Académique Documents
Professionnel Documents
Culture Documents
DOI 10.1007/s10999-008-9070-6
Received: 22 November 2006 / Accepted: 3 March 2008 / Published online: 1 April 2008
Ó Springer Science+Business Media, B.V. 2008
Abstract The presence of masonry infill walls in analyzed with a simplified plane model, for each
RC buildings is very common. However, and even direction, and the existing infill panels were looked at
today, in the design of new buildings and in the in accordance with their dimensions and location.
assessment of existing ones, these infill walls are The earthquake action adopted in this study was
usually considered as non-structural elements and simulated through three series of artificially gener-
their influence in the structural response is generally ated earthquakes, for a medium/high seismic risk
ignored. For horizontal loading, infill panels can scenario in Europe.
drastically modify the response, attracting forces to
parts of the structure that have not been designed to Keywords Infill masonry Non-linear modeling
resist them. This paper presents an improved non- RC buildings Seismic behavior
linear numerical simulation model for the influence Structural vulnerability
of the masonry infill walls in the seismic behavior of
structure. The model is implemented in the PORANL
program. After the implementation and calibration of
the proposed masonry model, a series of non-linear 1 Influence of infill masonry walls
dynamic analyses of a building representative of on the structural seismic response
Modern Architecture in Europe were carried out. The
main objective was to investigate the behavior of this It is incorrect to assume that masonry infill panels are
type of building, and any weakness under seismic always beneficial to the structural response of build-
loading. The building geometry and the dimensions ings. The contributions of the infills to the building’s
of the RC elements and infill walls were set in the seismic response can be positive or negative, depend-
original project, and confirmed in the technical visits. ing on a series of phenomena and parameters such as,
The building under study has nine storeys and the for example, relative stiffness and strength between
structure is mainly composed of 12 plane frames the frames and the masonry.
oriented in the transversal direction. The building was In recent earthquakes, numerous buildings have
been severely damaged or have collapsed because of
the structural modifications to the basic structural
H. Rodrigues H. Varum (&) A. Costa system induced by the non-structural masonry parti-
Civil Engineering Department, University of Aveiro,
tions (Fig. 1). Even if they are relatively weak,
Campus Universitário de Santiago,
3810-193 Aveiro, Portugal masonry infill walls can drastically modify the
e-mail: hvarum@civil.ua.pt structural response.
123
124 H. Rodrigues et al.
Masonry infill panels can substantially increase the bi-diagonal-strut model. It takes into account the
global stiffness of the structure. Consequently, its interaction of the masonry panel’s behavior in the
natural period will decrease. Vulnerable behavior two directions. In order to represent a masonry panel
may also result as a function of the extent to which we considered four support strut-elements with rigid-
masonry walls are built, such is the case when the linear behavior; and a central element in which the non-
walls only extend to part of the storey-height (short- linear hysteretic behavior is concentrated (Fig. 2a).
columns), leaving a relatively short portion of the The non-linear behavior is characterized by a
columns exposed. Frequently, a column is shortened multi-linear envelop curve, defined by nine
by elements which have not been taken into account
in the global structural design such as, for example, (a)
window openings or staircase landing slabs (Varum
2003).
123
Non-linear masonry infill macro-model 125
parameters (Fig. 2b), representing: cracking, peak accordance with the material’s behavior (defined by
strength, stiffness decreasing after peak strength and the envelop curve and hysteretic parameters). The
residual strength, for each direction, which allows for hysteretic rules are briefly exemplified in Fig. 3.
the consideration of the non-symmetrical behavior. Loading rules: The loading stiffness depends on
The hysteretic rules calibrated for masonry models the maximum force and displacement value reached
are controlled by three additional parameters, in the previous cycle (Fmax and dmax). The loading
namely: a—stiffness degradation; b—‘‘pinching’’ begins at the point corresponding to null-force (dr)
effect; and, c—strength degradation. and its stiffness is defined by the Eq. 1:
The non-linear behavior of the central element is Fmax
characterized by hysteretic rules based on the Takeda Kr ¼ ð1Þ
dmax dr
model (Costa 1989), as illustrated in Fig. 3. This thus
permits the determination of response to the loading Unloading rules: Unloading takes place when a
history as a function of the material’s behavior load inversion occurs. The unloading stiffness
(defined by the envelop curve and hysteretic param- depends on the previously reached maximum dis-
eters), and represents mechanical effects such placement. Before the yielding-point has been
stiffness and strength degradation, pinching effect, reached, the unloading stiffness (Kd) will be equal
and internal cycles (Rodrigues et al. 2005b). to the initial stiffness (K0). If the maximum displace-
The monotonic behavior curve of each panel ment reached is larger than the yielding displacement,
depends on the panel’s dimensions, the dimensions but smaller than the cracking displacement (dcr), the
and positions of any openings, material properties unloading stiffness (Kd) will depend on the parameter
(bricks, mortar, and plaster), quality of the building a, and on the maximum displacement reached in that
work, and interface conditions between the panel and cycle, defined by:
surrounding RC elements. This behavior may be Fcr a Fy
determined from empirical expressions or experi- Kd ¼ K0 ð2Þ
K0 dcr þ a Fy
mental results (Zarnic and Gostic 1998).
If the maximum displacement reached is larger than
dcr, the unloading stiffness (Kd) will depend only on the
3 Hysteretic rules of the infill masonry model parameter a. The unloading stiffness is given by Eq. 3:
Fcr a Fy
The non-linear behavior of the central element is Kd ¼ aFy K0 ð3Þ
dcr K0 K0
characterized by hysteretic rules which allows the
determination of its response to cyclic loads, in
123
126 H. Rodrigues et al.
X
N
di ∆5 ∆4
c¼c ð5Þ
i¼1
df x'
y' ∆6
df ¼ dy l ð6Þ
nc
dir
e 1 PD2 ∆2 lp
PD1 ¼ n
1 ð7Þ
e 1 f ∆1
123
Non-linear masonry infill macro-model 127
7 13
12 13A
2
12A
8 14
23 6 3
− 1 +
18 19 Dy− Dc Dy
22 +
Dc D
21 20 5
17 11
15 4 9
10
16
characteristics of the cross-sections, reinforcement buildings feasible. In order to investigate the vulner-
and its location, and material properties. ability of these types of buildings, which are very
The non-linear behavior of the plastic hinge common in Lisbon, a building on the Avenue Infante
elements is controlled through a modified hysteretic Santo, which is representative of Modern Architec-
procedure, based on the Takeda model, as illustrated ture, was subject to analysis. The building was
in Fig. 6. This model, developed by Costa (1989), studied using the non-linear dynamic analysis pro-
represents the response evolution of the global RC gram PORANL, which allows the carrying out of a
element to seismic actions and accounts for such safety evaluation in accordance with the recently
mechanical behavior effects as stiffness and strength proposed standards (ATC-40 1996; and VISION-
degradation, pinching effect, slipping, internal cycles, 2000 1995).
etc. The 3rd Unity Type A lot in the Infante Santo
Avenue in Lisbon is a singular example of a Modern
Housing Project in Portugal (see Fig. 7). This build-
5 Case study ing was part of an Urban Plan launched by Alberto
José Pessoa (1919–1985), in 1947, when he joined
The development and calibration of refined numerical the ‘‘Urbanization Study on the Protection Area of
tools, together with assessment and design codes, the Palácio das Necessidades’’ at the Câmara
make the structural safety assessment of existing Municipal of Lisbon (CML). This project involved
123
128 H. Rodrigues et al.
expropriating and dividing up parcels of land for the probability of simultaneous occurrence of wind and
new Avenue Infante Santo, which completed the ring seismic actions in the same direction and at their
created by De Gröer’s Municipal Plan (1938–1948). maximum intensity’’. Dias concludes that ‘‘this
The block plan is rectangular with a width of building-type has superior safety conditions than the
11.10 m and a length of 47.40 m. The building has majority of Lisbon’s buildings’’.
the height of 8 habitation storeys plus the pilotis At a later stage, the importance of taking seismic
height at the ground floor. The ‘‘free plan’’ also actions into account in the design of structural
serves as a reference point as the house was designed elements was recognized. A new design project was
with flexibility of use in mind. The main structural delivered, according to an article by Maria Amélia
system (12 parallel plane frames) restricts the archi- Chaves and Bragão Farinha published in ‘‘Técnica’’
tecture. The layout of the units in the building block Magazine. Horizontal forces, proportional to the
(floor type with 6 duplex apartments) was defined in floor’s mass were considered in the frame nodes.
accordance with the structural system. The distance However, structural analysis was only carried out in
between the frame’s axes is 3.80 m. Each frame is the transversal direction. The structural engineer
supported by two columns and has one cantilever concluded that wind forces induce larger demands
beam on each side with a span of 2.80 m, resulting in than seismic loads, resulting in larger cross-sections.
thirteen modules. The engineer, Ramos Cruz, responsible for the
The structural plan is a good solution to the construction, developed a new design project for the
architectural objectives. ‘‘From the beginning of the 3rd parcel. He presented new calculations based on
studies, there was a ever-present concern to design a the original project, but he changed the original
resistant structure that was simple, elegant and structural floor to a reinforced concrete slab. He
economic. And it looks like the objective was fully proposes that, with this continuous rigid slab, rigid
achieved because since the beginning of the work on diaphragm behavior is guaranteed. Another modifi-
the project there has been no need to change the cation is the introduction of reinforced concrete walls
primitive structure’’. in the staircases on the ground floor, one of which
The structural plan for the housing block of the extends the full height of the building. However,
Unity Type A is comprised of 12 transversal rein- these RC elements were not detected in the technical
forced concrete frames, each formed by two columns visit to the building (Miranda et al. 2005).
and three beams at each storey, two of which are in
cantilever (Infante Santo residential building 1956). 5.1 Description of the subject structure
The structural design was initially made for
vertical loads only, without the columns’ bending The building geometry and dimensions of the RC
moment being taken into consideration. Later, new elements and infill walls were given in the original
designs were developed which took into account, in a project (1950–1956), and were confirmed in the
simplified manner, the horizontal loads corresponding technical visits (Miranda et al. 2005). As already
to the wind, using the Cross method for calculating described in a previous section, the building under
the distribution of the bending moments J. V. Dias study has nine storeys and the structure is mainly
(Miranda et al. 2005) did not take seismic action into composed of 12 plane frames oriented in the trans-
consideration in his design. He refers to ‘‘the low versal direction (direction Y, as shown in Fig. 8). The
(plan)
0,15
3,70
1,00 1,00
0,30 0,30
1,15
2,55
1,00
1,00
1,70
0,30 0,30
2,25 0,80 0,80 2,25
0,40
0,15
2,50
Y
3,55
3,55
123
Non-linear masonry infill macro-model 129
123
130 H. Rodrigues et al.
where the coefficients b and a are calculated such that transversal direction), which constitutes the first
1% damping ratio in the first two modes of vibration validation of the numerical model. Figure 11 shows
is achieved. [K] and [M] are the stiffness and mass the first natural mode for each direction.
matrices of the structure, respectively. From the analysis of the first vibration shape
modes in both directions it is clear that the seismic
5.3 Natural frequencies and modes structural response will induce soft-storey mechanism
behavior. This conclusion will be confirmed with the
A first validation of structural numerical models earthquake analysis results in the following sections
can be achieved by comparing the experimentally (Fig. 12).
measured and the analytically estimated natural
frequencies. 5.4 Earthquake input signals
Table 1 lists the first four natural frequencies
computed for each building direction. Three artificial earthquake input series were adopted
To validate the numerical building models, for the seismic vulnerability analysis of the building.
measurements were made in the two independent The first series (A) was artificially generated for a
directions of the first natural structural frequency, medium/high seismic risk scenario in Europe (Carv-
using a seismograph, and the ambient vibration. The alho et al. 1999), for various return periods (Table 2).
measured first frequency is indicated in brackets in The second and third series (B and C, respectively)
Table 1. were generated with a finite fault model for the
A good agreement was found between the exper- simulation of a probable earthquake in Lisbon
imentally measured frequencies (1.17 Hz for (Carvalho et al. 2004), calibrated with real seismic
longitudinal direction and 1.56 Hz for transversal actions measured in the region of Lisbon. The
direction—Miranda et al. 2005) and the frequencies earthquakes of the B and C series were scaled to
estimated with the structural numerical models the peak ground acceleration of series A, for each
(1.08 Hz for longitudinal direction and 1.75 Hz for return period. Table 2 details the peak ground
acceleration and the corresponding return period for
Table 1 Natural frequencies for directions X and Y each earthquake’s intensity.
123
Non-linear masonry infill macro-model 131
Aceleeration (cm/s2)
300
200
100
0
-100 0 2 4 6 8 10 12 14
-200
-300 Time (s)
Serie B
500
Aceleration (cm/s2)
300
100
-100
0 5 10 15 20 25
-300
-500 Time (s)
400 Serie C
300
Aceleration (cm/s2)
200
100
0
- 100 0 20 40 60 80 100
-200
-300
-400 Time (s)
Table 2 Reference earthquake action (peak ground accelera- first storey level. In fact, the absence of infill masonry
tion and corresponding return period) walls at the ground storey and the greater storey
Return period Peak ground height (5.60 m for the 1st storey and 3.00 m for the
(years) acceleration (9g) upper storeys), induces an important vertical struc-
tural irregularity, in terms of stiffness and strength.
73 0.091
For all the structural elements (columns and
475 0.222
beams), and for all the seismic input action levels,
975 0.294
the shear force demand assumes a value inferior to
2,000 0.380
the corresponding shear capacity, which confirms the
3,000 0.435
building’s safety in shear.
5,000 0.514
From the numerical analyses performed, it was
verified for the earthquake input motions that the
the first storey, while the upper storeys show very low infill masonry panels essentially reveal no damage. In
deformation levels. fact, due to the building structural system and its
Figures 13 and 14 illustrate, for the longitudinal behavior, on the one hand, and to the absence of
and transversal direction respectively, the numerical masonry panels at the ground floor, on the other, the
results in terms of envelop deformed shape, maxi- global deformation demand is concentrated at the
mum inter-storey drift, and maximum storey shear, ground floor, and, therefore, the majority of the infill
for each earthquake input motion of the A series (73, masonry panels present a linear behavior.
475, 975, 2000, 3000, 5000 years return period).
From the analysis of the results in terms of 5.5.1 Vulnerability curves
building envelop deformed shape and inter-storey
drift profile, for both directions, it can be concluded This section compares, for the three earthquake input
that the deformation demands are concentrated at the motions, the vulnerability curves in terms of
123
132 H. Rodrigues et al.
Storey
Storey
Storey
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
-50 150 350 0 2 4 6 0 1500 3000 4500
Displacement
p (mm)
( ) Storey maximum drift (%) Storey shear (kN)
5 5 5
Storey
Storey
Storey
4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
0 100 200 300 0 1 2 3 4 5 0 2500 5000 7500 10000
Displacement (mm) Storey maximum drift (%) Storey shear (kN)
maximum 1st storey drift, maximum 1st storey shear that, for the 1st storey, the maximum inter-storey
and maximum top displacement, for the longitudinal drift demand for the longitudinal direction is
and transversal directions. larger than for the transversal, thus confirming
Figures 15 and 16 compare the vulnerability that the former is the building’s most vulnerable
curves for the transversal and longitudinal direc- direction.
tions for the maximum 1st storey drift obtained Figures 17 and 18 show the vulnerability curves in
from the numerical analysis. The results show terms of maximum 1st storey shear force. Figures 19
123
Non-linear masonry infill macro-model 133
Fig. 16 1st storey drift versus peak acceleration (longitudinal Fig. 19 Top displacement versus peak acceleration (transver-
direction—X) sal direction—Y)
Fig. 17 1st storey shear versus peak acceleration (transversal Fig. 20 Top displacement versus peak acceleration (longitu-
direction—Y) dinal direction—X)
and 20 show the vulnerability curves obtained, in in the results relating to 1st storey drift and top
terms of maximum top displacement. Shear demand displacement. The results for certain earthquake
at the 1st storey does not increase for earthquake series, and for higher intensity levels, were not
input actions larger than the one corresponding to a included in the graphics, due to non-convergence of
return period of 475 years. These only induce the numerical analyses associated with the large
increasing deformation demands, as can be observed deformation demands.
123
134 H. Rodrigues et al.
Table 3 Basic
Fully operational Operational Life safe Near collapse
Performance Objectives for
buildings according to Earthquake Frequent (43-yrp)
VISION-2000 (1995) design level Occasional (72-yrp) X
Rare (475-yrp) X
Very rare X
(970–2000 yrp)
Table 4 Storey drift limits according to the (a) ATC-40 (1996), (b) VISION-2000 (1995)
Performance level
Immediate occupancy Damage control Life safety Structural stability
123
Non-linear masonry infill macro-model 135
Comparing the maximum storey drift demands with intervention in this typology can be instigated at the
the safety limits proposed in the ATC-40 and VISION- ground storey, usually without infill masonry walls,
2000 recommendations, the conclusion may be drawn for example with bracing systems and, eventually, in
that building safety is guaranteed in the transversal combination with energy dissipation devices. The
direction (Y), for the three earthquake input series building assessment and design of a retrofitting
considered. For the longitudinal direction (X), safety is solution should always be followed by an efficiency
guaranteed for series A and B, but not for series C. evaluation.
123