi-point
constraints (rigidLink option in OpenSEES). The horizontal shear spring was modeled using a
OpenSEES). Since the shear strength for each wall is difficult to estimate and the ACI 318-05
(Eq. (21-7)) expression for shear strength is very conservative, the shear properties of each wall
in Figure 3.9. This curve consists in an initial stiffness computed according to the uncracked
properties of each wall, a cracking point, yielding point, and ultimate point that considers the
56
Cracking point
The cracking point was computed from the data shown in Appendix A, selecting
intermediate and squat walls with shear-dominant responses, and following the guidelines
summarized in other studies ((Hidalgo et al., 2002), (Martinez, 1999)). For each wall presented
in Appendix A the ratio of cracking shear (Vcr) to yielding shear (Vy) was calculated and median
values were obtained as follows: 0.57 for squat walls, and 0.55 for intermediate walls. Therefore
a value of Vcr/Vy of 0.6 was taken to compute the cracking shear from the yielding shear for both
Yielding point
The yielding displacement (δy) was obtained from assuming a post-cracking stiffness of
60% of the initial stiffness, based approximately on available data and similar to the shear model
proposed by (Hidalgo et al., 2002). The yielding shear is computed by modifying the strength
estimated using ACI 318-05 provisions. The nominal shear strength of an RC wall according to
where Acv is the cross-sectional web area of the wall; αc is 3.0 for height-to-length ratio
(hw/lw) less or equal than 1.5, 2.0 for hw/lw greater or equal than 2.0, and varies linearly between
2.0 and 3.0 for hw/lw between 2.0 and 1.5; f’c is the compressive strength of the concrete; ρt (=ρh)
is the transverse reinforcement ratio; and fy is the yield strength of the transverse reinforcement
steel. Since Eq. 3.1 is known to be highly conservative, a correction to this equation must be
57
made in order to obtain a more suitable value for the shear strength of the walls, which will be
In order to modify Eq. 3.1 for use as a suitable shear strength model, the estimation
proposed in (Krolicki et al., 2011) is incorporated in the shear model proposed here. In this
estimation the contribution of the concrete, steel, and axial load to the shear strength is separated
into three different expressions. Hence, the shear strength is calculated from:
(3.2)
where Vc is the concrete contribution, Vs is the steel contribution, and Vp is the axial load
contribution to the shear strength of the wall. Krolicki et al., (2011) propose the following
where , , and γp depends on the ductility
level of the wall. In the wall models used in this study, walls with AR of 0.3 were considered to
have low ductility, and walls with AR of 2.5 or greater were considered to have high ductility;
therefore the γp factor assigned was 0.29MPa for AR = 0.3, 0.05MPa for AR = 2.5, and a linear
interpolation between these two values was made in order to obtain the γp factor for walls with
AR 1.2, 1.5, and 1.75. Additionally, hcr is the projected vertical height of the critical flexure-shear
inclined crack, which is dependent on the average diagonal crack angle; M and V are the base
58
moment and base shear of the wall, respectively; c is the depth of the compression zone of the
wall, which was taken, as recommended, as 0.2lw; and P is the axial load.
Using Eq. 3.1 and 3.2, the proposed model for estimate Vy is divided in two different
Pre-1985 walls:
(3.3a)
Post-1985 walls:
(3.3b)
where βs is a factor that was derived from a regression analysis of data from experimental
tests, the details of which are explained in subsequent paragraphs. The following expression was
obtained:
(3.4)
In equation 3.4, M/Vlw is the shear-span ratio of the wall, which in the case of cantilever
walls loaded at the top is equal to hw/lw (aspect ratio of the walls), tw/lw is the ratio between the
thickness and the length of the wall, ρv is the vertical (longitudinal) reinforcement ratio of the
wall web, ρh is the horizontal (transverse) reinforcement ratio of the wall web, and η is the axial
load ratio computed as P/(Agf’c), where Ag is the gross cross-section of the wall. The values of ρv,
ρh, and η are specified as percentage (%). The bi factors were calculated from a regression
analysis using the databases from references (Gulec & Whittaker, 2009) and (Orakcal et al.,
2009). In these databases the value of VTEST/VACI is given, as it is shown in Table A.1 of
59
Appendix A, where VTEST is the shear strength obtained from experimental tests. Table A.1 also
shows the properties of the 112 walls tests used for the regression analysis.
Equation 3.4 was chosen from 27 equations, where the 5 different wall parameters were
related in various forms. The final equation to be used was selected based on the best value of R2
for each expression. It is important to note that two different regressions were developed, one for
intermediate walls and one for squat walls. In the case of slender walls, since their overall
behavior is governed by a flexural response, the horizontal shear spring was assumed to remain
elastic; hence, the derivation of the properties for a nonlinear shear model was not needed. Table
3.5 shows the final estimations of the bi factors for each model type.
Wall type b1 b2 b3 b4 b5 b6
Intermediate 0.192 -2.572 -0.708 0.004 -0.008 0.005
Squat 0.416 -0.841 -0.210 -0.001 -0.011 -0.001
Figure 3.10 illustrates a comparison between shear strength values predicted by the ACI
code (Eq. 3.1) and the proposed equation (Eq. 3.3) for the walls described in Table A.1. The
vertical axis shows the ratio of the predicted value of the shear strength over the experimentally
observed value. As seen from the figure, the proposed model (Equation 3.3) provides closer
estimates of the experimental values than the ACI equation and will form the basis for estimating
the properties of the shear springs to be used in simulations of the Chilean walls in the next phase
of the study.
60
4
Proposed Model
tes t
3
ACI
/V
2
pred
1
V 0
0 20 40 60 80 100 120
Specimen
(a)
3
Proposed Model
tes t
2 ACI
/V
pred
1
V
0
0 20 40 60 80 100 120
Specimen
(b)
Figure 3.10: Comparison of the proposed model and the ACI equation for the estimation of the
shear strength: a) Intermediate walls, b) Squat walls.
In the case of pre-1985 walls, the model proposed in Eq. 3.3a will be used to estimate
shear spring properties since it leads to estimations of the shear strength close to Krolocki et al.
(2011) model and because the database used for the regression analysis is larger than the one
used in the Krolocki study. Figure 3.11 shows a comparison between ACI, Eq. 3.3a, and the
Krolocki model estimations of the shear strength for the pre-1985 walls studied in this section
(Table 3.4). As observed in the figure, the estimations computed through Eq. 3.3a and Krolocki
model are very similar. The ACI equation converges to estimates of the proposed model only for
61
5000
ACI
4000 Krolocki
V (kN)
Eq. 3.3a
y
3000
2000
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Aspect ratio (h /l )
w w
(a)
12000
ACI
10000
Krolocki
V (kN)
6000
4000
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Aspect ratio (h /l )
w w
(b)
Figure 3.11: Comparison of the proposed model, ACI, and Krolocki model of shear strength
estimations for pre-1985 walls: a) Type 1, b) Type 2
Additionally, Eq. 3.3a presents a threshold value for the shear strength for squat and
intermediate walls. This value was generated under the assumption that the shear capacity is
controlled primarily by the horizontal steel of the wall web. From the data presented in Table
A.1, a range between 0.2% and 0.5% was set for the horizontal web reinforcement ratio (ρh)
which is consistent with the data on the walls used in this part of the study (Table 3.4). An
average estimate for the selected range was obtained from the data and is proposed as the
minimum value for the shear strength of any particular wall. Values of 1.5VACI and 1.0VACI were
computed for squat and intermediate walls, respectively. Figure 3.12 illustrates the data used for
each wall type to obtain the threshold ratios. In the case of post-1985 walls, the Krolocki model
seems to predict the shear strength better than the model proposed in Eq. 3.3a, because the axial
62
load becomes an important component of the capacity. Therefore, Eq. 3.3b utilizes the shear
V te s t/V AC I 1.5
0.5
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
ρ (%)
h
(a)
5
4
AC I
3
/V
te s t
2
V
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
ρ (%)
h
(b)
Figure 3.12: Averages values for the threshold of shear strength value for pre-1985 walls: a)
Intermediate walls, b) Squat walls
Ultimate point
The last point in the shear curve was computed assuming that the shear resistance is
completely lost at the ultimate deformation. The decay of the wall response is calculated
following results of squat and intermediate wall tests (Hidalgo et al., 2002). From observed data,
trend lines of the post-yielding stiffness were computed based on the shear span ratios of the
walls (M/Vlw); using these trend lines, values for the post-yielding stiffnesses were calculated as
a percentage of the initial stiffnesses, and are summarized in Table 3.6 as k3. Table 3.6 shows the
characteristic parameters of each model. Values for AR, hele (height of each element used), and
63
the shear spring properties are presented. In all the walls models m=8 number of elements
stacked along the walls height were used, n=10 fibers along the walls length for pre-1985, Type
1 walls; n=12 for pre-1985, Type 2 walls; n=8 for post-1985, Type 1 walls; and n=12 for post-
1985, Type 2 walls were considered. The shear spring properties are described in the form of
stiffness values and the force-deformation pairs that describe the shear spring behavior of each
element of each wall. The walls with aspect ratio corresponding to slender walls only present
initial stiffness value, because in these cases an elastic material was used to represent the shear
Shear Behavior
AR k1 Vcr δcr k2 Vy δy k3 δult
Model hele (cm)
(hw/lw) (kN/mm) (kN) (mm) (kN/mm) (kN) (mm) (kN/mm) (mm)
1985-1a 0.3 11.25 54997 2531 0.05 32998 4219 0.10 550 7.77
1985-1b 1.2 45 13749 1303 0.09 8250 2171 0.20 275 8.10
1985-1c 1.5 56.25 10999 1303 0.12 6600 2171 0.25 220 10.12
1985-1d 1.75 65.625 9428 1209 0.13 5657 2015 0.27 189 10.96
1985-1e 2.5 93.75 6600 - - 3960 - - - -
1985-2a 0.3 28.125 65996 6548 0.10 39598 10914 0.21 660 16.75
1985-2b 1.2 112.5 16499 3423 0.21 9899 5704 0.44 330 17.72
1985-2c 1.5 140.625 13199 2869 0.22 7920 4782 0.46 264 18.57
1985-2d 1.75 164.125 11309 2588 0.23 6786 4313 0.48 226 19.55
1985-2e 2.5 234.375 7920 - - 4752 - - - -
2010-1a 0.3 11.25 43998 3815 0.09 26399 6358 0.18 440 14.63
2010-1b 1.2 45 10999 1555 0.14 6600 2592 0.30 220 12.08
2010-1c 1.5 56.25 8800 1358 0.15 5280 2263 0.33 176 13.18
2010-1d 1.75 65.625 7542 1250 0.17 4525 2083 0.35 151 14.16
2010-1e 2.5 93.75 5280 - - 3168 - - - -
2010-2a 0.3 28.125 54997 11964 0.22 32998 19941 0.46 550 36.72
2010-2b 1.2 112.5 13749 5258 0.38 8250 8763 0.81 275 32.67
2010-2c 1.5 140.625 10999 4688 0.43 6600 7813 0.90 220 36.42
2010-2d 1.75 164.125 9424 4391 0.47 5655 7318 0.98 188 39.81
2010-2e 2.5 234.375 6600 - - 3960 - - - -
64
3.2.3 Simulation Results
As described in the previous sections, twenty different RC walls (Table 3.4) were
modeled as single cantilever walls. The analyses were performed by applying an increasing
monotonic load and recording the resultant displacement at the top along with the base shear
until a 5% drift (ratio between the lateral displacement and the height of the wall) was achieved.
Figure 3.13 to Figure 3.17 show the results obtained for the three aspect ratios considered.
From Figure 3.13 to Figure 3.17 it can be observed that for slender and intermediate
cases, pre-1985 walls demonstrate a larger lateral load capacity than post-1985 walls. The lower
resistance to lateral load of the post-1985 walls can be attributed to the reduction in wall
thickness observed in the walls constructed after the 1985 earthquake. On the other hand, the
results also indicate that for the case of squat walls the deformation capacity of post-1985 walls
will be more than that of pre-1985 walls. This becomes an important issue on buildings built
after the 1985 earthquake, as is the case with the lower horizontal reinforcement web ratio in pre-
1985 walls. Additionally for intermediate and slender post-1985 walls, the response is always
governed by their flexural component, which confirms what was observed after the 2010
earthquake. In the case of intermediate pre-1985 walls, a combination between flexural and shear
behavior can be observed for Type 1 walls. The peak response is controlled by the flexural
capacity, but the decay shown in their response is part of the shear response, and the shear
component becomes more important when the aspect ratio of the wall decreases. On the other
hand, Type 2, pre-1985, squat and intermediate walls are controlled by shear, and flexural
65
1400
1200
800
pre-1985 wall
post-1985 wall
600
400
200
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(a)
5000
4500
4000
pre-1985 wall
3500
Base shear (kN)
post-1985 wall
3000
2500
δ
2000 F
1500
h
1000
500
0
0 1 2 3 4 5
Drift (δ /h) (%)
(b)
66
2500
2000
Base shear (kN)
1500
1000
500
pre-1985 wall
post-1985 wall
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
4500
4000
3500
Base shear (kN)
3000
1500
1000
500
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
67
2500
pre-1985 wall
post-1985 wall
2000
Base shear (kN)
1500
1000
500
0
0 0.5 1 1.5 2 2.5 3 3.5
Drift (δ /h) (%)
(a)
5000
4500
4000
pre-1985 wall
3500
Base shear (kN)
post-1985 wall
3000
2500
2000
1500
1000
500
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
68
2500
2000
Base shear (kN)
1500
1000
500
pre-1985 wall
post-1985 wall
0
0 0.5 1 1.5 2
Drift (δ /h) (%)
(a)
7000
6000
5000
Base shear (kN)
4000
pre-1985 wall
post-1985 wall
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
69
7000
pre-1985 wall
6000 post-1985 wall
5000
Base shear (kN)
4000
3000
2000
1000
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
20000
18000
16000
14000
Base shear (kN)
12000
pre-1985 wall
10000
post-1985 wall
8000
6000
4000
2000
0
0 1 2 3 4 5
Drift (δ /h) (%)
(b)
70
3.2.4 Axial Load Variation
In a supplementary part of the study the effects of axial load variation were evaluated. In
this case, the same pre-1985 and post-2010 walls from the previous were modeled using axial
load ratios of 0.10Agf’c and 0.25Agf’c. These ratios were chosen to match realistic axial forces in
the walls and are comparable to the values given elsewhere (Massone et al., 2012) for RC walls
constructed before and after 1985. The increase in the axial load ratios of the post-1985 walls has
been identified as one of the possible causes for the types of damage experienced by RC walls
during the 2010 earthquake in Chile, hence an evaluation of the effect of the level of axial load in
the final response of the walls is needed. In Table 3.7 the axial load levels and the shear
properties of each wall element considered in the study are illustrated. It is important to note that
the cross-sections of the walls modeled are identical to the walls described before (Table 3.4).
Additionally, since the MVLEM is used, no shear-flexure interaction is considered; only axial-
flexural interaction can be modeled using this macro-model. The influence of the axial load in
the shear capacity is included in the estimate of the yield shear as described in section 3.2.2.
For this part of the study, the shear model presented in section 3.2.2 was modified in
order to consider the effect of the axial load in the shear capacity of the wall when its
contribution becomes important. That is to say, for the yielding shear, Eq. 3.3a is applied to walls
with axial load ratio equal to 0.10Agf’c, and Eq. 3.3b to walls with axial load ratios of 0.25Agf’c,
regardless of the period that the wall represents. This modification was included because
Krolicki model (Krolicki et al., 2011) provides a more suitable estimation of the shear strength
when axial load ratios are larger. The cracking and ultimate points are not affected by this
consideration.
71
Table 3.7: RC walls models for axial load study
Shear Spring
η AR k1 Vcr δcr k2 Vy δy k3 δult
Model
(%) (hw/lw) (kN/mm) (kN) (mm) (kN/mm) (kN) (mm) (kN/mm) (mm)
1985-1a-10 0.3 54997 2531 0.046 32998 4219 0.097 550 7.77
1985-1b-10 1.2 13749 1303 0.095 8250 2171 0.200 275 8.10
10 1985-1c-10 1.5 10999 1303 0.118 6600 2171 0.250 220 10.12
1985-1d-10 1.75 9428 1209 0.128 5657 2015 0.271 189 10.96
1985-1e-10 2.5 6600 - - 3960 - - - -
1985-1a-25 0.3 54997 4786 0.087 32998 7976 0.184 550 14.69
1985-1b-25 1.2 13749 2095 0.152 8250 3492 0.322 275 13.02
25 1985-1c-25 1.5 10999 1867 0.170 6600 3111 0.358 220 14.50
1985-1d-25 1.75 9428 1747 0.185 5657 2912 0.391 189 15.84
1985-1e-25 2.5 6600 - - 3960 - - - -
1985-2a-10 0.3 65996 6548 0.099 39598 10914 0.209 660 16.75
1985-2b-10 1.2 16499 3423 0.207 9899 5704 0.438 330 17.72
10 1985-2c-10 1.5 13199 2869 0.217 7920 4782 0.459 264 18.57
1985-2d-10 1.75 11309 2588 0.229 6786 4313 0.483 226 19.55
1985-2e-10 2.5 7920 - - 4752 - - - -
1985-2a-25 0.3 65996 13899 0.211 39598 23165 0.445 660 35.54
1985-2b-25 1.2 16499 5124 0.311 9899 8540 0.656 330 26.54
25 1985-2c-25 1.5 13199 4363 0.331 7920 7272 0.698 264 28.25
1985-2d-25 1.75 11309 3930 0.348 6786 6551 0.734 226 29.70
1985-2e-25 2.5 7920 - - 4752 - - - -
2010-1a-10 0.3 43998 1876 0.043 26399 3127 0.090 440 7.20
2010-1b-10 1.2 10999 923 0.084 6600 1539 0.177 220 7.17
10 2010-1c-10 1.5 8800 923 0.105 5280 1539 0.222 176 8.97
2010-1d-10 1.75 7542 849 0.113 4525 1414 0.238 151 9.61
2010-1e-10 2.5 5280 - - 3168 - - - -
2010-1a-25 0.3 43998 3815 0.087 26399 6358 0.183 440 14.63
2010-1b-25 1.2 10999 1555 0.141 6600 2592 0.299 220 12.08
25 2010-1c-25 1.5 8800 1358 0.154 5280 2263 0.326 176 13.19
2010-1d-25 1.75 7542 1250 0.166 4525 2083 0.350 151 14.16
2010-1e-25 2.5 5280 - - 3168 - - - -
2010-2a-10 0.3 54997 7671 0.139 32998 12785 0.294 550 23.54
2010-2b-10 1.2 13749 4408 0.321 8250 7347 0.677 275 27.39
10 2010-2c-10 1.5 10999 3257 0.296 6600 5428 0.625 220 25.30
2010-2d-10 1.75 9424 3023 0.321 5655 5038 0.677 188 27.40
2010-2e-10 2.5 6600 - - 3960 - - - -
2010-2a-25 0.3 54997 11964 0.218 32998 19941 0.459 550 36.72
2010-2b-25 1.2 13749 5258 0.382 8250 8764 0.807 275 32.68
25 2010-2c-25 1.5 10999 4688 0.426 6600 7814 0.900 220 36.42
2010-2d-25 1.75 9424 4391 0.466 5655 7319 0.984 188 39.81
2010-2e-25 2.5 6600 - - 3960 - - - -
72
As in the case of the previous models, the response of these walls was obtained from the
analysis of single cantilever walls loaded at the top. The results are presented in Figure 3.18 to
Figure 3.22, where it can be observed that walls with greater axial load ratios have higher
capacity, and in the majority of the cases pre-1985 walls show a higher strength than post-1985
walls, with exception of Type 2, squat walls, and axial load ratio of 0.10Agf’c (Figure 3.22(b)).
Once again, the reduction in capacity observed for post-1985 walls can be attributed to the
reduction of the thickness of the walls that occurred after the 1985 earthquake.
When the behavior of pre-1985 walls is inspected, as expected, Type 1 wall responses are
more influenced by their flexural component and Type 2 by the shear component. Additionally
for Type 1 walls, the shear behavior becomes more important when the aspect ratio decreases,
and when the axial load increases their response shifts from primarily shear controlled to flexure
controlled. The reason for this shift can be attributed to the larger axial load contribution to the
shear capacity, which makes the flexural capacity govern the final behavior. For Type 2 walls,
the final response for squat and intermediate walls is still governed by the shear component
In the case of post-1985 intermediate walls, the flexural behavior governs in most of the
cases, but when the aspect ratio of the wall is lower, a small influence of the shear component
can be observed for Type 1 walls with lower axial load ratio. The flexural response of post-1985
walls is consistent with the observed behavior of RC wall during the 2010 earthquake.
73
1800
1600
1400
Base shear (kN)
1200
1000
800
600
η = 0.1 (pre-1985)
400 η = 0.25 (pre-1985)
200 η = 0.1 (post-1985)
η = 0.25 (post-1985)
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
6000
5000
Base shear (kN)
4000
3000
2000
η = 0.1 (pre-1985)
η = 0.25 (pre-1985)
1000
η = 0.1 (post-1985)
η = 0.25 (post-1985)
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
Figure 3.18: Axial load effects in the load-displacement response for AR walls of 2.5: a) Type 1, b)
Type 2
74
2500
2000
Base shear (kN)
1500
1000
η = 0.1 (pre-1985)
500 η = 0.25 (pre-1985)
η = 0.1 (post-1985)
η = 0.25 (post-1985)
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
7000
η = 0.1 (pre-1985)
6000 η = 0.25 (pre-1985)
η = 0.1 (post-1985)
5000 η = 0.25 (post-1985)
Base shear (kN)
4000
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
Figure 3.19: Axial load effects in the load-displacement response for AR walls of 1.75: a) Type 1, b)
Type 2
75
3000
2500
1500
1000
η = 0.1 (pre-1985)
η = 0.25 (pre-1985)
500 η = 0.1 (post-1985)
η = 0.25 (post-1985)
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
8000
η = 0.1 (pre-1985)
7000 η = 0.25 (pre-1985)
η = 0.1 (post-1985)
6000 η = 0.25 (post-1985)
Base shear (kN)
5000
4000
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
Figure 3.20: Axial load effects in the load-displacement response for AR walls of 1.5: a) Type 1, b)
Type 2
76
3500
1500
1000
500
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
9000
η = 0.1 (pre-1985)
8000 η = 0.25 (pre-1985)
7000 η = 0.1 (post-1985)
η = 0.25 (post-1985)
Base shear (kN)
6000
5000
4000
3000
2000
1000
0
0 0.5 1 1.5 2 2.5 3
Drift (δ /h) (%)
(b)
Figure 3.21: Axial load effects in the load-displacement response for AR walls of 1.2: a) Type 1, b)
Type 2
77
8000
η = 0.1 (pre-1985)
7000 η = 0.25 (pre-1985)
η = 0.1 (post-1985)
Base shear (kN) 6000 η = 0.25 (post-1985)
5000
4000
3000
2000
1000
0
0 1 2 3 4 5
Drift (δ /h) (%)
(a)
25000
20000
Base shear (kN)
15000
10000
η = 0.1 (pre-1985)
5000
η = 0.25 (pre-1985)
η = 0.1 (post-1985)
η = 0.25 (post-1985)
0
0 1 2 3 4 5
Drift (δ /h) (%)
(b)
Figure 3.22: Axial load effects in the load-displacement response for AR walls of 0.3: a) Type 1, b)
Type 2
78
The results summarized in Figure 3.18 to Figure 3.22 demonstrate the important of the
axial load effect in the final response of RC walls. Larger axial load ratios lead to greater
capacities, but this it is not sufficient to ensure that post-1985 walls will have similar capacity as
pre-1985 walls. The changes in the characteristics of the cross-sections (reinforcement ratios and
thicknesses) through the years appear to have led to a decrease in the overall capacity of the wall
79
Chapter 4
the MVLE Model (MVLEM) for the simulations of RC walls. It is desirable that the constitutive
shear behavior be based on geometrical and material properties of the wall to be modeled, as
well as on the applied axial load. The MVLEM was chosen from the other macro-models
reviewed in Chapter 3 due to the simplicity in its implementation and its ability to provide an
understanding of the overall shear behavior. Although this macro-model does not consider
interaction of the shear and flexure behavior or the interaction between shear and the axial
forces, considering the effects of axial load in the shear modeling could help to overcome part of
this problem. The final response of the walls will depend greatly on its geometry. Slender walls
will respond in a flexural manner; hence the shear behavior imposed is not important to the final
performance (using an elastic material for the horizontal spring is adequate in the MVLEM). On
the other hand, squat walls response will be dominated by the shear behavior; therefore axial-
shear interaction should be included in the model. Finally, the response of intermediate walls will
be a combination of their flexure and shear behavior, which the MVLEM is capable of predicting
if the correct shear behavior is specified for the lateral shear spring, even though no flexure-shear
80
interaction is included in the model (Chapter 3). Another important feature of this model is its
ability to simulate cyclic loading without implementing many changes to the modeling approach
Having access to a large database of experimental results on single wall tests with
different axial load ratios and geometrical configurations would be extremely helpful to develop
an adequate shear model. However, large scale-testing programs are difficult to carry out due to
many factors including cost, available test facilities and time. Additionally, it is difficult to find
reliable test programs in the literature with adequate variation in model parameters needed to
suitable develop a shear model. Given these facts, a finite element simulation program is a very
good alternative to generate a database of results. Finite element models need to first be
large-scale simulations.
In this study finite element models of RC walls are used in an extensive parametric study
to develop phenomenological models for characterizing shear behavior of squat and intermediate
walls. The finite element platform used in this research work is the commercial software LS-
DYNA (Hallquist, 2005), which is a nonlinear, explicit finite element program with the ability of
microscopic and macroscopic representations, namely finite elements and structural based
elements, respectively. Finite element representation helps in the modeling of the concrete,
where confined and unconfined concrete can be considered. Structural based elements refer to
beam, shells, among others, which facilitates the modeling of the reinforcement distributed along
the wall cross-section and height. Currently, LS-DYNA incorporates adequate material models
81
for steel and concrete that helps to predict reasonably well the behavior of the reinforced
Full RC walls and web sections of RC walls were modeled in LS-DYNA. The geometry
of the cross-sections was modeled using solids elements for the concrete part, and beam elements
for the wall reinforcement along the reinforcements lengths connected to nodes of the solid
elements. Figure 4.1 shows a representation of a full RC wall. At the top of the wall rigid solids
elements were included to ensure the correct transfer of the lateral displacement imposed at the
(a) (b)
82
The elements used to model concrete are 8-node solid elements with a constant stress
solid element formulation, and the reinforcement steel is modeled by means of Hughes-Liu beam
elements with cross section integration formulation. More information on these formulations can
With respect to the material models, unconfined concrete was modeled using the material
“MAT_CSCM” (Continuous Surface Cap Model) included in the LS-DYNA framework. With
this material, the program is capable of considering failure of solid elements and removal of the
elements when the specified failure criterion is reached. The failure threshold is defined by the
user and represents the maximum attainable strain prior to failure of the material. More details
about this material can be found elsewhere (Murray, 2007). The material model used to represent
elastic-plastic material with failure based on a plastic strain and a stress-strain curve that can be
The visualization of different stress and strain states is possible through the post-
processor LSPrePost (LSTC, 2012). This is an interactive program where the user can prepare
the input data for LS-DYNA and process the results from LS-DYNA analyses. The post-
processor can display stresses and strains for each material (steel and concrete) separately for all
elements in the model at any time step. Animation of results is also feasible.
For all the simulations presented here, the walls were modeled as isolated walls fixed
only at the base. A lateral displacement at the top of the wall in the direction of its length was
imposed until 2.5% drift was reached. For models of squat walls the maximum lateral
displacement applied was 5% drift to ensure that the results contain sufficient details about the
83
shear behavior near failure. The lateral displacement was applied monotonically, and the axial
load was applied at the beginning of the simulations and remained constant during application of
the monotonic lateral displacement. Two levels of axial load were investigated in order to
evaluate its effect on the shear response of RC walls: 0.10Agf’c, and 0.25Agf’c.
Before running the simulations a mesh size study was carried out to attain a reasonable
balance between computational cost and model accuracy. For this, a post-1985, AR 0.3, Type 1
wall was selected and the force-displacement capacity curves were compared with three different
mesh sizes. The meshes were built considering the spacing between the horizontal reinforcement
position along the height (z-direction), the longitudinal reinforcement position across the length
(x-direction), and 5 subdivisions along the thickness of the wall (y-direction). Along the x-
direction one element is considered between each longitudinal rebar of the boundary elements,
and two elements between each longitudinal rebar along the wall web.
The meshing study was performed by varying the mesh size only in the z-direction and
three meshes were modeled. Mesh 1 had the largest elements and considered only one element
between two horizontal reinforcing bars (1 element = 1 horizontal reinforcement spacing). Mesh
2 consists of two elements between two lines of horizontal reinforcement up to half the wall
height, and one element from mid-height until the top of the wall. Mesh 3 considers three
elements between two horizontal reinforcement lines from the bottom up to half height of the
wall, and then two elements until the top of the wall height. Figure 4.2 shows the three different
84
(a) (b) (c)
The results of this study are shown in Figure 4.3 and Figure 4.4, where the difference in
the pushover responses for the three mesh sizes can be observed. The computational time for
each simulation (using a Xeon 2.40 GHz computer with16 CPU cores) was as follows: 30
minutes for Mesh 1, 43 minutes for Mesh 2, and 48 minutes for Mesh 3. It is important to
mention that this convergence study was performed using the smallest wall configuration. For
taller and/or longer walls the simulations took several hours, therefore the amount of time of
6
x 10
3
2.5
Base Shear (N)
1.5
Mesh 1
0.5 Mesh 2
Mesh 3
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Displacement (m)
85
(a)
(b)
(c)
86
As it can be seen form Figure 4.3 and Figure 4.4, the stress distribution using Mesh 1 is
significantly different compared to Mesh 2 and 3. On the other hand, Meshes 2 and 3 show some
variations in the distribution of stresses – these variations are minimal implying that further
reduction in the mesh size is unlikely to result in a change in the response but only lead to an
increase in the computational time of the simulation. Therefore, based on this study, the mesh
size chosen for all the simulations is Mesh 3, which gives accurate results at an appropriate
From the analyses performed in LS-DYNA for the current study two different files were
generated: nodal displacements along the mid-section of the thickness at the top of each wall,
and the reaction forces at the base nodes. From this data the pushover curve was generated for
each wall. The ensuing parametric study of different wall configurations will form the basis of
the development of the empirical shear model representing the shear component in the MVLE
In order to develop the shear model based on the geometric and reinforcement
characteristics of typical walls, a parametric study was carried out. This parametric study is
based on the statistical information presented in Table 3.2 and Table 3.3, where values of
and are taken as the main wall web features. Models of several wall panels were built in LS-
DYNA (in accordance with the statistical information of the parameters mentioned above) so as
to have an adequate number of data points to develop the shear model. These panels only
87
consider the wall web, without the boundary elements, since the shear response is controlled by
the properties of the web section. Table 4.1 shows a summary of the values extracted from Table
3.2 and Table 3.3, and Table 4.2 illustrates the statistical information used as a basis to generate
different models – in the present study, the configurations were generated based on the average
() and standard deviation () values for each of the parameters.
From Table 4.2 twelve sections for each type and year of construction were created and
they are listed in Table 4.3. Since the cross-sections must represent realistic values that can be
found in the field, such as the reinforcement ratio, wall thicknesses, diameter of reinforcing bars,
the generated cross-section information was suitably modified. The actual cross-section
88
information is summarized in Table 4.4 to Table 4.7, where the proposed and the actual web
Pre-1985 T1
lw = 300 cm
lw/tw tw (cm) ρw (%) proposed spacing (cm) db (mm) ρw (%) actual
S11 15 20 0.75 15 12 0.75
S12 15 20 0.60 18 12 0.63
S13 15 20 0.40 20 10 0.39
S14 15 20 0.20 20 8 0.25
S21 12 25 0.75 12 12 0.75
S22 12 25 0.60 10 10 0.63
S23 12 25 0.40 15 10 0.42
S24 12 25 0.20 20 8 0.20
S31 10 30 0.75 10 12 0.75
S32 10 30 0.60 12 12 0.63
S33 10 30 0.40 20 12 0.38
S34 10 30 0.20 25 10 0.21
89
Table 4.5: Cross-sectional data for Type 2 pre-1985 walls
Pre-1985 T2
lw = 750 cm
lw/tw tw (cm) ρw (%) proposed spacing (cm) db (mm) ρw (%) actual
S11 30 25 0.30 20 10 0.31
S12 30 25 0.25 25 10 0.25
S13 30 25 0.20 20 8 0.20
S21 25 30 0.30 18 10 0.29
S22 25 30 0.25 20 10 0.26
S23 25 30 0.20 25 10 0.21
S31 20 38 0.30 15 10 0.28
S32 20 38 0.25 25 12 0.24
S33 20 38 0.20 20 10 0.21
Post-1985 T1
lw = 300 cm
lw/tw tw (cm) ρw (%) proposed spacing (cm) db (mm) ρw (%) actual
S11 18 17 0.70 20 12 0.68
S12 18 17 0.50 12 8 0.50
S13 18 17 0.40 15 8 0.40
S14 18 17 0.25 20 8 0.30
S21 15 20 0.70 16 12 0.71
S22 15 20 0.50 22 12 0.51
S23 15 20 0.40 20 10 0.39
S24 15 20 0.25 20 8 0.25
S31 12 25 0.70 13 12 0.70
S32 12 25 0.50 18 12 0.50
S33 12 25 0.40 16 10 0.39
S34 12 25 0.25 25 10 0.25
Post-1985 T2
lw = 750 cm
lw/tw tw (cm) ρw (%) proposed spacing (cm) db (mm) ρw (%) actual
S11 38 20 0.60 20 12 0.57
S12 38 20 0.50 22 12 0.52
S13 38 20 0.40 20 10 0.40
S14 38 20 0.25 20 8 0.25
S21 30 25 0.60 15 12 0.60
S22 30 25 0.50 18 12 0.50
S23 30 25 0.40 15 10 0.42
S24 30 25 0.25 25 10 0.25
S31 25 30 0.60 12 12 0.63
S32 25 30 0.50 15 12 0.50
S33 25 30 0.40 20 12 0.38
S34 25 30 0.25 22 10 0.24
90
As before, all the wall panels were modeled using a concrete strength of 25 MPa and a
yield strength of the steel of 420 MPa. All the simulations considered a wall panel fixed at the
bottom, with a lateral load applied at the top until a drift of 2.5% or 5% was reached, for aspect
ratios of 1.5 and 0.3, respectively. Additionally, two axial load levels ( and )
were studied to represent typical gravity load levels observed in Chilean buildings. In all,
considering all cross-sections, aspect ratios and axial load levels, a total of 180 wall panels were
modeled in LS-DYNA. The results of the pushover simulations are then classified in order to
develop shear model parameters based on certain characteristics of the wall panel. The two axial
load levels and two aspect ratios considered in this study should be considered an initial effort to
develop a framework for modeling the shear behavior of walls and these studies need to be
supplemented in the future with additional aspect ratios and axial load levels.
4.2.1 Results
The pushover curves obtained from the simulations are shown in Figure 4.5 to Figure 4.8.
They are classified according to year of construction, type, aspect ratio, values, and
normalized by wall height on the abscissas and axial load level on the ordinates.
The figures show different types of curves, which includes a descending branch for Type
2, AR= 1.5 walls at 25% of axial load level, generally independent of the year of construction.
The rest of simulations show a plateau after the peak strength has been reached. Therefore,
according to the results obtained from the simulations, four groups were created in order to
classify the different sections: N10AR03, N25AR03, N10AR15, and N25AR15. The following
notation was used: N refers to the axial load level (10% or 25%) and AR refers to the aspect ratio
(0.3 or 1.5). Four different shear models were developed as described in the following section.
91
AR 0.3 AR 1.5
2.5 0.6
0.5
2
η=10%
η=10%
Base Shear / Axial Load
(a)
AR 0.3 AR 1.5
2.5 0.6
0.5
2 η=10% η=10%
Base Shear / Axial Load
0.4
(b)
92
AR 0.3 AR 1.5
2.5 0.6
0.5
2
η=10% η=10%
Base Shear / Axial Load
1.5
0.3
η=25%
1
η=25%
0.2
S31-ρw = 0.75% S31-ρw = 0.75%
(c)
Figure 4.5 (continued): Pushover curves for Type 1, Pre-1985 wall panels:
a) =15, b) =12, c) =10.
AR 0.3 AR 1.5
1.8 0.45
η=10%
1.6 0.4
η=10%
1.4 0.35
Base Shear / Axial Load
1.2 0.3
1
η=25% 0.25 η=25%
0.8 0.2
0.6 0.15
S11-ρ = 0.31% S11-ρ = 0.31%
w
w
0.4 S12-ρ = 0.25% 0.1
w S12-ρ = 0.25%
w
S13-ρw = 0.20% S13-ρ = 0.20%
0.2 0.05 w
0 0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(a)
93
AR 0.3 AR 1.5
1.8 0.45
η=10%
1.6 0.4
η=10%
1.4 0.35
Base Shear / Axial Load
1
η=25%
0.25 η=25%
0.8 0.2
0.6 0.15
0 0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(b)
AR 0.3 AR 1.5
1.8 0.45
η=10%
1.6 0.4
η=10%
1.4 0.35
Base Shear / Axial Load
1.2 0.3
1
η=25% 0.25 η=25%
0.8 0.2
0.6 0.15
0 0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(c)
Figure 4.6 (continued): Pushover curves for Type 2, Pre-1985 wall panels:
a) =30, b) =25, c) =20.
94
AR 0.3 AR 1.5
2.5 0.6
0.5
2 η=10% η=10%
Base Shear / Axial Load
1.5
0.3
η=25%
η=25%
1
0.2
S11-ρ = 0.68%
S11-ρ = 0.68% w
w
S12-ρ = 0.50%
0.5 S12-ρw = 0.50% w
0.1 S13-ρ = 0.40%
S13-ρ = 0.40% w
w
S14-ρw = 0.30%
S14-ρw = 0.30%
0 0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(a)
AR 0.3 AR 1.5
2.5 0.6
0.5
2 η=10% η=10%
Base Shear / Axial Load
0.4
1.5
0.3
η=25%
η=25%
1
(b)
95
AR 0.3 AR 1.5
2.5 0.6
0.5
2 η=10% η=10%
Base Shear / Axial Load
1.5
0.3
η=25%
η=25%
1
0.2
S31-ρ = 0.70% S31-ρ = 0.70%
w w
(c)
Figure 4.7 (continued): Pushover curves for Type 1, Post-1985 wall panels:
a) =18, b) =15, c) =12.
AR 0.3 AR 1.5
2.5 0.6
0.5
2
η=10% η=10%
Base Shear / Axial Load
0.4
1.5
0.3
η=25%
1 η=25%
0.2
S11-ρw = 0.57%
S11-ρw = 0.57%
S12-ρ = 0.52%
0.5 S12-ρw = 0.52% w
0.1
S13-ρw = 0.40%
S13-ρ = 0.40%
w
S14-ρw = 0.25%
S14-ρw = 0.25%
0 0
0 1 2 3 4 5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(a)
96
AR 0.3 AR 1.5
2.5 0.6
0.5
2
η=10% η=10%
Base Shear / Axial Load
1.5
0.3
η=25%
1 η=25%
0.2
S21-ρ = 0.60% S21-ρ = 0.60%
w w
(b)
AR 0.3 AR 1.5
2.5 0.6
0.5
2
η=10% η=10%
Base Shear / Axial Load
0.4
1.5
0.3
η=25%
1 η=25%
0.2
S31-ρ = 0.63% S31-ρ = 0.63%
w w
(c)
Figure 4.8 (continued): Pushover curves for Type 2, Post-1985 wall panels:
a) =38, b) =30, c) =25.
97
4.3 Developing the Shear Model
As mentioned before, the generated pushover curves were classified into four different
groups so as to develop four different empirical shear models based on the axial load level and
the aspect ratio of the walls investigated. Each of the shear models consist of a four-point multi-
linear relationship as the one illustrated in Figure 4.9, where refers to the cracking point of
the concrete, is the yielding point, is the starting point of the descending branch, and
Vy=Vu α
Vf
k2
Vcr
k1
-δf -δδu -δy -δcrr
k1 δcr δy δu δf
-Vcr
k2
-Vf
α
-Vy= -Vu
The points described in Figure 4.9 were developed based on observed trends in the
generated pushover curves. The final models presented in this work relate base shear normalized
98
by the axial load () applied to the wall panel with the displacement at the top normalized by the
Cracking point
The cracking base shear is estimated from: , where is obtained by taking the
average value of for the different sets of simulations, that is to say, one value for each
one of the four groups (N10AR03, N25AR03, N10AR15, and N25AR15). The standard
deviation of the averages was checked in every case, in order to ensure that the value estimated
is a good representation of the sets of walls inspected. The cracking displacement was computed
from the value and considering an initial stiffness of for AR=0.3 and for
AR=1.5, where is the gross section of the wall, and is the shear modulus. This difference in
the initial stiffness values is based on the first slope observed in the simulations.
Yielding point
The yielding base shear is defined at the peak of the response obtained from the
simulations, before a visible plateau or a descending branch can be observed in the response. The
yielding base shear normalized by the axial load applied and the corresponding yielding drift
levels are estimated from the simulations using the following form of a regression model:
(4.1)
where, is the reinforcement ratio of the wall panel measured as a percentage value. A
unique set of factors were obtained for the estimation of yield shear and yield drift,
respectively. To ensure an appropriate regression model of the two variables were obtained,
99
values were monitored. In the four models developed, the range for the estimate of the yield
shear was 0.9 – 0.8, and for the yield drift the range was 0.8 – 0.6.
Descending branch
The final point on the force-displacement response defines the descending branch from
the yield plateau. This point is defined as a fraction of the yield base shear ( ). The
procedure to obtain the factor ‘’ is the same as the one used to obtain ‘’ for the cracking point,
but in this case the average of the values from the simulations is computed. In the case of
the drift value at the failure point, the estimate is based on regression model similar to the one
In this case, the ultimate base shear is taken equal to and the ultimate drift is
computed from:
(4.2)
were is estimated using a regression model similar to the format used for the estimation
Table 4.8 and Table 4.9 summarize the results of the model formulation. Table 4.8 lists
the factors of the regression model of the four different variables estimated through this
approach. Table 4.9 presents the factors that multiply to obtain and . Additionally, Table
4.10 specifies the different equations to estimate the base shear and drift pairs for each point of
100
Table 4.8: Computed factors for regression models
Variable b1 b2 b3 r2
0.048 0.719 0.256 0.72
2.463 -0.021 0.266 0.93
N10AR03
3.371 -0.021 0.021 0.01
-0.395 -0.188 -0.003 0.01
0.058 0.795 0.466 0.71
0.593 0.004 0.354 0.94
N10AR15
1.322 0.096 0.099 0.15
-0.078 -0.117 0.103 0.00
0.048 0.616 0.054 0.59
1.271 -0.047 0.108 0.83
N25AR03
1.800 0.173 0.047 0.15
-0.0001 2.438 0.984 0.55
0.042 0.760 0.317 0.77
0.306 0.013 0.190 0.80
N25AR15
0.832 0.185 0.064 0.06
-36.505 -1.839 -0.658 0.53
Table 4.9: Mean and standard deviation values for estimated variables
101
4.3.2 Validating the proposed shear model
From the model formulation presented in Table 4.10, random wall panels from the LS-
DYNA simulations were chosen to assess the accuracy of the proposed shear model. For each
type and wall group (based on period of construction) selected, the shear model was evaluated
and compared with the results from the finite element simulations. Figure 4.10 to Figure 4.13
illustrate the comparison for the different walls selected. In each plot the first 3 characters of the
plot title refers to the section name (please see Table 4.4 to Table 4.7), the next character
specifies the aspect ratio of the wall (a for AR=0.3 and c for AR=1.5), and the last two characters
(a) (b)
(c) (d)
Figure 4.10: Comparison of proposed shear model versus FE simulation: pre-1985 Type 1 wall
panels: a) AR=0.3, axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3,
axial load ratio = ; d) AR=1.5, axial load ratio =
102
(a) (b)
(c) (d)
Figure 4.11: Comparison of proposed shear model versus FE simulation: pre-1985 Type 2 wall
panels: a) AR=0.3, axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3,
axial load ratio = ; d) AR=1.5, axial load ratio =
103
(a) (b)
(c) (d)
Figure 4.12: Comparison of proposed shear model versus FE simulation: post-1985 Type 1 wall
panels: a) AR=0.3, axial load ratio = ; b) AR=1.5, axial load ratio = ; c) AR=0.3,
axial load ratio = ; d) AR=1.5, axial load ratio =
104
(a) (b)
(c) (d)
Figure 4.13: Comparison of proposed shear model versus FE simulation: post-1985 Type 1 wall
panels: a) AR=0.3, axial load ratio = ; b) AR=0.3, axial load ratio = ; c)
AR=0.3, axial load ratio = ; d) AR=1.5, axial load ratio =
In most of the cases, the plots in Figure 4.10 to Figure 4.13 show that the proposed shear
model reproduces reasonably the shear response obtained in the detailed finite element
simulations. In all cases, the estimates of the cracking and yielding point estimations are very
close to the values obtained in the simulations. Additionally, when the simulation illustrates a
steep descending branch, the shear model is capable of representing the expected behavior.
Based on the findings of the study, for wall sections with aspect ratios of 0.3 or 1.5, and
axial load levels of and ; the shear model proposed is capable of predicting
105
the shear response of the wall element. The proposed shear model can be used to represent
completely the shear response of the horizontal spring in the MVLE model.
106
Chapter 5
Chile Earthquake
The most recent earthquake that hit Chile on February 27th, 2010 was the sixth largest
that has been recorded until now, and therefore it presents an unprecedented opportunity to study
its effect on the performance of existing structures. Given the observed widespread damage to
shear wall buildings; it is of great importance to carry out investigative studies regarding the
structural behavior of these buildings for improving future design. To achieve this objective it
was imperative to obtain the structural layouts as well as damage reports of each of the two
In this chapter, sections of actual damaged buildings during 2010 Chilean earthquake are
OpenSEES platform. In order to ensure that the wall section modeled represents accurately the
part of the selected building, a preliminary calibration study with the elastic properties of the full
building is performed. The elastic dynamic properties (modal frequencies) are obtained through a
3D model built in SAP2000. The OpenSEES simulations are first calibrated to match the
dynamic properties of the 3D elastic model. The calibrated models then forms the basis for
107
nonlinear simulations of the damaged wall sections and thereby also evaluate the performance of
During July and August 2010 a field trip was made by the author to Chile where
information about structural layouts and the corresponding reports on the degree of damage to a
group of buildings were collected. Field trips were organized to the most damaged cities: Talca,
Constitución, Viña del Mar, Santiago and Concepción (Figure 5.1). This allowed evaluating
which city had the most representative set of RC buildings to carry out this research. The city of
Talca was not appropriate because many of the damaged buildings were composed of adobe
was mostly a consequence of the tsunami. Therefore, the study focused on the other three major
Chilean cities: Viña del Mar, Concepción, and Santiago, where shear-wall RC buildings
Two trips were made to Concepción. The first trip was orientated to get in contact with
the city government and obtain the permits to enter to the damaged buildings. The second was a
two-day trip in order to get data from damaged buildings. Six damaged building were inspected
In addition, two additional field trips were accomplished in Viña del Mar. At this
location, with the assistance of structural engineers in the city more information was collected
about the damaged buildings in the city as well as other regions of Chile, such as Concepción.
Structural drawings and damage reports were obtained. Finally, focusing on the largest city in
108
Chile, Santiago, some buildings were severely damaged and others behaved exceptionally well.
The final data included the structural layouts and damage reports of 33 buildings, as
shown in Table 5.1 (Note: names and addresses of the buildings have been omitted). Within this
sample set, slightly, moderately and severely damaged buildings can be distinguished. The
buildings in Chile.
109
Table 5.1: Description of the gathered data during the field trip
Two post-1985 buildings were modeled in SAP2000 to obtain their elastic pre-damaged
dynamic properties through a 3D model representation. The buildings represent the modern
design practice that was prevalent in Chile before the 2010 earthquake.
110
5.2.1 Building 1
ongoing demolition order. The lateral resistant system consists of reinforced concrete walls
distributed in both directions. It was built in 2003, has one-basement and 19 floors, and was a
residential-use building.
Its plan configuration is almost rectangular. The layout is symmetric in one direction, but
asymmetric in the perpendicular direction. The distribution of the walls is similar in both
directions. The plan is reduced in both directions in the two upper floors. Typical wall
thicknesses are 15-25 cm in the basement to the 13th floor, 15-20 cm in the upper floors, with an
overall typical thickness of 20 cm in all the floors. The type of concrete used in the construction
was specified as: = 25 MPa from basement to floor 4, = 20 MPa for floors 5 to 19, and =
35 MPa for two of the four perimeter sides. The type of steel used for the reinforcement has a
Significant damage was observed in RC walls from the first basement to the third floor,
although the damage was visible up to the fifth floor. It included spalling of the concrete,
buckling and fracture of the longitudinal bars and shear damage. Some walls in the first floor
experience out-of-plane damage. It could be observed flexural plus compression behavior, and
transverse reinforcement with U and L shape close at the boundary elements of the RC walls.
Figure 5.2 illustrates some of the damage reported during the 2010 earthquake for this building.
111
Figure 5.2: Observed Damage in RC walls (Building 1)
A simulation model for this building was developed in SAP2000. All the beams and
columns were modeled using frame sections, and RC walls were modeled using area sections
with a mesh size of approximately 50x50 cm for each element. A convergence study was carried
out in order to establish the maximum mesh size for the wall elements in the building model.
Base moment, base shear, and top displacement for two different walls were inspected. Figure
5.3 shows the displacement profile of the wall for different mesh sizes. As can be seen,
considering smaller meshes did not improve the results, therefore a mesh size of 50x50cm was
112
(a)
(b)
Figure 5.3: Convergence study for SAP2000 models: (a) Wall A; (b) Wall B
The simulation model also included the basement levels of the building but lateral
restraints were specified at the ground level. The joints at the base of the building were
considered as fixed supports. Slabs were not modeled, but independent diaphragm constraints
were specified on each floor so that lateral motion in two orthogonal directions and torsion of the
floor slab was permitted. Additional mass source considering superimposed dead loads (besides
113
self-weight) and 25% live loads were added to the model by creating a special joint at the
centroidal point of the floor plan at each level. Figure 5.4 shows a general view of the SAP2000
model for Building 1. An eigenvalue analysis in SAP2000 yielded the following modal periods
in the two principal directions: = 1.04 sec., and = 1.38 sec. These estimates will be the
target periods for the 2D nonlinear wall models to be simulated in OpenSEES (OpenSEES,
2014).
A wall section (identified in the building drawings as Axis 6) that experienced significant
damage in the 2010 earthquake was selected to be modeled in OpenSEES. This axis is identified
in Figure 5.5, where both the elevation view and the SAP2000 representation are shown.
Along this axis the second floor wall was subjected evident shear damage, as can be
observed in Figure 5.6. This wall is considered a flag wall with a high degree of discontinuity
between the second and the third floor. Damage was also present in the first floor, but the second
floor wall was selected for the simulation using the wall model and shear model described in
Chapter 3 and Chapter 4. The wall is oriented in the x-direction, therefore the period to be
114
Figure 5.4: SAP2000 model for Building 1
115
Wall to be
inspected
(a) (b)
Figure 5.5: Frame selected for detailed modeling: a) Elevation view, b) SAP2000 model detail
116
5.2.2 Building 2
The second building is located in Viña del Mar. As in the case of Building 1, the lateral
resisting system consists of reinforced concrete walls distributed in both directions. After the
2010 earthquake this building was repaired and is currently in use. It was built in 1996, has one
The plan can be considered rectangular in all the floors. In the basement, the plan is
larger, to allow for parking. The amount of walls is not the same in both directions. Wall
The type of concrete used for the design has a specified strength = 22.5 MPa for the
entire building and the type of steel used for the reinforcing steel has a yield strength of 420MPa
Shear cracking and flexural damage was observed in walls that formed the emergency
stairwell from the basement to floor 4. Tension damage was evident in the south walls of the
building in the first floor. In the north side of the building some walls experienced compression
failure, where buckling and fracture of the longitudinal bars had occurred. This compression
failure caused an overall rotation of the building. Figure 5.7 shows some of the damage reported
117
Figure 5.7: Damage in RC walls for Building 2 (provided by Prof. Jorge Carvallo)
features as for Building 1. Beams and columns were modeled with frame sections, and RC walls
with 50x50cm shell elements composed of area sections. The basement was also modeled and
the joints at the base were fixed to the ground. Floor slabs were not modeled explicitly since
diaphragm constraints were imposed on each floor level. As in the case of Building 1,
supplementary mass sources to account for superimposed dead load and 25% of the live load was
included in special centroidal joints in the floor plan at each level. Figure 5.8 shows a general
118
view of the SAP2000 model for Building 2. A modal analysis in SAP2000 yielded the following
The frame selected for detailed nonlinear analysis is identified in the building drawings
as Axis 1, which corresponds to the y-direction of the SAP2000 model. Figure 5.9 provides
details specified in the building drawing of the selected section including an elevation view of
the SAP model. Following the 2010 earthquake, widespread flexural-compression damage was
observed in the first floor. The selected wall section along Axis 1 was among the damaged walls
in the building with significant damage to the base as well as buckling of the wall boundary
elements. The wall was classified as a flag wall with a cross-sectional discontinuity along the
length between the first and second floor. Additionally, the wall did not have confinement, and L
and U shape hoops were observed. Since the wall to be considered in the nonlinear simulation is
oriented in the y-direction, the target period for the wall section in the OpenSEES model is 0.67
sec.
119
Figure 5.8: SAP2000 model for Building 2
120
Wall to be
inspected
(a) (b)
Figure 5.9: Frame selected for detailed modeling: a) Elevation view, b) SAP2000 model detail
The selected frames from each building were modeled using OpenSEES to examine the
effectiveness of the simplified models to reproduce the observed wall damage. The walls were
modeled as MVLEMs, columns and beams were modeled as ‘beamWithHinges’ elements, and
elements.
121
For developing the MVLE models to be used in the wall sections, the guidelines from
Chapter 3 were followed, with the exception of the horizontal shear spring, which will be
explained later. In summary, several fiber sections were considered along the wall length. Each
one of the fibers was modeled using nonlinear trusses elements, and consists of a portion of the
wall cross-section. Figure 5.10 shows an example of the representative details of the wall fibers.
The minimum number of fibers considered was 6, with a maximum of 8 elements stacked along
the wall height. The walls with a larger number of elements along the height correspond to the
walls located at the bottom of the frame where much of the nonlinearity was concentrated.
Master
M
nodes
n
k1 k2 kn-1 kn
Figure 5.10: Fibers distribution scheme
The master nodes connect each element to one another; therefore the shear springs
remain in the same horizontal position (with respect to the wall length) for the total height of the
frame. Details of the elements used in the modeling are described in the following sections.
With respect to the horizontal shear springs, ‘zeroLength’ elements were used. Spring
properties in the three local coordinate directions need to be assigned to this type of element. For
122
the directions orthogonal to the imposed displacement of the wall, elastic materials with
negligible stiffnesses were specified, and for the spring in the direction of the wall displacement,
described by the four points generated from the shear model developed in section 4.3.1 and
Figure 4.9. Since the curve was generated using the normalized variables and , the
base shear is obtained from the estimated axial load applied to each wall, and the drift
according to the maximum displacement experienced by each floor during the simulations.
The next step was to calibrate the OpenSEES model to the SAP2000 3D building model
by matching the fundamental periods in both models. Masses were assigned to the master nodes
at the elements located at the top of each floor. These masses were computed from the
distributed axial load acting on each floor and considering the tributary area corresponding to the
modeled walls. Additionally, the masses were proportionally reduced based on the location of
the selected frame in relation to the total floor plan area. The axial load was applied at each floor
level to the master node of each corresponding element based on the calculated distributed axial
load of . This value represents the average axial load in Chilean buildings built
after 1985.
The frame models were fixed at the bottom, free at the top, but the node at the top of the
basement was also restrained against moving in the direction of the applied loading. This
restriction was imposed because the basement was located below the ground level.
The simulations were performed by means of a pushover analysis, wherein the simulation
model was loaded laterally using an inverted triangular distribution up to a top drift of 5% of the
123
frame height. The frame was loaded in both positive and negative directions of the loading axis
The frame selected to represent wall damage in Building 1 has 17 floors and one
basement. Walls from the basement to the third floor were modeled with 8 elements stacked
along the wall height (corresponding to the value of in Chapter 3), walls from the 4th floor to
the 11th floor have 5 elements, and walls from floor 12 to floor 17, 4 elements. With respect to
the modeling of the wall length, (corresponding to the value of in Chapter 3): the basement, 1st,
and 2nd floors consist in 10 fibers, the 3rd to the 14th floor walls each has 12 fibers, and from the
As mentioned before, the masses applied to the frame model were placed in each master
node of the elements located at the top and bottom of each floor. The mass magnitudes are listed
in Table 5.2. These masses represent the magnitudes estimated previously based on average axial
load ratios in post-1985 Chilean walls and had to be reduced by a factor of 1.8 in order to
Table 5.3 presents the parameters and that are needed for modeling the shear
124
Table 5.3: Wall parameters, Building 1
Floor
B-2 0.25 32
3-14 0.25 37
15-17 0.25 30
The pushover response of the entire frame-wall system is presented in Figure 5.11 for
both the positive and negative x-directions. While the lateral strengths in both directions are
approximately similar, the ductility capacity is lower when the frame is loaded in the negative x-
direction. The particular wall whose damage is being investigated is the 2nd floor wall. The
flexural response of the 8 sub-elements of this wall is illustrated in Figure 5.12. Not
unexpectedly, similar behavior is observed for the 8 elements stacked along the height of the
inspected wall. In the figure, element 1 corresponds to the bottom part of the wall and element 8
is the top segment of the wall. Figure 5.13 shows the variation of the axial forces in the axial
springs in the wall. These axial forces are used to estimate the critical section of the wall that
failed in shear since the shear capacity is controlled by the axial force in the wall as presented in
As shown earlier in Figure 5.6, the type of damage observed after the 2010 earthquake in
the 2nd floor wall was classified as a typical shear failure. The results obtained from the
OpenSEES simulations indicate that the wall is currently failing in flexure, as presented in
Figure 5.14 and Figure 5.15 which shows a comparison of the shear model of the 2nd floor wall
and its behavior during the simulation of the lateral response of the entire wall-frame segment
across the entire height of the building. It is seen that the predicted shear capacity is higher than
125
the lateral capacity when flexural failure occurs. Note that Figure 5.14(b) and Figure 5.15(b)
presents three different shear capacity models. This reflects the changing axial state in the wall
due to the imposed lateral displacement. In the present study three axial states of the wall are
considered: a preliminary state referred to as the Initial Shear Model, a secondary state when the
flexural limit state is reached and a final state referred to as the Critical State when tension
failure of the wall initiates (note the axial force fluctuation in the axial springs shown in Figure
5.13).
The observed shear failure of the wall may be attributed to the structural layout of the
building. From the modal analysis it was observed that the predominant dynamic mode was a
coupled lateral-torsional mode (Figure 5.16). This coupled motion likely induced a torsional
response of the wall that led to a sudden decrease in the shear strength predicted by the proposed
600 600
500 500
Base Shear (kN)
Base Shear (kN)
400 400
300 300
200 200
100 100
0 0
0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4 0.5
Drift (%) Drift (%)
(a) (b)
126
7
x 10
3
2.5
0.5
0
0 0.2 0.4 0.6 0.8 1
Drift (%)
(a)
6
x 10
2
Bending moment (kN-mm)
-2
-4
ELE 1
-6 ELE 2
ELE 3
-8 ELE 4
ELE 5
ELE 6
-10 ELE 7
ELE 8
-12
0 0.1 0.2 0.3 0.4 0.5
Drift (%)
(b)
Figure 5.12: Bending moment variation along the height of 2nd floor wall, Building 1
(ELE 1: element at the bottom; ELE 8: element at the top of the wall)
a) positive - direction, b) negative - direction
127
500
-500 1-left
Axial force (kN)
2
-1000 3
4
-1500 5
6
-2000 7
8
-2500 9
10-right
-3000
0 5 10 15 20
Displacement (mm)
(a)
500
1-left
Axial force (kN)
-500 2
3
-1000 4
5
-1500 6
7
8
-2000
9
10-right
-2500
0 2 4 6 8
Displacement (mm)
(b)
Figure 5.13: Axial load distributions of the axial springs along the wall length: a) positive -
direction, b) negative - direction
128
600 5000
4500 Model A
500 Model B
4000
Model C
3500
400
Shear (kN)
3000
Shear (kN)
300 2500
2000
200
1500
1000
100
500
0 0
0 5 10 15 20 0 5 10 15 20 25
Displacement (mm) Displacement (mm)
(a) (b)
Figure 5.14: Comparison between the response of the 2nd floor wall and the shear capacity models
for positive -direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear
model, Model B = Shear model at flexural limit state, Model C = Shear model at critical state
600 6000
500 5000
400 4000
Shear (kN)
Shear (kN)
300 3000
200 2000
Model A
100 1000 Model B
Model C
0 0
0 2 4 6 8 10 0 5 10 15 20 25
Displacement (mm) Displacement (mm)
(a) (b)
Figure 5.15: Comparison between the response of the 2nd floor wall and the shear capacity models
for negative -direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear
model, Model B = Shear model at flexural limit state, Model C = Shear model at critical state
129
Figure 5.16: Combined lateral-torsional response of Building 1
This frame consists of 10 floors plus one basement. From the basement to the third floor
8 elements were stacked along the wall height (corresponding to the value described to
Chapter 3), from the 4th floor to the 6th 5 elements were used, and finally 4 elements were used
from the 7th floor to the 10th floor. Additionally, different number of fibers was considered based
on the wall length (corresponding to the value defined in Chapter 3): the basement wall was
divided into 17 fibers, the first floor wall into 10 fibers, walls in floors 2 through 10 had 13
The masses were calculated according to the axial load applied to the tributary area
corresponding to the wall in each floor, and these magnitudes are presented in Table 5.4. These
130
masses represent 1/3 of the average axial load ratios in post-1985 Chilean walls and were the
required mass magnitudes to calibrate the frame model and obtain the target period. The frame is
located at one of the building edges, therefore the effective mass in this wall is slightly lower
The modeling of the shear component (horizontal spring) of the MVLEs follows the
model described in Chapter 3. Table 5.5 shows the parameters and of each wall as
Floor
B 0.36 44
1 0.36 14
2-5 0.36 21
6-10 0.20 21
Once again, the main objective of the simulation was to reproduce the type of failure
observed after the 2010 earthquake. In the case of the frame selected from Building 2, the wall
located on the first floor (Figure 5.9) experienced compression and flexural failure at the bottom.
131
The first part of the simulation examines the pushover response of the entire frame
(Figure 5.17). The capacity of the frame in the positive -direction is slightly lower than its
strength in the negative -direction, though the ductility capacity is approximately the same in
both directions. In examining the flexural behavior of the damaged first floor wall, the bending
capacity in the eight elements across the wall height is seen to vary by about 15% (Figure 5.18).
In the figure, element 1 corresponds to the bottom of floor 1, and element 8 is top segment of the
floor. Figure 5.19 shows the axial force variation in the axial elements along the length of the
selected wall. As discussed in the assessment of Building 1, the net axial force on the critical
Results from the OpenSEES simulations of the entire wall indicate a flexural/
compression behavior, as seen in Figure 5.20 and Figure 5.21. Additionally, when the 1st floor
wall behavior is compared with the shear capacity model (shown in Figure 5.20b and Figure
5.21b) it can be observed that the failure of the wall was controlled by flexure. Hence it can be
concluded that the observed failure mode was correctly reproduced in the OpenSEES simulation.
250 300
250
200
Base Shear (kN)
200
150
150
100
100
50
50
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
Drift (%) Drift (%)
(a) (b)
Figure 5.17: Pushover curve for full frame, Building 2:
a) positive - direction, b) negative - direction
132
6
x 10
16
14
10 ELE 1
ELE 2
ELE 3
8
ELE 4
ELE 5
6
ELE 6
ELE 7
4
ELE 8
0
0 0.5 1 1.5 2 2.5
Drift (%)
(a)
6
x 10
12
10
Bending moment (kN-mm)
4
ELE 1
ELE 2
2 ELE 3
ELE 4
0 ELE 5
ELE 6
-2 ELE 7
ELE 8
-4
0 0.5 1 1.5 2 2.5
Drift (%)
(b)
Figure 5.18: Bending moment variation along the height of 1st floor wall, Building 2
(ELE 1, element at the bottom; ELE 8, element at the top of the wall):
a) positive - direction, b) negative - direction
133
1000
500
1-left
0 2
Axial force (kN)
3
-500 4
5
-1000 6
7
-1500 8
9
-2000 10-right
-2500
0 20 40 60 80
Displacement (mm)
(a)
1500
1000
1-left
2
500
3
Axial force (kN)
4
0
5
6
-500
7
8
-1000
9
10-right
-1500
-2000
0 10 20 30 40 50
Displacement (mm)
(b)
Figure 5.19: Axial load variations in the axial springs along the wall length:
a) positive - direction, b) negative - direction
134
350 350
300 300
250 250
Shear (kN)
Shear (kN)
200 200
150 150
100 100
50 50 Model A
Model B = Model C
0 0
0 10 20 30 40 50 60 70 80 0 5 10 15 20 25
Displacement (mm) Displacement (mm)
(a) (b)
Figure 5.20: Comparison between the response of the 1st floor wall and the shear capacity models
for positive -direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear
model, Model B = Shear model at flexural limit state, Model C = Shear model at critical state
350 400
300 350
300
250
250
Shear (kN)
Shear (kN)
200
200
150
150
100
100
50 50 Model A
Model B = Model C
0 0
0 10 20 30 40 50 60 70 80 0 5 10 15 20 25
Displacement (mm) Displacement (mm)
(a) (b)
Figure 5.21: Comparison between the response of the 1st floor wall and the shear capacity models
for negative -direction: a) Response, b) Shear capacity models proposed: Model A = Initial shear
model, Model B = Shear model at flexural limit state, Model C = Shear model at critical state
135
Chapter 6
Conclusions
The main objective of this research is the modeling of RC shear-walls by developing an
efficient yet reliable shear model to be used as input into a macro-model representation of the
wall element. Currently, only complex and computationally intensive high fidelity models can
predict reasonably well the behavior of RC walls under severe loadings, and the time, effort and
knowledge needed to build such models accurately hinders the proper use of nonlinear finite
elements in general practice. Therefore, the development of suitable simple models that work
with several types of walls will have a significant impact on the state-of-the-art in shear wall
modeling.
The first task carried out to achieve the proposed research objective was an evaluation of
different existing macro-models. The performance of the different models was assessed by
comparing predicted numerical responses with available experimental data so as to select the
most suitable ‘base’ macromodel for the present study. The two severe earthquakes in Chile in
1985 and 2010 provide an excellent opportunity to validate new models against observed
behavior of damaged shear walls. Furthermore, the study also attempts to gain a better
136
In the following sections, the main conclusions of this study are outlined. The
conclusions are divided into different sub-sections based on the main theme of the previous
Earthquakes
A summary of the regulations that existed in Chile prior to the 1985 earthquake are
compared to the post-1985 changes with respect to the design of shear walls. The main goal of
this phase of the work was to assess the influence of changes to design practice that may have
affected the performance of the post-1985 walls. The main findings from this part of the study
• Pre-1985 code provisions in Chile were not so important because most of the shear
wall design was based on past tradition and experience rather than adherence to code-
based guidelines.
• Buildings built before 1985 were designed without any particular provisions for shear
reinforcement. It can be stated that shear demands were generally met through the
provision of an appropriate amount of wall thickness. This design criteria was not
enforced by the code, but was part of the design tradition among architects and
engineers.
• Since the behavior of RC buildings during severe earthquakes in Chile has generally
been satisfactory, the evolution of the codes has relied more on tradition and past
practice rather than input from foreign sources (such as ACI). This is an important
137
fact to consider when examining the changes imposed in the 1996 Chilean seismic
provisions.
• The degree and extent of damage to buildings in Chile after the two earthquakes was
much lower than expected, considering the severity of the seismic events. The
damage observed changed from flexural-shear type failures in the 1985 earthquake, to
observed response of different shear walls, a comparison between three macro-models was
undertaken. To obtain suitable simulations, available test data of three shear-walls of varying
cross-section tested under cyclic loading was used. The main findings from this study are
described below:
• As expected, the three macro-models are capable of predicting reasonably well the
strength and the global response of slender walls. However, when the wall enters the
inelastic range, some disagreement between the simulated response using the three
macromodels and the experimental data is evident. This discrepancy, especially for
the Panel model and the MVLEM, is diminished when the aspect ratio of the wall
increases.
138
• In the case of intermediate and squat walls, only the MVLEM is capable of predicting
the full response of the wall, reaching the expected strength and following the plateau
the input of the shear spring properties. In the present study, this was determined from
estimating the shear behavior of the wall sections. The EBM and the Panel model are
shear component. Whereas the Panel model tends to underestimate the wall strength
and is not capable of reproducing post-yield behavior, the EBM tends to predict the
• With respect to the aspect ratio study, the results obtained with the MVLEM tends to
provide the most reasonable and consistent estimates of strength and initial stiffness.
accurate estimate of the shear behavior, it may be stated that the EBM estimates of
post-yield response for the squat walls may be unreliable. Given the observations in
section 3.1.3, it may be concluded that the Panel model predictions of post-yield
• The specified force-deformation behavior of the shear spring has a major influence in
the predicted response of both the MVLEM and EBM. Therefore, further
needed to obtain more reliable and consistent results. The Softened Membrane Model
139
used in this study to derive the shear properties offers a good starting point but
The response of the two types of walls was carried out using the MVLEM with horizontal
shear spring properties obtained from an estimation made from an experimental database of RC
walls. The main conclusions from the results of the simulations are:
• When the behavior of pre-1985 and post-1985 walls is analyzed, almost in every case,
pre-1985 walls were found to have higher strength capacities than post-1985 walls.
The only exception was observed for squat walls, in which the higher strength for
post-1985 walls could be attributed to higher axial load ratios as well as an increase in
• The reduction in the strength capacity of post-1985 walls can be attributed to the
reduction in the wall thickness over the years since the 1985 earthquake.
• The typical post-1985 walls that were investigated presented a response governed by
their flexural component (expect for squat walls, as expected), which is consistent
with the behavior observed during the 2010 earthquake, and confirms that MVLEM is
• In the case of pre-1985 walls, the shear component becomes less important when the
aspect ratio increases. For of larger wall lengths (short aspect ratios), the shear
140
• The axial load ratio on the walls had a significant effect on the wall response. Higher
axial load ratios increased the strength of walls in all cases. Pre-1985 walls still show
greater strength capacity at the same level of axial load than post-1985 walls.
• The behavior of pre-1985 walls with larger aspect ratios and higher axial load ratios
are more influenced by their flexural component, which decreases when the aspect
ratio decreases. In the case of wide walls with low aspect ratios, as expected, the
shear component is the one that governs the behavior of the wall.
• In the case of post-1985 walls with high aspect ratios (greater than 1.5), the results
applied axial load. In the case of walls with lower aspect ratios, the response is
controlled primarily by shear independent of the amount of the axial load applied.
• Although MVLEM does not incorporate true shear-axial load interaction, this can be
implicitly incorporated if the shear parameters specified in the MVLEM takes into
account the critical axial load level on the wall during seismic action.
From the statistical data gathered on existing Chilean buildings, wall parameters were
varied in order to generate different sets of Chilean wall characteristics. The specific
characteristics considered in this study were: wall thickness, wall length, web reinforcement
ratios, axial load applied, and aspect ratio (height-to-length). Based on these control parameters,
several wall panel (without considering boundary elements) cross-sections were generated, and
141
All the cross-sections were part of a simulation program where the wall panels were
pushed laterally at the top until 2.5% or 5% drift was reached. With the result of the simulations
four wall groups were created to develop four different shear models. The shear models were
generated based on comprehensive regression analyses using different functional forms of the
model parameters. The statistical study indicated that the key model parameters were the ratio
between the wall length and the wall thickness, and the web reinforcement ratio. Based on these
two critical parameters, empirical shear models were expressed as a four-point multi-linear
curve, wherein the cracking, yielding and failure points are defined.
Based on comparisons between the empirical models and the simulation results from
detailed finite element analyses of wall panels using LD-DYNA, it was observed that the
proposed shear model works well for the majority of cases, but also exhibits a loss of accuracy
Since the proposed shear model is a function of the axial load on the wall, the model has
the ability to consider axial-load interaction in an implicit manner, meaning that the critical axial
load level during seismic loading needs to be established prior to determining the properties of
the shear spring. The goal of the study is to establish the parameters of the shear spring in the
142
6.4 Assessment of Buildings in Chile with Damaged Walls after the
2010 Earthquake
In the final phase of the study, two Chilean buildings that experienced varying degrees of
damage were modeled in SAP2000. The dynamic properties (lateral periods and mode shapes)
were extracted from the SAP2000 simulations to calibrate two-dimensional nonlinear models of
portions of those buildings developed in OpenSEES. The OpenSEES models were utilized to
assess the behavior of typical wall configurations that experienced damage during the 2010
earthquake. The walls of the selected wall-frames in each building were modeled using the
The wall-frame model for Building 1, which incorporated the shear model with
parameters estimated using the proposed model in this dissertation, was incapable of reproducing
the observed shear failure in the wall at the second floor. The computed critical shear capacity
(as shown in Figure Figure 5.14) is slightly higher than the flexural capacity thereby inhibiting
shear failure of the wall. However, as explained in the previous chapter, the initiation of tension
in the left lower segment of the wall combined with the torsional response of the building may
have resulted in premature shear failure of the wall. On the other hand, the flexural failure of the
wall in Building 2 was predicted more accurately. Even in this case, the estimated critical shear
capacity is fairly low and the model suggests the possibility of a shear-flexure failure.
Findings from this part of the study highlight several limitations of two-dimensional
modeling and the challenges in estimating the critical axial force in the wall. The presence of a
dominant torsional mode can significantly affect the ability of a 2D model to predict the real
behavior of shear walls in a building. Likewise, since axial force is not constant across the entire
143
wall panel, it is essential to identify the critical wall panel and the corresponding critical axial
At the outset of this study it was believed that the shift in the failure mode of Chilean
and higher axial load ratios. It was shown that increased axial loads do improve the shear
performance of the wall. Likewise boundary elements in the post-1985 walls experienced very
Despite these shortcomings the application of the shear model is relatively simple and it
is possible to run a large range of simulations are low computational cost. Additionally, since the
shear component is in series with the flexural component, the MVLEM can correctly identify the
The shear model developed in this study is an initial effort to improve the estimation of
the properties of the nonlinear shear spring. The study revealed important parameters that control
the behavior of a shear wall and has the potential for further improvement as itemized below:
• The finite element (FE) simulations in the present study were carried out over a
limited range of axial load levels and aspect ratios. Additional simulations
considering a full spectrum of aspect ratios and axial load levels are still needed to
improve the predictive capability of the proposed model for a wider range of wall
configurations.
144
• Determining the critical axial load on the wall panel was a challenging issue in the
present study since the wall is represented by multiple axial springs distributed across
the wall length. Further studies are needed to identify a critical shear panel zone that
controls the shear strength of the wall and the corresponding axial load on the wall.
• The finite element and MVLEM simulations were limited to monotonic loading only.
While the FE analyses for monotonic loads may be adequate to estimate shear
properties, it is suggested to extend the MVLEM simulations to cyclic loads and also
the proposed empirical model. For example, the membrane model such as the
Softened Membrane Model (Hsu & Mo, 2010) along with its cyclic formulation is
worthy of further investigation though in the present study the current version of the
softened membrane model was found to be ineffective. This can also lead to studies
issue that could be included by considering modifications to the MVLEM that permit
of wall slenderness which was beyond the scope of the present study.
145
Bibliography
ACI318-05. (2005). Building code requirements for structural concrete (ACI318-05) and
ACI318-83. (1983). Building code requirements for structural concrete (ACI318-83) and
ACI318-95. (1995). Building code requirements for structural concrete (ACI318-95) and
Cassis, J. H., & Bonelli, P. (1992). Lessons Learned from the March 3, 1985 Chile Earthquake
and Related Research. Tenth World Conference of Earthquake Engineering, (pp. 5675-5680).
Balkema, Rotterdam.
Cruz, E., Lüders, C., & Vásquez, J. (1988). Lessons from the March 3, 1985 Earthquake (in
DIN. (1953). Concrete and Reinforced Concrete. DIN 1045, Beuth Verlag GmbH for the
Fischinger, M., Vidic, T., & Fajfar, P. (1992). Nonlinear seismic analysis of structural walls
146
seismic analysis and design of reinforced concrete buildings (pp. 191–202 ). New York,
Gulec, C. K., & Whittaker, A. S. (2009). Performance-Based Assessment and Design of Squat
Tall Buildings in Chile. Civil Engineering Thesis (in Spanish). University of Chile.
Hidalgo, P. A., & Riddell, R. (2001). Fundamentals of Structural Engineering for Architecture
Hidalgo, P. A., Jordan, R. M., & Martinez, M. P. (2002). An analytical model to predict the
Hidalgo, P. A., Ledezma, C. A., & Jordan, R. M. (2002). Seismic Behavior of Squat Reinforced
Hsu, T. C., & Zhu, R. R. (2002). Softened membrane model for reinforced concrete elements in
Hsu, T. T., & Mo, Y. L. (2010). Unified Theory of Concrete Structures. Houston, TX: John
147
INN. (1961). Instituto Nacional de Normalización. NCh430.Of61: Reinforced Concrete: Part II
Kabeyasawa, T., Shiohara, H., Otani, S., & Aoyama, H. (1983). Analysis of the full-scale seven-
story reinforced concrete test structure. Journal (B), The Faculty of Engineering, University
Krolicki, J., Maffei, J., & Calvi, G. M. (2011). Shear Strength of Reinforced Concrete Walls
Lagos, R., & Kupfer, M. (2012). Performance of High Rise Buildings Under the February 27th
Lessons Learned from the 2011 Great East Japan Earthquake, March 1-4, 2012, pp. 1754-
Lagos, R., Marianne, K., Lindenberg, J., Bonelli, P., Saragoni, R., Guendelman, T., et al.
148
LSTC. (2014). LS-DYNA® - Theory Manual. Livermore, CA, USA: Livermore Software
Technology Corporation.
http://www.lstc.com/lspp/
Lisbon, Portugal.
buildings structured with reinforced concrete walls. Pontificia Universidad Católica de Chile.
Massone, L. M., Bonelli, P., Lagos, R., Lüders, C., Moehle, J., & Wallace, J. W. (2012).
Seismic Design and Construction Practices for RC Structural Wall Buildings. Earthquake
Massone, L. M., Orakcal, K., & Wallace, J. W. (2009). Modeling of Squat Structural Walls
Massone, L. M., Orakcal, K., & Wallace, J. W. (2006). Shear–flexure interaction for structural
walls. Deformation capacity and shear strength of reinforced concrete members under cyclic
149
Murray, Y. D. (2007). Users Manual for LS-DYNA Concrete Material Model 159 . U.S.
Prencite Hall.
NIST. (2012). Comparison of U.S. and Chilean Building Code Requirements and Seismic
OpenSEES. (2014, February). Open System for Earthquake Engineering Simulation. Pacific
Orakcal, K., Massone, L. M., & Wallace, J. W. (2009). Shear Strength of Lightly Reinforced
Orakcal, K., Wallace, J. W., & Conte, J. P. (2004). Flexural Modeling of Reinforced Concrete
Pang, X. D., & Hsu, T. T. (1995). Behavior of reinforced concrete membrane elements in shear.
Paulay, T., & Priestley, M. (1992). Seismic Design of Reinforced Concrete and Masonry
Pilakoutas, K., & Elnashai, A. (1995). Cyclic Behavior of RC Cantilever Walls, Part I:
150
Riddell, R., Wood, S. L., & De La Llera, J. C. (1987). The 1985 Chile Earthquake: Structural
Characteristics and Damage Statistics for the Building Inventory in Viña del Mar. Urbana,
Sozen, M. (1989). Earthquake Response of Building with Robust Walls. 5th Chilean
Sritharan, S., Beyer, K., Henry, R. S., Chai, Y. H., Kowalsky, M., & Bull, D. (2014).
Takayanagi, T., Derecho, A. T., & Corley, W. G. (1979). Analysis of Inelastic Shear
Telleen, K., Maffei, J., Heintz, J., & Dragovich, J. (2012a). Practical Lessons for Concrete Wall
Design Based on Studies of the 2010 Chile Earthquake. 15th World Conference of
Telleen, K., Maffei, J., Willford, M., Aviram, A., Huang, Y., Kelly, D., et al. (2012b). Lessons
for Concrete Wall Design from the 2010 Maule Chile Earthquake. Proceedings of the
International Symposium on Engineering Lessons Learned from the 2011 Great East Japan
151
Cross- Sections. Department of Civil Engineering, Clarkson University. Postdam, NY:
Vecchio, F. J., & Collins, M. P. (1986). The modified-compression field theory for reinforced-
K. H. Fafjar P, & K. H. Fafjar P (Ed.), Nonlinear seismic analysis and design of reinforced
Vulcano, A., Bertero, V., & Colotti, V. (1988). Analytical modeling of RC structural walls. 9th
Wallace, J. W., & Moehle, J. P. (2012). Behavior and Design of Structural Walls - Lessons from
Engineering Lessons Learned from the 2011 Great East Japan Earthquake, (pp. 1132-1144).
Tokyo, Japan.
Wallace, J. W., & Moehle, P. J. (1993). An Evaluation of Ductility and Detailing Requirements
of Bearing Wall Buildings Using Data from the March 3, 1985, Chile Earthquake.
Wilson, E. L., & Habibullah, A. (1995). SAP2000, Structrual Analysis Program, Computers and
Wood, S. L. (1991). Performance of Reinforced Concrete Buildings During the 1985 Chile
Earthquake: Implications for the Design of Structural Walls. Earthquake Spectra , 7 (4), 607-
638.
152
Wood, S. L., Wight, J. K., & Moehle, J. P. (1987). The 1985 Chile Earthquake: Observations on
Urbana-Champaign.
Wyllie, L. A., Bolt, B., Durkin, M. E., Gates, J. H., McCornick, D., Smith, P. D., et al. (1986).
Zhou, J., Hirosawa, M., Kondo, T., & Shimizu, Y. (2000). Effect of the torsional moment on the
shear strength of reinforced concrete columns due to eccentric jointing of beam to column .
12th World Conference on Earthquake Engineering 12WCEE. Auckland, New Zealand: New
153
In this appendix the database used to generate the proposed model for the shear behavior
of the RC walls modeled is presented. The database is an extract of the wall test presented in by
other researchers in previous studies (Gulec & Whittaker, 2009), (Orakcal et al., 2009).
Table A.1: Database used for the estimation of the shear strength
Test ρv η
Researcher Specimen ID M/Vlw tw/lw ρh (%) VTEST (kN) VACI (kN) VTEST/VACI
No. (%) (%)
1 Cardenas SW-7 1.08 0.040 0.85 0.27 0.00 519 359 1.44
2 Cardenas SW-8 1.08 0.040 2.87 0.27 0.00 570 418 1.36
3 Cardenas SW-10 1.08 0.040 0.00 0.00 0.00 306 207 1.48
4 Cardenas SW-11 1.08 0.040 0.00 0.69 0.00 609 605 1.01
5 Cardenas SW-12 1.08 0.040 0.00 0.93 0.00 658 746 0.88
6 Cardenas SW-13 1.08 0.040 2.87 0.93 0.00 632 853 0.74
7 Pilakoutas SW4 2.13 0.100 0.50 0.39 0.00 105 206 0.51
8 Pilakoutas SW5 2.13 0.100 0.59 0.31 0.00 112 149 0.75
9 Pilakoutas SW7 2.13 0.100 0.59 0.39 0.00 129 211 0.61
10 Pilakoutas SW8 2.13 0.100 0.50 0.28 0.00 95 146 0.65
11 Pilakoutas SW9 2.13 0.100 0.50 0.56 0.00 98 214 0.46
12 Greifenhagen M1 0.69 0.100 0.34 0.37 2.20 202 222 0.91
13 Greifenhagen M2 0.69 0.100 0.34 0.00 2.20 203 109 1.87
14 Greifenhagen M4 0.69 0.088 0.39 0.26 5.00 135 155 0.87
15 Hidalgo 1 1.00 0.120 0.25 0.13 0.00 198 290 0.68
16 Hidalgo 2 1.00 0.120 0.25 0.25 0.00 270 409 0.66
17 Hidalgo 4 1.00 0.120 0.25 0.38 0.00 324 532 0.61
18 Hidalgo 6 0.69 0.093 0.26 0.13 0.00 309 298 1.04
19 Hidalgo 7 0.69 0.093 0.13 0.25 0.00 364 462 0.79
20 Hidalgo 8 0.69 0.093 0.26 0.25 0.00 374 447 0.84
21 Hidalgo 9 0.69 0.077 0.26 0.26 0.00 258 344 0.75
22 Hidalgo 10 0.69 0.061 0.25 0.25 0.00 187 265 0.71
23 Hidalgo 11 0.50 0.071 0.26 0.13 0.00 235 191 1.23
24 Hidalgo 12 0.50 0.071 0.13 0.26 0.00 304 257 1.18
25 Hidalgo 13 0.50 0.071 0.26 0.26 0.00 289 263 1.10
26 Hidalgo 14 0.35 0.047 0.25 0.13 0.00 255 127 2.00
27 Hidalgo 15 0.35 0.047 0.13 0.26 0.00 368 173 2.13
28 Hidalgo 16 0.35 0.047 0.25 0.25 0.00 362 169 2.14
29 Hidalgo 21 0.69 0.077 0.00 0.00 0.00 258 209 1.23
30 Hidalgo 22 0.69 0.077 0.00 0.00 0.00 222 176 1.26
154
31 Hidalgo 23 0.69 0.077 0.00 0.25 0.00 333 396 0.84
32 Hidalgo 24 0.69 0.077 0.25 0.00 0.00 232 208 1.12
33 Hidalgo 25 0.50 0.071 0.00 0.00 0.00 352 157 2.25
34 Hidalgo 26 0.50 0.071 0.00 0.00 0.00 262 135 1.95
35 Hidalgo 27 0.50 0.071 0.00 0.25 0.00 491 298 1.65
36 Hidalgo 28 0.50 0.071 0.25 0.00 0.00 258 154 1.67
37 Hidalgo 29 0.35 0.053 0.00 0.00 0.00 400 88 4.55
38 Hidalgo 30 0.35 0.053 0.00 0.00 0.00 356 77 4.61
39 Hidalgo 31 0.35 0.053 0.00 0.25 0.00 391 169 2.32
40 Hidalgo 32 0.35 0.053 0.25 0.00 0.00 344 88 3.90
41 Hirosawa 72 1.00 0.094 0.51 0.26 11.40 772 486 1.59
42 Hirosawa 73 1.00 0.094 0.51 0.26 9.40 775 509 1.52
43 Hirosawa 74 1.00 0.094 0.51 0.57 9.40 790 810 0.98
44 Hirosawa 75 1.00 0.094 0.51 0.57 14.30 812 761 1.07
45 Hirosawa 76 1.00 0.094 0.51 1.08 13.30 794 1245 0.64
46 Hirosawa 77 1.00 0.094 0.51 1.08 10.70 875 1270 0.69
47 Hirosawa 79 1.00 0.094 0.51 0.61 14.30 591 798 0.74
48 Hirosawa 82 2.00 0.189 0.40 0.57 9.40 318 788 0.40
49 Hirosawa 83 2.00 0.189 0.40 0.57 11.00 304 773 0.39
50 Rothe T10 1.50 0.100 0.71 0.51 0.00 89 303 0.29
51 Wiradinata Wall-1 0.58 0.050 0.59 0.26 0.00 532 160 3.33
52 Pilette Wall-4 0.58 0.050 0.59 0.80 0.00 401 359 1.12
53 Pilette Wall-5 0.58 0.050 1.07 1.20 0.00 544 530 1.03
54 M.-Doostdar Wall-7 0.82 0.050 0.59 0.80 0.00 375 623 0.60
55 M.-Doostdar Wall-8 1.09 0.067 0.51 0.80 0.00 225 602 0.37
56 Salonikios MSW3 1.60 0.083 0.28 0.28 7.00 173 458 0.38
57 Salonikios MSW6 1.60 0.083 0.57 0.57 0.00 187 736 0.25
58 Salonikios LSW1 1.10 0.083 0.57 0.57 0.00 260 440 0.59
59 Salonikios LSW2 1.10 0.083 0.28 0.28 0.00 185 275 0.67
60 Salonikios LSW3 1.10 0.083 0.28 0.28 7.00 251 281 0.90
61 Sheu SWN-1D 0.65 0.100 0.43 0.57 12.00 299 198 1.51
62 Sheu SWN-5D 0.90 0.100 0.43 0.57 12.00 245 299 0.82
63 Sheu SW-5 0.65 0.100 0.77 1.03 0.00 242 313 0.77
64 Sheu SW9 0.65 0.100 0.79 0.57 0.00 248 197 1.26
65 Sheu SW11 0.65 0.100 0.79 0.57 0.00 222 197 1.12
66 Sheu SW12 0.65 0.100 0.78 0.57 0.00 248 198 1.26
67 Sheu SW13 0.65 0.100 0.76 0.00 0.00 243 71 3.43
68 Sheu SW17 0.90 0.100 0.79 0.57 0.00 180 295 0.61
69 Sheu SW19 0.90 0.100 0.76 0.00 0.00 157 92 1.69
70 Synge Wall-1 0.57 0.033 0.81 1.61 0.00 773 936 0.83
71 Yoshizaki 165 1.08 0.075 0.22 0.23 0.00 98 95 1.03
72 Yoshizaki 166 1.08 0.075 0.73 0.82 0.00 142 206 0.69
73 Yoshizaki 167 1.08 0.075 0.44 0.41 0.00 130 129 1.01
74 Yoshizaki 168 1.08 0.075 0.73 0.82 0.00 161 206 0.78
75 Yoshizaki 169 1.08 0.075 1.17 1.17 0.00 173 271 0.64
76 Yoshizaki 170 0.72 0.050 0.24 0.23 0.00 152 91 1.67
77 Yoshizaki 171 0.72 0.050 0.78 0.82 0.00 210 195 1.08
78 Yoshizaki 172 0.72 0.050 0.44 0.41 0.00 205 123 1.67
79 Yoshizaki 173 0.72 0.050 0.78 0.82 0.00 243 195 1.24
80 Yoshizaki 174 0.72 0.050 1.17 1.17 0.00 254 257 0.99
81 Yoshizaki 175 0.54 0.037 0.22 0.23 0.00 178 87 2.06
82 Yoshizaki 176 0.54 0.037 0.80 0.82 0.00 322 185 1.74
83 Yoshizaki 177 0.54 0.037 0.37 0.41 0.00 279 116 2.40
84 Yoshizaki 178 0.54 0.037 0.80 0.82 0.00 391 185 2.11
85 Yoshizaki 179 0.54 0.037 1.17 1.17 0.00 399 243 1.64
86 Massone wp111-9 0.44 0.111 0.25 0.27 10.00 754 387 1.95
155
87 Massone wp111-10 0.44 0.111 0.25 0.27 10.00 821 398 2.06
88 Massone wp1105-8 0.44 0.111 0.25 0.27 5.00 649 400 1.62
89 Massone wp1105-7 0.44 0.111 0.25 0.27 5.00 683 400 1.71
90 Massone wp110-5 0.44 0.111 0.25 0.27 0.00 405 393 1.03
91 Massone wp110-6 0.44 0.111 0.25 0.27 0.00 325 397 0.82
92 Sugano 140-1 0.45 0.030 0.66 0.66 12.00 2354 1797 1.31
93 Sugano 141-2 0.45 0.030 0.66 0.66 23.00 2942 1794 1.64
94 Sugano 142-3 0.45 0.030 0.66 0.66 16.00 3138 1824 1.72
95 Sugano 143-4 0.45 0.030 0.33 0.33 10.00 1814 1428 1.27
96 Sugano 144-5 0.45 0.030 0.33 0.33 10.00 1912 1438 1.33
97 Sugano 145-6 0.45 0.030 0.69 0.66 11.00 2138 1425 1.50
98 Sugano 146-7 0.45 0.030 0.69 0.66 11.00 1981 1415 1.40
99 Sugano 147-8 0.45 0.030 0.77 0.74 12.00 2305 1801 1.28
100 Barda B1-1 0.50 0.052 0.50 0.50 0.00 1276 729 1.75
101 Barda B2-1 0.50 0.052 0.50 0.50 0.00 965 639 1.51
102 Barda B3-2 0.50 0.052 0.50 0.50 0.00 1112 736 1.51
103 Barda B6-4 0.50 0.052 0.25 0.50 0.00 872 457 1.91
104 Barda B7-5 0.25 0.052 0.50 0.50 0.00 1140 722 1.58
105 Barda B8-5 1.00 0.052 0.50 0.50 0.00 889 706 1.26
106 Ryo 31 0.63 0.035 0.17 0.18 0.00 608 344 1.77
107 Sugano 71 0.63 0.036 0.07 0.07 0.00 804 372 2.16
108 Aoyagi 150 0.56 0.059 0.58 0.62 0.00 1553 1451 1.07
109 Aoyagi 152 0.56 0.059 0.58 0.62 0.00 2308 1443 1.60
110 Aoyagi 148 0.56 0.029 0.71 0.76 0.00 931 789 1.18
111 Aoyagi 149 0.56 0.029 0.71 0.76 0.00 1029 823 1.25
112 Aoyagi 151 0.56 0.029 0.71 0.76 0.00 1495 813 1.84
156
In this appendix, the calculation scheme to compute the shear stress-strain relationship
1) Select a value of (smeared strain in principal 2-direction when Hsu/Zhu ratios are
2) Assume a value of (smeared shear strain in 2-1 coordinate of applied stress). A
3) Assume a value for (smeared strain in principal 1-direction when Hsu/Zhu ratios
principal tensile stress). An array of tensile strains is obtained after the procedure is
finished.
157
where:
= smeared strain in l-direction of longitudinal steel bars when Hsu/Zhu ratios are
steel bars.
= smeared strain in t-direction of transverse steel bars when Hsu/Zhu ratios are
steel bars.
= Hsu/Zhu ratio (ratio of resulting tensile strain to source compressive strain).
= Hsu/Zhu ratio (ratio of resulting compressive strain to source tensile strain).
= orientation angle, or angle between the direction of larger principal stress (1-
= smeared strain of steel bars that yield first, taking into account Hsu/Zhu ratios.
158
where:
159
where:
= Deviation angle from the angle of the applied principal stresses in the 1-2
coordinate and the angle of the principal concrete stresses in the r-d coordinate
160
where:
= smeared yield strain of the bilinear model idealization for mild steel formulated
= smeared steel stress greater than at which the material fails.
= smeared yield stress of the bilinear model idealization for mild steel formulated
Step 7.
where:
161
11) Check if . If not assume a different value for
13) Check if . If yes, stop. If not, select another value for and return to
An example of this procedure results is presented in Figure B.1, where the SMM curve is
obtained for the cross-section shown in Figure 3.5. The plot also illustrates the idealized multi-
linear curve to be used as part of the PINCHING4 material in OpenSEES. The shear-deformation
curve calculation scheme relies in finding equilibrium at every point which is difficult to achieve
for every combination of the variables , , and . Therefore, the model worked only in
limited cases. Additionally, Figure B.2 shows the idealized model used as part of the
PINCHING4 material for different aspect ratios, based on the shear-deformation curve obtained
162
Figure B.1: Shear-deformation curve from softened membrane model (SMM) for section RW2 and
idealized curve used with Pinching4 model in OpenSEES.
163