Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

EARTHQUAKE ENGINEERING & STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2014; 43:159–179


Published online 22 July 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/eqe.2335

A nonlinear macroelement model for the seismic analysis of


masonry buildings

Andrea Penna1,*,†, Sergio Lagomarsino2 and Alessandro Galasco1


1
Department of Civil Engineering and Architecture, University of Pavia, Via Ferrata 3, I-27100 Pavia, Italy
2
Department of Civil, Environmental and Chemical Engineering, University of Genoa, Via Montallegro 1, I-16145
Genova, Italy

SUMMARY
The macroelement technique for modelling the nonlinear response of masonry panels is particularly efficient
and suitable for the analysis of the seismic behaviour of complex walls and buildings. The paper presents a
macroelement model specifically developed for simulating the cyclic in-plane response of masonry walls,
with possible applications in nonlinear static and dynamic analysis of masonry structures. The model,
starting from a previously developed macroelement model, has been refined in the representation of
flexural–rocking and shear damage modes, and it is capable of fairly simulating the experimental response
of cyclic tests performed on masonry piers. By means of two internal degrees of freedom, the two-node
macroelement permits to represent the coupling of axial and flexural response as well as the interaction of
shear and flexural damage. Copyright © 2013 John Wiley & Sons, Ltd.

Received 24 September 2012; Revised 1 May 2013; Accepted 3 June 2013

KEY WORDS: macroelement; masonry; seismic analysis; nonlinear; equivalent frame

1. INTRODUCTION

Masonry constructions are often regarded as nonengineered structures whose structural behaviour can
be only approximated under restrictive and prudential hypotheses. This mainly derives from the
intrinsic complexity of the behaviour of masonry buildings and elements. Significant nonlinearities
due to the two-phase composite material affect the structural response and cracking may occur also
for low levels of stress. Moreover, the interaction between vertical and horizontal structures is often
complex and unclear. For these reasons, a great interest on seismic performance assessment of new
and existing masonry structures has more and more developed in the last 30 years.
The high seismic vulnerability of existing unreinforced masonry buildings, evident after past and recent
earthquakes (e.g., [1]) in many developed, emerging and developing world countries, has boosted the
research activity on modelling strategies suitable for static and dynamic analysis of masonry structures,
from single walls to complex masonry buildings. Several approaches are proposed, with different levels
of abstraction, hence introducing some interpretation of the phenomenon to simplify the problem for a
specific analysis aim and accepting some approximation in the results.
Some of those approaches try to model the problem without any simplification; this is the case of
micromodelling techniques (e.g., [2, 3]) in which a refined discretisation of units, mortar and
interfaces by nonlinear finite elements is proposed. Several authors (e.g., [4–6]) proposed simplified
micromodels in which only unit and joint elements are present. Such an approach would in principle

*Correspondence to: Andrea Penna, Department of Civil Engineering and Architecture, University of Pavia, Via Ferrata
3, I-27100 Pavia, Italy.

E-mail: andrea.penna@unipv.it

Copyright © 2013 John Wiley & Sons, Ltd.


160 A. PENNA, S. LAGOMARSINO AND A. GALASCO

allow to model and analyse any masonry structure, but there are actually several considerations that
significantly limit the field of applicability of these models to small substructures, typically
bidimensional.
Indeed, the high computational load and the irregular geometry of both in-plane layout and
thickness composition (e.g., stone masonry) make often preferable the adoption of equivalent
continuum models with an equivalent homogenized material derived from the mechanical properties
of the components. Homogenised material models usually regarded as macromodels have been
proposed by several authors (e.g., [7–9]). Still, the problem of a significant computational burden
often limits the applicability of these techniques to the analysis of small structures, typically under
static monotonic loadings.
Depending on size and detail of the models, discrete and/or rigid element approaches [10–12] can be
regarded as possible alternatives to the previously mentioned finite element models. Even in this case,
the increasing dimension of the modelled structures causes a demand in computation time, which often
cannot be matched without reducing the level of detailing with a coarser mesh.
All the aforementioned modelling strategies can be profitably applied in research and/or in the case
of special structures, but can be hardly considered for the analysis of current masonry buildings. On the
contrary, the modelling and analysis methods, which can be considered as fully suitable for common
engineering practice, are typically on the basis of strong simplifications aiming at reproducing some
aspects of the response.
Simplified methods, based on limit analysis [13–15], equivalent truss models [16] and storey
mechanisms [17–19] have been developed and also used by practitioners. They allow to obtain the
approximate evaluation of some aspects of the seismic response and/or they can be adopted under
restrictive hypotheses only.
An intermediate approach, which permits to obtain a reasonable compromise between accuracy of
results and computational effort, consists in the so-called macroelement modelling strategy, typically
implemented in equivalent-frame models of walls. A single macroelement is used to model the
behaviour of a masonry structural member (pier or spandrel beam). Assembling macroelements, it is
possible to model the in-plane behaviour of entire masonry walls, even with irregular distributions
of openings, by means of the equivalent frame technique (e.g., [20–23]).
Magenes et al. [24] report a rather complete collection of different models for frame-type and
macroelement modelling. In addition, models implemented in [25] and [26] should be also mentioned.
The availability of an effective nonlinear model for the seismic analysis of masonry structural
elements would represent a crucial step towards a proper assessment of new and existing masonry
buildings. The aim of this paper is to present an innovative macroelement model specifically
developed for such a scope.

2. MACROELEMENT MODEL REPRESENTATIVE OF A MASONRY PANEL

As illustrated in Galasco et al. [27], the strategy presented for the three-dimensional analysis of entire
masonry buildings is based on the in-plane equivalent frame modelling of walls, whose out-of-plane
contribution to strength and stiffness is neglected. Hence, the macroelement model representative of
piers and spandrel beams to be assembled to form the equivalent frame can also be a two-
dimensional model capable of simulating the in-plane response of masonry panels. In particular,
because the aim of this work is to provide a nonlinear model for seismic analysis of masonry
structures, the macroelement model has to fulfil the following requirements:
1. It has to adequately represent elastic properties of the masonry panel;
2. It has to properly evaluate lateral strength associated with the main failure modes (shear and
flexure-rocking);
3. It has to provide a physically consistent cyclic response when subjected to lateral loading.
As presented in Figure 1, it is important that the macroelement is able to simulate both shear and
flexural failure modes, including the effect on shear failure modes of flexural deformation with
possible cross section partialisation.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 161

Figure 1. Main failure modes of masonry panel subjected to lateral in-plane loading (mainly uncompressed
portion of the panel coloured in grey): (a) flexure-rocking with possible toe-crushing, (b) shear with sliding
along mortar bed-joints and (c) shear with diagonal cracking through units and mortar.

Among other macroelement models, the one presented by Gambarotta and Lagomarsino [28] with
its mechanics based formulation has proved the capability of complying to some extent with the
aforementioned requisites when used for the simulation of experimental tests [27, 29]. The
mechanics based formulation makes such a model a preferable starting point for cyclic and dynamic
analyses in which the panel response is not governed by predetermined values of strength and
stiffness, but it depends instead on mechanical properties, their possible deterioration, actual
boundary conditions and applied forces, which may vary in each step of the analysis.
The macroelement model proposed in [28, 29] is a two-node element capable to represent the in-
plane cyclic behaviour of a whole masonry panel. On the basis of its kinematical structure with two
additional internal degrees of freedom, it allows to describe both the shear behaviour and the
coupled axial–flexural one at the two nodes.
This macroelement model includes a nonlinear shear stress–strain cyclic relation that is derived by
the macroscopic integration of the Gambarotta and Lagomarsino [30] continuum model for masonry
and a nonlinear description of the coupled relation between flexural and axial degrees of freedom,
which allows to explicitly evaluate the cracking effects on rocking motion. The main advantages of
such macroelement model are the completely mechanical description and the capability of
representing the cyclic shear and flexural response of masonry panels.
Further studies [31, 32] have modified some aspects of the original formulation, with a relevant
improvement of the capabilities. The macroelement model presented in this paper includes, in
particular, a nonlinear degrading model for rocking damage, which permits to keep into account the
effect of limited compressive strength (toe-crushing).
Figure 2 illustrates the basic ideas of the macroelement formulation. The panel can be ideally
subdivided into three parts: a central body where only shear deformation can occur and two

Figure 2. Kinematics of the macroelement.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
162 A. PENNA, S. LAGOMARSINO AND A. GALASCO

interfaces, where the external degrees of freedom are placed, which can have relative axial
displacements and rotations with respect to those of the extremities of the central body.
The two interfaces can be considered as infinitely rigid in shear and with a negligible thickness.
Their axial deformations are due to distributed system of zero-length springs.
These assumptions simplify the macroelement kinematics and compatibility relations allow to
obtain a reduction of the actual degrees of freedom of the model. Because the central part is
considered as a rigid body with only shear deformation capability, under small displacement
hypotheses, the axial displacements and rotations of the ends can be considered equal to the centroid
ones (we, φe), whereas the transversal displacements on the central body ends must be equal to the
corresponding nodal displacements (ui, uj).
Therefore, the macroelement kinematics can be described by means of eight degrees of freedom, six
nodal displacement components (ui wi, φi, uj wj, φj) and two internal components (we, φe).
No distributed transversal actions are considered and so the internal shear force is constant along the
element axis (Vi = Vj = V).
The shear failure mode has been modelled through a constitutive law depending on three
displacement components (ui, uj, φe), the stress condition, evaluated considering an uniform axial
force, and the plasticity model variables (the shear damage variable α and the inelastic sliding
displacement s). This shear model is considered to be macroscopically representative of both the
shear failure modes, diagonal cracking and sliding on bed-joints (Figure 1), with an equivalent
approach, and the damage associated to such mechanisms can be measured by means of the inelastic
sliding displacement s that occurs when the Coulomb frictional limit is exceeded. The application of
an equivalent Coulomb-type criterion to the shear resistance of a masonry wall requires the adoption
of equivalent, generally low, values of cohesion and friction coefficient, in particular when the entire
cross section of the element is considered. This is also illustrated in [33] from the regression on
experimental values and it can be theoretically justified according to the interpretation provided in [34].
The panel flexural-rocking failure mode is represented adopting an unilateral contact model. A no
tension model has been attributed to zero-length springs at the interfaces, with a bilinear degrading
constitutive model in compression.

3. AXIAL AND FLEXURAL BEHAVIOUR

The axial and flexural behaviour of the two extremity joints is studied separately. The static and
kinematic variables involved in joint model are the element forces N and M for the considered node
and the relative displacement components w and φ. The relations between such variables are directly
derived from the constitutive model. If the whole cross section is compressed, no cracking occurs
and the problem is uncoupled, with linear relations linking N and M with w and φ respectively,
(
N ¼ kltw
k ; (1)
M ¼ tl3 φ
12
where l is the wall length; t is the wall thickness; k ¼ 2 E=h is the spring axial stiffness for surface unit,
with E compressive Young modulus of masonry and h wall height (to reproduce the axial stiffness of
the panel); w is the relative axial displacement (w = wiwe, at node i and w = wewj at node j) and φ is
the relative rotation (φ = φiφe, at node I and φ = φeφj at node j) (Figure 3).
The cracking condition can be simply expressed in terms of kinematic variables (Figure 4),

2w
jφj > : (2)
l
In cracked condition, the axial force can be calculated as

1
N ¼  ktl′ d; (3)
2

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 163

Figure 3. Kinematic variables at node i interface.

Figure 4. Kinematic representation of node i interface in cracked conditions.

where d is the relative axial displacement at the compressed edge of the considered interface joint and l′
is the compressed length of the cross section.
Under the small displacement hypothesis, it is immediate to obtain

d ¼ jφjl′ (4)

and therefore the expression of |w|


 
′ l
jwj ¼ l  jφj; (5)
2

derived according to the similar triangles rule can be inverted to obtain the length of the compressed
part of the cross section, l′,

w l ljφj þ 2w
l′ ¼ þ ¼ : (6)
jφj 2 2jφj

Expression (6) can be substituted in (4) and (5) expressing the axial force as

kt
N¼ ðljφj þ 2wÞ2 : (7)
8jφj

Imposing the rotational equilibrium around the mid-point of the cross section the bending moment
expression is then derived as

kt ðljφj þ wÞ
M¼ ðljφj þ 2wÞ2 : (8)
24 φjφj

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
164 A. PENNA, S. LAGOMARSINO AND A. GALASCO

Equations (7) and (8) can also be rewritten separating the elastic contribution and the cracking
corrections

kt
N ¼ ktlw  ðjφjl þ 2wÞ2 H ðjφj þ 2w=lÞ
8jφj
; (9)
1 3 kt ðjφjl  wÞ 2
M ¼ ktl φ  ðjφjl þ 2wÞ H ðjφj þ 2w=lÞ
12 24 φjφj
where H is a Heaviside function that activates the inelastic terms of the equations if the cross section is
cracked.
It is evident that, in case of cracked cross section, axial force and bending moment are not uncoupled
and, therefore, it is possible to obtain the relation between w and φ for a given axial force N,
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
jφjl 2jφjN
w¼  ; (10)
2 kt
which is valid out of the uncracked range of φ values.
The cracking condition is given by ∂w
∂φ ¼ 0, with

sffiffiffiffiffiffiffiffiffiffiffiffi!
∂w φ l N
¼  ; (11)
∂φ jφj 2 2jφjkt

and so it can be expressed as a linear function of the axial compression force N,

2N
jφ j ¼ : (12)
ktl2
Figure 5 shows the interaction between vertical displacement and rotation for different values of axial
load.
For node i, for example, (9) can be written in matrix form as
2 3
  ktl 0  w  w   N  ðw ; w ; φ ; φ Þ 
Ni
¼4 ktl3 5
i e i i e i e
 
; (13)
Mi 0 φi  φe M i ðw i ; w e ; φ i ; φ e Þ
12
where N i and M i are the inelastic corrections to N and M due to the cross section cracking.

Figure 5. Interaction between w and ϕ for a sample wall subjected to different values of axial force (a is
constant).

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 165

They can be evaluated as

kt
N i ¼ ðjφjl þ 2ðwi  we ÞÞ2 H ðjφj þ 2ðwi  we Þ=lÞ
8jφj
: (14)
 kt ðjφi  φe jl  ðwi  we ÞÞ 2
Mi ¼ ðjφi  φe jl þ 2ðwi  we ÞÞ H ðjφi  φe j þ 2ðwi  we Þ=lÞ
24 ðφi  φe Þjφi  φe j

Similar constitutive equations can be derived for the interface joint at node j.
As discussed before, (14) are inelastic corrections to the nodal forces, which have been derived
considering the no tension behaviour typically assumed of masonry structures. The homogenised
masonry material can provide a tensile strength, which is usually much smaller than the compressive
one, and, in case of seismic excitation, the tensile strength is lost after the first load cycles.
Masonry compressive strength is usually high compared with the vertical stress because of static
vertical loads. Nevertheless, from experimental tests (e.g., [35–37]), it has been often observed that
in-plane rocking mechanisms are characterized by toe-crushing phenomena at the base of masonry
piers, causing the limitation of the ultimate bending moment and some stiffness degradation in the
further cycles.
To include such effects into the nonlinear model, a phenomenological bilinear constitutive model
with stiffness degradation has been assigned to the interface joint springs [31], as illustrated in
Figure 6.
If the displacement threshold δy is exceeded the compressive strength in the spring is limited at the
value fm and the spring stiffness k is degraded to the secant value at the maximum displacement
attained during the previous load history, δmax. A local compression displacement ductility can be
defined as

δ max
μf ¼ ; (15)
δy

with

δy ¼ f m =k; (16)

and the secant spring stiffness can be expressed as

k  ¼ k=μf : (17)

Figure 6. Constitutive model, with no tensile strength, limited compressive strength and compressive stiff-
ness degradation, for interface joint springs.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
166 A. PENNA, S. LAGOMARSINO AND A. GALASCO

Two damage variables for each cross section edge have been introduced to perform the nonlinear
correction associated with the toe-crushing phenomena and to update the compressive stiffness
associated with the bending–rocking behaviour: μc = dmax/δy, the ductility demand in the external
compressed spring, and ζ = (μc  1)δy /φl, the ratio between the length of the portion of the cross
section involved in the crushing nonlinearity (Figure 7).
In the following load steps, the nonlinear corrections to N and M have then to include the influence
of the crushing damage variables ζ and μc.
The degraded stiffness can be indeed expressed as a function of the maximum values of such
parameters:

 1
 μ 1
k ðx; μc ; ζ Þ ¼ k 1 þ c ½x þ lðζ  0:5Þ (18)
ζl

 
where x∈ ð0:5  ζ Þl; 2l is the position of the spring with respect to the mid-point of the cross section.
The axial compressive stress is linear with the relative displacement, in the compressed part of the
cross section where the value of δy has not been exceed, and it has a more complex trend in the
degraded part.

>
> l
>
< k ðw  φxÞ x∈  ; ð0:5  ζ Þl
2
σðxÞ ¼ 
(19)
>
> ζl l
>
: k ð w  φxÞ x∈ ð 0:5  ζ Þl;
ðμc  1Þ½x þ lðζ  0:5Þ þ ζ l 2

Nevertheless, a linear approximation of the stiffness degradation does not induce significant
differences both in the stress path (Figure 8) and in the nonlinear corrections of the nodal forces.
Based on this last observation, it has been possible to derive the following corrections to N and M,
which must be applied to the values calculated considering the cracking corrections (20):

Figure 7. Displacement and compression stress distribution along the cross section in case of nonlinear com-
pressive behaviour (in case of monotonic loading).

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 167

Figure 8. Theoretical (continuous line) and linearised (dashed line) correction of the stress path.

 
** μc  1 l
N ðμc ; ζ ; w; φÞ ¼ k ζ lt w þ φ
2μ 2
 c  (20)
ζ 1
M ** ðμc ; ζ ; w; φÞ ¼   lN **
3 2

These corrections, as shown in the following, are valid both in case of stress reduction due to the
stiffness degradation effect and in case of new springs reaching the strength limit.
For each load step, four combinations of previous (maxima) and currently calculated damage
variables, μc′ and ζ ′, are possible at each corner of the macroelement (Figure 9):
a. μc ′ ≤ μc , ζ ′ ≤ ζ ;
b. μc ′ ≤ μc , ζ ′ > ζ ;
c. μc′ > μc, ζ ′ ≤ ζ ;
d. μc′ > μc, ζ ′ > ζ .
In the (a) condition, the damage variables have not to be updated because a new compressive
nonlinearity condition has not been reached: the damaged region has not been extended, and the
ductility demand is equal or lower than the previous maximum value. In all the other cases, the
damage variables have to be updated (both in the (d) condition).
As represented in Figure 9 for condition (b) and (c), although the linearization of the stress path in
the damaged area does not follow the possible discontinuities, it however provides a good
approximation of the theoretical trend.
Figure 10(a) shows the moment–rotation curve for the joint model with limited compressive
strength, and Figure 10(b) shows the influence of masonry compressive strength on the curve. As it
can be noted, the typically used assessment/design value for the ultimate bending moment for
masonry piers (e.g., [38]),

 
N N
Mu ¼ l ; (21)
2 κf m t

represents, for κ = 1 (values of κ = 0.8–0.9 are typically adopted for assessment/design purpose), an
upper bound for the curves, which tend to this value for high values of φ.
The compressive damage model, moreover, influences the interaction between axial displacement
and rotation. For monotonic loading, the relation is given by

N jφj 1 
w¼  δy  jφjl ; (22)
ktδy 2

2ðδy þwÞ δ2y kt


which is valid for jφj > l ¼ 2N .

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
168 A. PENNA, S. LAGOMARSINO AND A. GALASCO

Figure 9. Stress and displacement distribution along the element cross section for different combinations of
damage variables.

Figure 10. Moment–rotation curves for the joint model with limited compressive strength in case of constant
axial force: (a) identification of the flexural cracking and to toe-crushing conditions and horizontal lines in-
dicating the maximum strength given by (21) with κ = 1 and (b) effect of increasing masonry compressive
strength.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 169

Figure 11. Interaction between vertical displacement and rotation for constant axial load. Coupling due to
the no tension behaviour with (continuous line) and without (dotted line) cyclic toe-crushing damage.

The cyclic interaction between w and φ, as shown in Figure 11, is strongly influenced by the effect
of stiffness degradation with a limitation in the variation of w and a clear dependency of the current
value of w on the previous displacement history.

4. SHEAR DAMAGE MODEL

In case of masonry piers, horizontal mortar joints are considered as possible sliding surfaces where a
criterion, based on the two contributions of friction and cohesion between the elements, provides
shear strength to the central body of the macroelement model.
The central body kinematics (Figure 12), where shear deformability and damage phenomena are
concentrated, is described only by a relative transversal displacement, u, which can include an
elastic component, uel, and a plastic sliding one, s,

u ¼ uel þ s: (23)

If an incremental monotonic shear force is applied, after the elastic range, in which the constitutive
equation
Glt
V¼ u (24)
h

Figure 12. Panel shear deformation and plastic sliding.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
170 A. PENNA, S. LAGOMARSINO AND A. GALASCO

provides the correlation between the shear force, V, and the relative transversal displacement, u, as a
function of the elastic shear modulus of masonry and the geometrical characteristics of the wall, a
plastic displacement component s (sliding) occurs when |V | > Vf, with

V f ¼ μN > 0; (25)

maximum static frictional strength. μ is the equivalent friction coefficient.


Then, the cohesion provides additional shear strength till to the maximum value

V ¼ V f þ V c; (26)

where Vc is the shear force due the material cohesion.


The force–displacement diagram in Figure 13 allows to identify the different force and displacement
components.
To describe the shear behaviour in the friction sliding range, an additional damage variable, α, is
considered. The value of α is zero until no damage occurs to the panel and it can only increase,
reaching α = 1 by the peak shear strength of the panel and α > 1 in the post-peak softening branch.
The cohesive force Vc is a function of the sliding displacement s. The adopted function is derived
from fracture mechanics considerations for brittle materials, already implemented in the Gambarotta
and Lagomarsino [30] constitutive model for mortar joints

V c ¼ kðαÞs; (27)

where

lt
k ðα Þ ¼ ; (28)
ct hα

is a stiffness parameter depending on the nonlinear shear deformability parameter, ct, and the length of
the cross section at the centre of the panel.
It is then possible to write s as

ct hα 
s¼ V  Vf : (29)
lt

Figure 13. Shear force versus displacement: identification of shear and displacement components.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 171

The internal friction force has to comply with the sliding limit condition (N ≤ 0, as no axial tensile
strength is assumed for the macroelement)


Glt slt
φs ¼ jV  V c j  V f ¼ ðu  sÞ   μN≤0: (30)
h ct hα

A displacement increment beyond this limit condition would be necessarily an increment of the
inelastic sliding displacement s.
A second limit condition is given by the cohesion damage model, because the exceeding of the
failure condition may cause an increment of the material degradation if a previous higher damage
level, in terms of α parameter, was not reached before.
From (28), it is possible to derive the energy dissipation due to the material inelastic deformations

1 lt 2
ED ¼ s ; (31)
2 ct hα2

whereas the maximum available energy dissipation consistent with the damage level is described by
the toughness function R(α):

8
>
< 1 ct hV c; max 2 α ¼ 1 f vo 2 ct hltα α≤1
RðαÞ ¼ 2 lt 2
; (32)
>
: f ct hltα
1 β
2
α>1
2 vo

where fvo is the masonry shear strength without compression (cohesion), Vc,max = fvolt is maximum
shear capacity due to material cohesion and β is a model parameter governing the slope of the
softening branch (as evident from Figure 14 and from left part of Figure 15).
Therefore, the limit condition, which also must not be exceeded, associated with the damaged joints
is given by

1 lt 2
ϕ d ¼ E D  R ðα Þ ¼ s  RðαÞ≤0: (33)
2 ct hα2

Because the restraint functions ϕ s and ϕ d cannot be positive, any combination of parameters inducing
ϕ d > 0 and/or ϕ s > 0, provides residuals Δϕ that need to be compensated by increments of the damage
variables α and s, therefore modifying the associated limit domain surfaces.

Figure 14. Toughness function R(α).

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
172 A. PENNA, S. LAGOMARSINO AND A. GALASCO

Figure 15. Role of parameters β (left) and ct (right) on the shape of the nonlinear shear model.

By means of an iterative procedure, the damage variables have to be updated to guarantee that

(
Δϕ d þ dϕ̇d ¼ 0
; (34)
Δϕ s þ dϕ̇s ¼ 0

with

8  
>
> ∂ϕ d ∂ϕ d lts dRðαÞ lts
>
< dϕ̇d ¼ ∂α α̇ þ ∂λ λ̇ ¼  c hα3 þ dα α̇ þ λ̇
t ct hα2
  ; (35)
>
> ∂ϕ s ∂ϕ s lts Glt lt
>
: dϕ̇s ¼ α̇ þ λ̇ ¼ α̇  þ λ̇
∂α ∂λ ct hα2 h ct hα

where α̇ and λ̇ ¼ sign V f ṡ are the increments to the previous damage variables.
In matrix form, it can be written

2 3
( ) lts dRðαÞ lts ( ) 
þ  
α̇ 6 ct hα3 dα ct hα2 7 α̇ Δϕ d
M 6
¼4 7 ¼ : (36)
λ̇ lts Glt lt 5 λ̇ Δϕ s
 þ
ct hα2 h ct hα

Hence, the updated values of the damage variables, α′ and s′, can be evaluated as

( )   ( )   " #  
α′ α α̇ α 10 1
Δϕ d
¼ þ ¼ þ  M : (37)
s′ s ṡ λ 0 sign V f Δϕ s

The correction to the shear force in the macroelement can be then derived

Glt Glt  Glt


V i ¼ V j ¼ V ¼ V el  V  ¼ ðu  sÞ ¼ ui  uj þ φe h  s: (38)
h h h

The effect of the parameters β and ct on the nonlinear shear model is represented in Figures 15 and
16, on the monotonic and cyclic response, respectively. As it can be noted from right parts of
Figures 15 and 16, the amplitude of the inelastic displacement component depending on the
cohesive behaviour is proportional to ct. In particular, as it can be easily noticed from displacements
at maximum strength, the inelastic component s is proportional to the Gct product.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 173

Figure 16. Influence of model parameters β (left) and ct (right) on the cyclic shear response.

5. MACROELEMENT IMPLEMENTATION

5.1. Constitutive equations


In the previous sections, the constitutive model relations between the eight kinematic variables and the
six nodal forces (Ni, Vi, Mi, Ni, Vj, Mj) have been derived. Internal equilibrium equations provide the
forces acting of the additional degrees of freedom

Ne ¼ Nj  Ni
(39)
Me ¼ Mj  Mi þ V ih

In matrix form, subdividing the elastic and inelastic terms, the macroelement constitutive equations
can then be written as
2 3
klt 0 0 0 0 0 klt 0 8 9
8 9 8 9 ** >
7> wi > >

> Ni > 6 Glt Glt >
> N þ N >
>
>
> > 6
> 6 0  Glt 7 > > > i i
>
> 0 0 0 0 7> > > >
Vi > > > > >

>
> >
> 6 h h 7>
> u >
> >
> V >
>
>
> >
> 6 7>
> >
i >
> > > i
>
>
>
> >
> 6 0
1
ktl3
1
 ktl 7>3 > > >
>  ** >
>
>
> Mi >>
0 0 0 0 0
7>
> φi >> >
> M þ M i >>
>
> > 6 12 12 > > > i
>
<N > = 66 0 0 0 klt 0 0 klt 0 7> =
<
7 wj
> > <  ** > =
N þ N
¼6 7
j
6 Glt Glt 
7> u > > 
j j ;
>
> Vj >>  7> >
>
> >
>
6 0
6
0 0 0 0 Glt > j>
7> > >
> >
>
Vj >
>
>
>
> >
>
h h
7>
> > > >
>
> M >
j > 6
6 0 >
> φj >>
>
>
>
> ** >
>
>
 ktl3 7

>
> >
> 6 0 0 0 0
1
ktl3 0
1
>
>
7> > > >
> >
> M þ M >
>
>
> Ne >> 6 12 12 7> we > >
j j
>
>
>
> >
> >
> >
> >
>  >
>
: ; 64 klt 0 0 klt 0 0 2klt 0 7:
5 φ ; >
>
N e >
>
Me 1 1 1 3 e
:  ;
0 Glt  ktl3 0 Glt  ktl3 0 Glt þ ktl M e
12 12 6
(40)
where the nonlinear correction terms corresponding to cracking, toe-crushing and shear damage
conditions are described in (14), (20) and (38), respectively.

5.2. Element expiration at maximum drift


In case, based on experimental results on masonry structural elements, deformation limits associated
with a sudden loss in lateral strength are identified (near collapse conditions), they can be also
accounted for in the macroelement model. In fact, the model allows to limit the lateral displacement
capacity of masonry panels and complete walls by controlling several deformation quantities. In
particular, ultimate drift values, δs and δr, can be set for the shear-sliding deformation component
and the bending–rocking one, respectively.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
174 A. PENNA, S. LAGOMARSINO AND A. GALASCO

The ultimate drift conditions have been derived as functions of the displacement components

8 u  u
> j i
< h þ φe ≥δs
>
: (41)
> φj þ φi
:
> þ φe ≥δr
2

If one of the (41) is reached, then the macroelement lateral stiffness is set to zero and a force
redistribution is then automatically operated among the still resisting elements of the equivalent
frame model. The macroelement, in the further load steps, is then transformed into a link element
maintaining its capacity of carrying vertical loads only.

6. EXAMPLES OF APPLICATION TO THE SIMULATION OF EXPERIMENTAL TESTS

To show the ability of the presented macroelement model to simulate the experimental behaviour of
real masonry piers, three examples are presented in the following sections. In particular, the
macroelement model is used to simulate the results of in-plane quasistatic cyclic tests performed on
full-scale masonry piers, one exhibiting a clear shear failure with stiffness degradation and strength
deterioration [36] and two other ones [36][39] with a lateral response dominated by the flexural
behaviour.
Other applications of the macroelement model showing its suitability for the simulation of the
experimental response of in-plane cyclic quasistatic tests on masonry piers are also reported in some
recent literature works, referring to different masonry typologies from regular autoclaved aerated
concrete block masonry with thin-layer mortar joints [40] to irregular tuff stone masonry [41].

6.1. Shear-dominated ‘squat’ panel tested by Anthoine et al. [36]


As a first example of application, a rather squat masonry panel (1 m long, 1.35 m high and 0.25 m
thick) made of solid clay bricks tested at European Laboratory for Structural Assessment laboratory
(Joint Research Center) in Ispra is considered.
The wall was tested under controlled boundary conditions, with fixed base, fixed rotation of the top
beam (double-bending condition) and an applied vertical force equal to 150 kN kept constant during
the whole test. Horizontal displacements where imposed to the top beam, with cycles of increasing
amplitude, which were repeated two or three times. A clear shear behaviour was observed with X-
shaped shear cracks.
Figure 17 shows the experimental results compared with those of the numerical simulation
performed imposing the same loading history to a macroelement model with mechanical properties
listed in Table I.
As it can be noted from the left part of Figure 17, the calibrated macroelement model is able to
capture the envelope curve of the experimental cyclic force–displacement response, although the
envelope of the numerical curve is obviously symmetric, and in this case, the experimental one is not.
The right part of Figure 17 reports a comparison of the hysteretic energy dissipation for each cycle
showing a fair agreement between the experimental and numerical trend, with a tendency of the
numerical model to slightly underestimate the experimental value for lower displacement values.
Also, the numerical model is able to simulate the loss of energy dissipation capacity occurring in the
first repetition of the cycles at the same peak displacement.
As it was mentioned before, in this case, the shear model parameters were obtained by direct
calibration against the experimental results.
Some of the model parameters can be directly evaluated on the basis of in situ or laboratory testing
on masonry panels. This is the case of some of the mechanical parameters, that is, Young’s modulus in
compression, E, shear modulus, G, density, ρ, and compressive strength, fm.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 175

Figure 17. Comparison of experimental (black) and numerical simulation (grey) results for the shear-
dominated response of a masonry wall [36]: cyclic force–displacement curve (left) and histograms
representing the hysteretic energy dissipation in each cycle (identified by average drift level and number
of repetition).

Table I. Mechanical properties and model parameters of the calibrated macroelement.


Nonlinear
shear
Young’s Shear Compressive Friction deformation Softening
modulus modulus Density strength Cohesion coefficient parameter parameter
E (MPa) G (MPa) ρ (kg/m3) fm (MPa) fvo (MPa) μ () Gct () β ()
4250 1000 1800 6.0 0.29 0.09 1.4 0.325

The nonlinear shear deformation parameter, ct, and the softening parameter, β, need instead to be
determined on the basis of experimental results. In case of a blind prediction of the response of a
masonry wall, if no specific experimental information is available, these shape parameters can be set
on the basis of previously calibrated values for similar masonry types. In most cases, however,
typical ranges are 1–4 and 0.1–0.4 for the product Gct and for β, respectively. Further systematic
calibration work should be performed to provide specific recommendations and typical values for
different masonry types.
The strength parameters of the macroscopic shear model, that is, the equivalent cohesion, fvo, and
the equivalent shear friction coefficient, μ, can be preliminarily set on the basis of a suitable strength
criterion selected as representative of the expected shear failure mode (e.g., diagonal tension failure).
A linear approximation of the shear strength criterion in a reasonable range of average compression
values (e.g., 0.1–1.0 MPa) or a first order Taylor expansion calculated at the design value of axial
compression can provide meaningful values for fvo and μ, able to provide a fair approximation of
the expected lateral strength. Magenes and Calvi [37], reported an interpretation of the lateral
strength of the considered specimen based on the application of the shear strength criterion proposed
by Mann and Müller [34], which accounts for the brick tensile strength, fbt, that is,

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f bt N
V res ¼ lt 1þ ; (42)
2:3ð1 þ αV Þ ltf bt

where (1 + αV) = (1 + M/Vl) = (1 + 0.5h/l) = 1.675 is a correction coefficient proposed in [37] to


encompass the effects of complex stress distribution, cracking and shear–flexure interaction. The
experimental value adopted in [37] for fbt was 1.22 MPa.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
176 A. PENNA, S. LAGOMARSINO AND A. GALASCO

If the first order Taylor expansion is applied to (42) at the value of the applied vertical force
No = 150 kN, it yields to the following values of fvo and μ:
8 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
>
> dV res ðN o Þ 1 f bt
>
> μ¼ ¼ ¼ 0:106
>
< dN 4:6ð1 þ αV Þ f bt þ N o =lt
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : (43)
>
> ð Þ
>
> V res N o N o f N o N o
>
: f vo ¼ μ ¼ bt
1þ  μ ¼ 0:323 MPa
lt lt 2:3ð1 þ αV Þ ltf bt lt

The values of fvo and μ obtained from (43) are close to those reported in Table I determined by
calibration.

6.2. Flexure-dominated ‘slender’ panels tested by Anthoine et al. [36] and Magenes et al. [39]
The experimental campaign carried out by Anthoine et al. [36] also included a taller (2.0 m high) panel
realized with the same masonry type and tested under the same boundary conditions and with the same
vertical load applied on top. The wall in-plane behaviour was dominated by a bending response with
flexural cracking at the edges.
Figure 18 reports the comparison of the experimental and numerical results in terms of force–
displacement cyclic curves. The parameters adopted for the numerical model are those reported in
Table I. It can be noted that the macroelement model is capable of simulating the experimental
envelope curve quite accurately. In this case, because the applied compression was relatively low in
comparison with the masonry compressive strength, the lateral response predicted by the model is
not significantly affected by toe-crushing effects.
Another example of slender pier with a clear flexural behaviour exhibited during in-plane cyclic
shear-compression test is considered in the following. The specimen was a full-scale wall with a
higher in-plane aspect ratio with respect to the previous example, that is, it was 1.25 m long, 2.5 m
high and 0.175 m thick. The wall was made of calcium silicate blocks and thin-layer mortar, with
both bed-joints and head-joints filled. The wall was tested at the EUCENTRE laboratory within the
framework of the ESECMaSE Project [39].
Also in this case, the wall was tested under controlled double-bending boundary conditions, with an
applied vertical force equal to 218.75 kN (1 MPa) constant during the test. Horizontal displacements of
increasing amplitude where imposed to the top beam, with three repetitions of each cycle. A pure
flexural behaviour was observed with significant partialization of the base and the top of the wall.
Figure 19 reports the direct comparison of the force–displacement curves obtained from the test
results and the numerical simulation. Mechanical properties and parameters of the macroelement

Figure 18. Comparison of the force–displacement curves obtained from in-plane cyclic test [36] on the slen-
der panel (black) and numerical simulation (grey).

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 177

Figure 19. Comparison of the force–displacement curves obtained from in-plane cyclic test (black dashed
line) on a calcium silicate masonry wall [39] and numerical simulation in case of infinite (light grey) and
limited compressive strength (dark grey).

Table II. Mechanical properties and model parameters of the calibrated macroelement.
Nonlinear
shear
Young’s Shear Compressive Friction deformation Softening
modulus modulus Density strength Cohesion coefficient parameter parameter
E (MPa) G (MPa) ρ (kg/m3) fm (MPa) fvo (MPa) μ () Gct () β ()

18000 6000 1900 7.5 0.3 0.2 2 0.4

model are listed in Table II. It should be noted that the parameters associated with the shear model
cannot be really calibrated because they did not affect the lateral response governed by bending–
rocking behaviour.
From the observation of Figure 19, it is possible to verify that also in this case the macroelement
model allows simulating the cyclic bending–rocking response of masonry piers. In particular, it is
possible to accurately reproduce the nonlinear envelope curve obtained from the test. In this case,
the role of masonry compressive strength in limiting the lateral strength of the model, with results
closer the experimental ones, can be clearly observed from Figure 19.
In both the examples presented in this section, the model underestimates the experimental hysteretic
energy dissipation, due to limited local inelasticity that cannot be detected by this model, which is in
any case almost negligible for flexural behaviour with respect to shear damage.

7. CONCLUSIONS

To study the seismic behaviour of unreinforced masonry structures, nonlinear modelling capabilities
are a mandatory requirement. Among other modelling strategies for masonry buildings, the
equivalent frame technique is an optimal compromise between accuracy and computational
efficiency that requires the definition of appropriate two-node nonlinear elements representative of
structural members.
The presented two-node macroelement is a mechanics based model allowing to represent the main
features of the lateral response of masonry piers, that is, the bending–rocking behaviour, the shear
behaviour and their mutual interaction. By means of the two internal degrees of freedom, the shear
response is concentrated in the central body and it is uncoupled from the axial–flexural response,
concentrated in the top and bottom interfaces.
The model is particularly suitable for representing the nonlinear cyclic behaviour of masonry
members subjected to static and/or dynamic lateral loads, only starting from boundary conditions,

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
178 A. PENNA, S. LAGOMARSINO AND A. GALASCO

axial force acting on the element and a limited number of model parameters (8). It is capable of
simulating the damage evolution associated with both shear and rocking behaviour with toe-crushing
and allows identifying damage conditions by means of specific indicators included in the model.
The simulation of experimental tests showed a fair agreement between test and numerical analysis
results, even in terms of hysteretic energy dissipation associated with shear damage. This
achievement is particularly important for the application of the macroelement model in time-history
analyses oriented at the simulation of the seismic behaviour of masonry structures.
The macroelement model can be used for meshing equivalent frames representative of entire
masonry walls. As already implemented in the TREMURI programme, the equivalent frame models
can be obtained assembling two-node pier and spandrel elements modelled with the presented
macroelement. To account for the typical orthotropic behaviour of masonry walls, appropriate values
for the mechanical properties shall be adopted for modelling spandrel elements. In particular,
Young’s modulus in compression and compressive strength must correspond to those associated
with masonry behaviour under in-plane horizontal compression (parallel to bed-joints).

ACKNOWLEDGEMENTS
Professor Guido Magenes of the University of Pavia is gratefully acknowledged for providing the experi-
mental data used for the comparison with numerical simulation in Section 6. The clarity of manuscript
benefited from the valuable contribution of two anonymous reviewers.

REFERENCES
1. Rota M, Penna A, Strobbia C, Magenes G. Typological seismic risk maps for Italy. Earthquake Spectra 2011;
27(3):907–926.
2. Alpa G, Monetto I. Microstructural model for dry block masonry walls with in-plane loading. Journal of the Mechanics
and Physics of Solid 1994; 47(7):1159–1175.
3. Anthoine A, Pegon P. Numerical analysis and modelling of the damage and softening of brick masonry. In Numerical
Analysis and Modelling of Composite Materials, Ball J (ed.). Blackie Academic and Professional: New York, 1996;
152–184. ISBN: 0 7940239 6.
4. Lofti HR, Benson Shing P. Interface model applied to fracture of masonry structures. ASCE Journal of Structural
Engineering 1994; 120:63–80.
5. Gambarotta L, Lagomarsino S. Damage models for the seismic response of brick masonry shear walls. Part I: the
mortar joint model and its applications. Earthquake Engineering and Structural Dynamics 1997; 26:423–439.
6. Lourenço PB, Rots J. A multi-surface interface model for the analysis of masonry structures. ASCE Journal of
Engineering Mechanics 1997; 123(7):660–668.
7. Pegon P, Anthoine A. Numerical strategies for solving continuum damage problems with softening: application to
the homogenization of masonry. Computers and Structures 1997; 64(1-4):623–642.
8. Lourenco PB, Rots JG, Blaauwendraad J. Continuum model for masonry: parameter estimation and validation, ASCE
Journal of Structural Engineering 1998; 124(6):642–652.
9. Calderini C, Lagomarsino S. A continuum model for in-plane anisotropic inelastic behaviour of masonry. ASCE
Journal of Structural Engineering 2008; 134(2):209–220.
10. Lemos JV. Discrete element modeling of masonry structures. International Journal of Architectural Heritage:
Conservation, Analysis, and Restoration 2007; 1(2):190–213.
11. Casolo S, Peña F. Rigid element model for in-plane dynamics of masonry walls considering hysteretic behaviour and
damage. Earthquake Engineering and Structural Dynamics 2007; 36(8), 1029–1048.
12. Caliò I, Marletta M, Pantò B. A new discrete element model for the evaluation of the seismic behaviour of
unreinforced masonry buildings. Engineering Structures 2012; 40:327–338.
13. Abruzzese D, Como M, Lanni G. On the lateral strength of multistory masonry walls with openings and horizontal
reinforcing connections. Proceedings of the 10th World Conference on Earthquake Engineering, Madrid, Balkema,
Rotterdam, 1992; 4525–4530.
14. Giuffrè A (ed.) Safety and conservation of historical centers: the Ortigia case. Laterza: Bari, 1993 (in Italian).
15. Milani G, Lourenço PB, Tralli A. 3D homogenized limit analysis of masonry buildings under horizontal loads.
Engineering Structures 2007; 29:3134–3148.
16. Calderoni B, Marone P, Pagano M. Modelli per la verifica statica degli edifici in muratura in zona sismica,
Ingegneria Sismica 1987; 3:19–27 (in Italian).
17. Tomaževič M. The computer program POR. Report ZRMK, Ljubljana, 1978 (in Slovenian).
18. Braga F, Dolce M. A method for the analysis of antiseismic multi-storey masonry buildings. Proceedings of the 6th I.
B.Ma.C., Roma, 1982; 1088-1099 (in Italian).
19. Tomaževič M. Dynamic modelling of masonry buildings: storey mechanism model as a simple alternative. Earth-
quake Engineering and Structural Dynamics 1987; 15(6):731–749.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe
NONLINEAR MACROELEMENT MODEL FOR SEISMIC ANALYSIS OF MASONRY BUILDINGS 179

20. Magenes G, Della Fontana A. Simplified non-linear seismic analysis of masonry buildings. Proceedings of the British
Masonry Society 1998; 8:190–195.
21. Roca P, Molins C, Marì AR. Strength capacity of masonry wall structures by the equivalent frame method. ASCE
Journal of Structural Engineering 2005: 1601–1610.
22. Belmouden Y, Lestuzzi P. An equivalent frame model for seismic analysis of masonry and reinforced concrete buildings.
Construction and Building Materials 2009; 23(1):40–53.
23. Penelis GG. An efficient approach for pushover analysis of unreinforced masonry (URM) structures. Journal of
Earthquake Engineering 2006; 10(3):359–379.
24. Magenes G, Braggio C, Bolognini D. Metodi Semplificati per l’analisi Sismica non Lineare di Edifici in Muratura.
CNR-GNDT: Roma, 2001.
25. Chen S-Y, Moon FL, Yi T. A macroelement for the nonlinear analysis of in-plane unreinforced masonry piers,
Engineering Structures 2008; 30(8):2242–2252.
26. Grande E, Imbimbo M, Sacco E. A beam finite element for nonlinear analysis of masonry elements with or without
fiber-reinforced plastic (FRP) reinforcements. International Journal of Architectural Heritage 2011; 5(5-6):693–716.
27. Galasco A, Lagomarsino S, Penna A, Resemini S. Non-linear seismic analysis of masonry structures. Proceedings of
the 13th Word Conference on Earthquake Engineering (WCEE), Vancouver Canada, 2004.
28. Gambarotta L, Lagomarsino S. On the dynamic response of masonry panels. Proceedings of the National Conference
“Masonry Mechanics Between Theory and Practice”, Messina, Italy, 1996 (in Italian).
29. Brencich A, Lagomarsino S. A macro-elements dynamic model for masonry shear walls. In Proceedings of the
STRUMAS IV—4th Int. Symp. On Computer Methods in Structural Masonry. E&FN Spon: London, 1998; 67–75.
30. Gambarotta L, Lagomarsino S. Damage models for the seismic response of brick masonry shear walls. Part II: the
continuum model and its applications. Earthquake Engineering and Structural Dynamics 1997; 26:441–462.
31. Penna A. A macro-element procedure for the dynamic non-linear analysis of masonry buildings. Ph.D. Dissertation,
Politecnico di Milano, 2002 (in Italian).
32. Lagomarsino S, Galasco A, Penna A. Non linear macro-element dynamic analysis of masonry buildings. Proceedings of
the ECCOMAS COMPDYN Conference, Rethymno, Crete, Greece, 2007.
33. Tomaževič M. Earthquake-Resistant Design of Masonry Buildings. Imperial College Press: London, 1999.
34. Mann W, Müller H. Failure of shear-stressed masonry: an enlarged theory, tests and application to shear walls.
Proceedings of the British Ceramical Society 1982; 30:223–235.
35. Abrams DP. Strength and behaviour of unreinforced masonry elements. Proceedings of the 10th World Conference
on Earthquake Engineering, Madrid, 1992.
36. Anthoine A, Magonette G, Magenes G. Shear compression testing and analysis of brick masonry walls. Proceedings
of the 10th European Conference on Earthquake Engineering, Vienna, 1995.
37. Magenes G, Calvi GM. In-plane seismic response of brick masonry walls. Earthquake Engineering and Structural
Dynamics 1997; 26:1091–1112.
38. EN 1998-3. Eurocode 8: Design of Structures for Earthquake Resistance - Part 3: Assessment and Retrofitting of
Buildings. CEN: Bruxelles, 2005.
39. Magenes G, Morandi P, Penna A. In-plane cyclic tests of calcium silicate masonry walls. 14th International Brick
Block Masonry Conference, Sydney, Australia, 2008.
40. Costa AA, Penna A, Magenes G. Seismic performance of autoclaved aerated concrete (AAC) masonry: from exper-
imental testing of the in-plane capacity of walls to building response simulation. Journal of Earthquake Engineering
2011, 15(1):1–31.
41. Rota M, Penna A, Magenes G. A methodology for deriving analytical fragility curves for masonry buildings based on
stochastic nonlinear analyses, Engineering Structures 2010; 32(5):1312–1323.

Copyright © 2013 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2014; 43:159–179
DOI: 10.1002/eqe

You might also like