Вы находитесь на странице: 1из 20

Construction and Building Materials 41 (2013) 276295

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Pushover analysis of large scale unreinforced masonry structures by means


of a fully 2D non-linear model
A.H. Akhaveissy a,, G. Milani b
a
b

Department of Civil Engineering, Faculty Engineering, Razi University, P.O. Box 67149-67346, Kermanshah, Iran
Dipartimento di Ingegneria Strutturale (DIS), Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy

h i g h l i g h t s
" Macroscopic non-linear model for the pushover analysis of masonry walls.
" A DSC model with modied hierarchical single yield surface (HISS) is used.
" Two HISS yield surfaces for compressive and tensile behavior are utilized.
" Three large scale walls are studied and results are compared with literature.
" Good predictions of both ductility and peak loads are obtained.

a r t i c l e

i n f o

Article history:
Received 8 June 2012
Received in revised form 30 November 2012
Accepted 6 December 2012
Available online 11 January 2013
Keywords:
Masonry
2D pushover analyses
Macro-modeling
Disturbed state concept
Softening
Compression and tension strength

a b s t r a c t
A simple 2D model for the evaluation of the seismic performance of large scale unreinforced masonry
structures in-plane loaded is presented. The approach is fully two dimensional and allows performing
pushover analyses on large scale structures without the reduction of the walls to an equivalent frame,
competing with 1D codes. In the model, a macroscopic approach is used, where the so called disturbed
state concept (DSC) with modied hierarchical single yield surface (HISS) plasticity are used to characterize the constitutive behavior of masonry in both compression and tension. Two HISS yield surfaces for
compressive and tensile behavior are utilized. The DSC model allows for the characterization of nonassociative behavior through the use of disturbance, and it computes micro-cracking during deformation,
which eventually leads to fracture and failure. The DSC model is validated at both (1) specimen and (2)
structural level. At a structural level, three large scale masonry walls subjected to incremental horizontal
loads are analyzed and pushover curves obtained with the model proposed are compared to those
obtained by means of standard equivalent frames and existing literature models.
2012 Elsevier Ltd. All rights reserved.

1. Introduction
In several regions of high seismic hazard (e.g. South-Europe,
Middle East, South and Center-America) the traditional form of
construction is constituted by unreinforced masonry, which sometimes exhibits very poor mechanical properties, being frequently
built with irregular stones interconnected by poor mortar or unbonded joints. Even masonry structures built in regular texture
with clay bricks and mortar with intermediate mechanical properties, generally regarded as less vulnerable, have in some cases proven to be highly vulnerable to horizontal loads, as conrmed by the
collapses of many masonry buildings after the seismic event recently occurred in Northern Italy (Emilia Romagna, 20th and
29th May 2012). Considering all these issues, seismic assessment
Corresponding author. Tel.: +98 831 427 4535; fax: +98 831 428 3264.
E-mail address: Ahakhaveissy@razi.ac.ir (A.H. Akhaveissy).
0950-0618/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2012.12.006

of existing masonry structures is a task of large societal importance


in all these regions.
Modern structural approaches specialized for the safety assessment of existing structures tend estimating the seismic performance by means of pushover analyses [1,2] which are based on
the numerical evaluation of the ultimate lateral force capacity
and the displacement capacity of the building.
On the other hand, to propose a comprehensive numerical model able to give information on the non-linear behavior of such
buildings but taking at the same time into account properly
the actual texture of the walls seems at present a prohibitive
challenge.
In this framework, a large number of studies specialized in
pushover analyses of masonry structures have been conducted in
the recent past see e.g. [312]. As a rule, such models base essentially on the substitution of the 2D heterogeneous material with
either a ctitious frame or an assemblage of macro-elements, so

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

that they all belong to the so-called equivalent frame or more


properly mono-dimensional approaches.
As well known, the approximations introduced by an equivalent
frame assumption are strong and not commonly accepted to be
effective, since masonry is far to behave as a mono-dimensional
element, especially in the inelastic range. In some cases of engineering practice, indeed, and generally for complex geometries, it
may become critical even to foresee the existence of a univocal
equivalent frame, raising once again crucial questions on the effectiveness of such simplied procedures. Furthermore, due to their
intrinsic nature, pushover analyses of a masonry building performed by means of equivalent frames, are generally unsuitable
to reproduce partial out-of-plane failures.
On the other hand, their simplicity and direct applicability is out
of the question and in some cases, they still represent the most
straightforward FE procedure for a fast assessment of existing
structures [10]. To corroborate this idea, it should be also noted that
many codes of practice, as for instance the Italian one [1], furnish
simplied formulas for the prediction of the load bearing capacity
of piers and spandrels in shear and bending, thus implicitly accepting to model masonry structural elements as equivalent beams.
In order to circumvent the typical drawbacks of a mono-dimensional discretization of a masonry wall subjected to horizontal actions, in this paper a novel, fully 2D, approach is proposed.
Within a continuous 2D discretization of real scale masonry
structures, the most diffused numerical strategies are essentially
based either on micro- or on macro-modeling.
The so-called micro-modeling approach [13,14] relies into a distinct representation of bricks and mortar, usually modeled respectively by means of plate and shell elements and non-linear
interfaces with zero thickness. While such approach is rather indicated for the analysis of laboratory specimens and small panels, it
results computationally very expensive, especially in the non-linear range. For this reason, it is hardily applicable to large scale
structures.
On the contrary, macro-modeling [15,16] seems more powerful
for the numerical analysis of large scale masonry structures. Essentially it relies into the substitution of the heterogeneous assemblage of blocks by means of a ctitious homogeneous material,
which macroscopically should exhibit the same elastic and inelastic behavior of a small sample constituted by few bricks jointed by
mortar.
As well known, masonry shows a different stressstrain behavior
in tension and compression and along material axes, with marked
softening, both in tension and compression. Therefore, to be reliable, a macro-model should utilize a ctitious homogeneous material with softening and distinct inelastic properties (peak strength
and fracture energy) for tension and compression. Mechanical properties to be used should be derived either by means of experimental
data on small specimens loaded in various ways or by means of
homogenization procedures, i.e. solving a suitable boundary value
problem in the inelastic range on a representative element of volume, which generates the whole structure by repetition.
In this framework, in the paper, a novel simple and truly 2D
numerical model for the pushover analysis of large scale masonry
walls is presented.
In the model masonry is substituted with a ctitious orthotropic material, within a macro-modeling strategy. In particular, a
so called modied disturbed state concept and hierarchical single
surface approach (hereafter called DSC/HISS-CT model, see [15])
is utilized. In order to adapt the model to masonry actual behavior,
different hierarchical single surfaces (HISS) are used to model the
compressive and tensile (CT) behaviors of brickwork. The base
model (HISS) was introduced a few decades ago by Desai et al.
[17] and Desai and Salami [18], but it is applied here for the rst
time to large scale masonry walls subjected to horizontal actions.

277

DSC/HISS model is a unied and hierarchical model that can be


used to characterize elastic, plastic and creep deformations, as well
as micro-cracking that leads to fracture and failure, degradation or
softening, and healing or strengthening. Such approach has a number of advantages compared with other available models that often
account for only specic behavioral aspects. The new feature here
utilized is the use of distinct HISS yield functions in compression
and tension. The DSC model allows for the characterization of
non-associative behavior through the use of disturbance, and it
computes micro-cracking during deformation, which eventually
leads to fracture and failure.
After a preliminary validation of the DSC model at a constitutive
(or specimen) level, three large scale structural applications are
discussed. The main object of the paper is, indeed, to show how
real scale walls may be easily studied in the non-linear range by
means of the 2D approach proposed, thus avoiding a questionable
equivalent frame discretization.
In particular, the rst and second example are masonry walls
with two stories and respectively two and seven bays, already
studied by Salonikios et al. in [12]. The third example is a ve-story
ancient masonry building already analyzed within the so called
Catania project by many authors through different numerical strategies, including equivalent frames, macro-elements and commercial 2D codes where masonry is modeled by means of smeared
crack approaches.
In all cases, very good agreement is found between present
numerical results and existing literature, meaning that the fully
2D approach proposed may represent a valuable numerical alternative to a discretization of the structure by means of 1D elements.
2. Masonry macro-model: disturbed state concept with HISS
model
When attention is focused on the global response of a large
structure, although brickwork exhibits a heterogeneous and anisotropic behavior, an idealization by means of a homogenous and
isotropic continuum [19] is usually preferred in practice. Indeed,

(a)

(b)
Fig. 1. (a) RI and FA states in DSC, and (b) disturbance as a coupling between the RI
and FA states.

278

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

constitutive models suitable for anisotropic materials, such as masonry, contain several parameters that have to be quantied on the
basis of non-standard experimental tests, where tedious inverse
identications must be carried out.
On the other hand, the assumption of isotropic softening is not
completely satisfactory for a material such as masonry, which can
be loaded up to the ultimate tensile strength even if damage in the
perpendicular direction already occurred. This problem has been
partially solved in Feenstra and de Borst [20] using kinematic softening, i.e. with a model where the yield surface is shifted in the
direction of the rst principal stress. It is noted that the response
of kinematic softening is also not completely realistic. As a matter
of fact, if the material, initially, is loaded along a certain direction
until softening is completed and, then, is loaded in a direction
orthogonal to the crack previously open, an ideally plastic behavior
is found. This is due to the fact that all the fracture energy has been
consumed during the opening of the rst crack. To partially circumvent this drawback, the utilization of independent softening
parameters along material axes and for tension and compression,
with the utilization of independent yield surfaces (Rankine and Hill
type) is recommended, as extensively described in Loureno et al.
[21].
A similar principle is utilized here, where a HISS model with different parameters in tension and compression is applied to model
the post peak behavior. In this manner, the actual behavior of a
brittle material in presence of cracks already open along an orthogonal direction may be modeled. No orthotropy is applied for the
sake of simplicity, with the aim of limiting the inelastic parameters
to set for the simulations. However, it is worth noting that minimal

conceptual difculties arise for an implementation of scale parameters along material axes.
The following description of the DSC model is adapted from various publications, e.g. [17,18,2224] and the reader is referred
there for a detailed analysis of the approach proposed.
Here only a brief review is reported. In such a model, a deforming material element is assumed to be composed by two (or more)
reference states, namely the relatively intact (RI) and the fully adjusted (FA) state, as schematically represented in Fig. 1. Under
external excitations, the material is assumed to transform continuously from the relative intact (RI) state to the fully adjusted
(FA) state, Fig. 1a, at randomly distributed locations. The transformation involves micro-structural changes that cause particle reorientation and relative motions. The observed behavior is expressed
in terms of the RI and FA states using the disturbance function, D,
which acts as a coupling or interaction mechanism between the RI
and FA states, Fig. 1b. The disturbance grows as the material deforms and the plastic strain (or work) accumulates. Thus, DSC
intrinsically includes coupling in which the micro-cracked (damaged) or fully adjusted part also contributes to the response of
the material. The RI and FA states can be dened using various
models. Continuum elasticity or plasticity can be used to model
the response of the RI state, while the FA state can be assumed
to carry only hydrostatic stress or can be modeled using the critical
state model [24]. Brief descriptions of the models for the RI and FA
states, and the disturbance used in the DSC model, are given below.
It may be mentioned that the use of plasticity theory is one of
the possible ways to characterize the RI behavior. However, the

Ultimate Envelope

J2D

Phase
Change Line
(Critical
State)

J1
3R

(a) J 2D J1
1

(a)
= 0.9
= 0.77
= 0.6
= 0.3

= 0.0

(b) Octahedral plane; ( < 0.756 for convexity)


Fig. 2. HISS yield surface in J 1 

p
J2D space.

(b)
Fig. 3. Schematization
p [35]. (a): Yield
p of compressive and tensile HISS yield surfaces
surfaces in J1  J2D stress space. (b): Yield surfaces in r1  2r2 space.

279

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

DSC can be formulated by simulating the RI response as nonlinear


with irreversible deformations, without invoking the theory of
plasticity. Various hierarchical models for the inclusion of additional factors have been developed: nonassociative d1 -model
[25,26], kinematic and anisotropic hardening d2 -models for sands
and clays [2729], and viscoplastic dv p models [3032]. It is found
that the basic d0 -model is usually sufcient to characterize the RI
response of many materials. With the disturbance parameter D
embedded in the DSC, and the RI response modeled using the
d0 -model, both non-associativity and anisotropic behavior may
be accounted for, at least partially. This is because the disturbance
parameter D causes the deviation from normality of the plastic
strains with respect to the yield surface [33]. A comparison between a failure surface predicted by the DSC/HISS model and
experimental data for concrete [24] shows that the model can simulate a 2D behavior as well. Furthermore, it should be noted that
failure surfaces can be dened for different fragile materials exhibiting different compressive and tensile strength. Finally, a comparison on masonry wallettes in biaxial stress state between the
failure surfaces found by means of the DSC/HISS model and those
provided by different approaches available in the literature [34]
shows that DSC/HISS model lack of orthotropy along material axes
is not critical and that the typical anisotropic behavior of masonry
at failure is reproduced with sufcient care.
2.1. Relative intact (RI) state
The hierarchical single-surface (HISS) plasticity model [17] provides a general formulation for the elasto-plastic characterization
of the material behavior. In the model, a single continuous yield
surface is utilized, which, compared with some previous models
that involved multiple (discontinuous) yield surfaces, appears
computationally much more stable. This model, which allows for
isotropic and anisotropic hardening as well as associated and
non-associated plasticity, can be used to represent material responses based on the continuum plasticity theory [17,24]. In the
HISS model, the RI state is usually dened by the associated plasticity; accordingly, the yield function, F, Fig. 2, is given by:

F J 2D  aJ n1 cJ 21 1  bSr 0:5 0 a


J 2D J 2D =p2a

J 1 J 1 3R=pa
p
3D
Sr 227  JJ1:5

the ultimate condition. The hardening or growth function for the


plastic yield can be expressed as:

a1
ng1

where a1 and g1 are the hardening parameters, and n is the trajectory or the accumulated plastic strains. Using F, Eq. (1)(a), the incremental stressstrain equations for the RI (plasticity) model are
derived as [24]:


dr C e 

cF

@F T e
C e @Q
@r
@r C
@F T C e @Q @F c
@r
@n F
@r



h T  i1=2
@Q
@Q
@r

de

a
3
b

@r

where r is the stress vector, C e is the elastic constitutive matrix, d


denotes an increment, and Q is the plastic potential function. When
an associated ow rule is adopted, then F  Q .
For masonry, the compressive stress is assumed to be positive.
The yield surface,
Fig. 2, is active in the compression region in
pF,
the positive J 1  J 2D space. Masonry behavior is tension and compression is rather different, hence such yield surface, Fig. 2, is not
appropriate on the negative J 1 -axis (tension). Often, an ad hoc
model, such as the stress transfer approach [36], is used. In this approach, the computed tensile stress divided by the tensile strength
is redistributed. The HISS model can be used for both compression
and tension if the material parameters are determined from appropriate laboratory tests under compression and tension. Such a
model, called DSC/HISS-CT, can account for both compression
and tension yield and was introduced in [15,3537]. However,
DSC/HISS-CT has not yet been used for the analysis of large scale
unreinforced masonry buildings, as the cases hereafter treated.
Fig. 3 shows a sketch of the tension and compression yield surfaces for different values of the material parameters. The yield surfaces in tension and compression are expressed in stress space in
terms of J1, J2D and J3D, which include all components of the stress
vector in the three-dimensional space. As can be seen, the yield
surfaces, although plotted in the same stress space for convenience, are quite different and their parameters are determined
from relevant compression or tension tests. The model is fully 3D
and its specialization for 2D problems, both in presence of inand out-of-plane loads, is very straightforward.
2.2. Fully adjusted (FA) state

2D

where J2D and J 3D are the second and third invariants of the deviatoric stress tensor, respectively, J1 is the rst invariant of the total
stress tensor, pa is the atmospheric pressure, R is the reference
stress and is used mainly to include the intercept (c), which is proportional to the cohesive strength, Fig. 2, and c and b are related to

Various characterizations of the behavior of the FA state are


given in Desai [24]. Very concisely, here it is be pointed-out that
the FA state for masonry is determined using residual stresses in
the stressstrain curve response, Fig. 1b. In what follows, a fully adjusted stress equal to approximately 20% of masonry compressive

Dc=0.9

Dc =0.9

0.8
Du=1.00

0.6
0.4
0.2
0

Disturbance

Disturbance

0.8
Du=1.00

0.6
0.4
0.2
0

0.001 0.002 0.003 0.004 0.005 0.006 0.007

0.0003

0.0006

0.0009

0.0012

Trajectory of plastic strain

Trajectory of plastic strain

(a)

(b)

0.0015

Fig. 4. Variation of D versus trajectory of plastic strains for masonry with compressive strength = 10.5 MPa for (a) compression and (b) tension.

280

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

fc

ft

'

fm

ft = f m'

0.85 f m'

0.2 f m'

cu

(b)
Tensile stress (MPa)

Compressive Stress
(MPa)

(a)
3.5
3
2.5
2
1.5
1
0.5
0
0

0.003

0.006

0.009

0.12
0.1
0.08
0.06
0.04
0.02
0
0.0002

0.012

0.0004

Axial Strain

Axial Strain

(c)

(d)

0.0006

0.0008

Fig. 5. (a) Compressive and (b) tensile behavior of a generic masonry prism. (c and d): Actual behavior used in the numerical simulations.

Table 1
Parameters for the DSC/HISS-CT model.
Compressive strength (MPa)

Kind of behavior

Masonry wall

3.00

Compression
Tension

HISS plasticity parameters

a1

g1

1650
1650

0.20
0.20

0.100
0.030

0.30
0.40

5.43
5.55

1.900
0.25

1.55e-9
5e-8

0.3000
0.4000

3500
70000

1.72
1.285

resistance in uniaxial tests [38] and a residual stress equal to 14% of


masonry compressive strength [39] are adopted.
2.2.1. Disturbance
Disturbance, D, can be dened in terms of measured stress, void
ratio, pore water pressure, and nondestructive properties, such as P
or S wave velocities [24]. In terms of stress, it is dened as:

ri  ra
ri  rc

DSC Parameters
(disturbance)

E (MPa)

Compressive Stress (MPa)

Kind of material

3
Constructed

2.5

DSC/HISS-CT

2
1.5
1
0.5
0
0

D Du 1  expAnZD 

where Du is the ultimate disturbance (often assumed to be equal to


1), nD is the trajectory of (deviatoric) plastic strains, and A and Z are
disturbance parameters. Parameters in Eq. (5) are determined on
the basis of the values of the measured ra at various points on
the stressstrain curve, Eq. (4), and the corresponding values of nD .
Variations of disturbance, D, for typical stressstrain curves for
a peak stress equal to 10.5 MPa are shown in Fig. 4. In the same gure, the value of the critical disturbance, Dc, is assumed to occur at
about 0.9, which is roughly the intersection between tangents of
the middle and ultimate zones of the plastic strain cure.
From a physical point of view, micro-cracks start when the disturbance is about 0.003 for compression and about 0.0001 for tension respectively. Then micro-cracks grow up to Dc = 0.9, where
fracture and macroscopic cracks occur [24]. As a consequence,

0.005

0.01

0.015

0.02

Strain

(a)
Tensile Stress (MPa)

where ri , ra and rc are RI, observed and FA stress values, respectively. To introduce D in the DSC model, D has to be expressed in
terms of basic variables, such as accumulated plastic strains or
work. In what follows, D is expressed in terms of the accumulated
deviatoric plastic strains using a Weibull [40] function:

0.12
0.1

Constructed

0.08

DSC/HISS-CT

0.06
0.04
0.02
0
0

0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 0.0007 0.0008

Strain

(b)
Fig. 6. Comparison between DSC/HISS-CT predictions and constructed data for
(a) compressive and (b) tensile behavior of a masonry prism.

D can be used as a measure to identify micro-cracks growth (based


on test data) leading to fracture and failure. For example, when the

281

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

0.5 (MPa)
2

Point D

0
-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0.5

1 (MPa)

-0.5

Point A
-1

Point B
-1.5
-2
DSC/HISS-CT model by mesh 8*8

-2.5

Point C
-3
-3.5

(a)
0
-0.01

-0.005

0
0

-0.01

-0.005

-0.01

-0.5

-0.5

-0.5

-1

-1

-0.005

-1

-1.5
-1.5

-1.5

1 (MPa)

-2
-2

-2

-2.5
-2.5

-2.5

-3
-3

-3

-3.5

-3.5

(b) point A

-3.5

-4

(d) point C

(c) point B

0.12

0.12

0.1
0.08

0.1
0.08

0.06

0.06

0.04

0.04
0.02

0.02

2 (MPa)

0
0

0.02

0.04

0.06

0.08

0.1

0.12

0.0002

0.0004

0.0006

0.0008

0.001

(e) point D
Fig. 7. (a) Predicted failure surface for a 800  800 mm specimen in biaxial state of stress meshed with 64 elements, (b) stressstrain curve for point A, (c) point B, (d) point C,
and (e) point D.

disturbance reaches 0.9 or higher values, it may reasonably stated


that fractures initiate and grow. Within a Finite Element
discretization, elements that reach this critical value are identied
at the end of each load iteration. This post-processing phase
allows to plot disturbance D contours, which reasonably represent
the state of damage of the structure during the non-linear a
nalysis.

Both the RI and FA states contribute to disturbance D (coupling


effect). The following DSC equations in incremental form show
such coupling [24]:

draij 1  Ddriij Ddrcij dDrcij  riij

where d denotes an innitesimal increment and rij is the stress tensor component.

282

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

Phase change parameter n can be found from the slopes of the


phase change line, ct , and the ultimate line, c, [15,37] as:

2.3. Numerical simulation of masonry prism behavior


Hereafter a concise review of the procedure utilized to both t
existing experimental data and assign internal variables values is
reported.
2.3.1. Laboratory tests
On the basis of a number of experimental tests on masonry
prisms, a series of closed form equations have been used by Kaushik et al. [38,41] and Pandey and Meguro [39] to characterize the
uniaxial compressive and tensile behavior of masonry, see also
Fig. 5. In particular, when dealing with the compressive behavior,
the following equations hold:

     
2
fm fm0 2 ee0  ee0
e 6 e0



e0 < e 6 e1
fm fm0 1  0:15 eecuee00
fm 0:2fm0

e > e1

a
b

Eq. (7)(a) is the Hognestads model for concretes compressive


behavior [42], which adapts well to masonry uniaxial compression.
Eq. (7)(b) is a linear function between peak stress and residual
stress, whereas Eq. (7)(c) is the residual stress, which is assumed
here equal to 20% of the peak stress.
On the other hand, when dealing with the tensile behavior, the
following equations hold:

ft Ec e
ft

kfm0

e 6 et

e  et
1
em  et

e > et e

Eq. (7)(d) is obtained from the elastic behavior of masonry in the


tensile region up to the peak tensile stress; softening behavior occurs after this point. This behavior is modeled by a linear function.
Here, a linear function, Eq. (7)(e), is assumed to determine DSC
parameters describing the softening behavior of masonry in tension. However, the DSC/HISS-CT model accounts for the nonlinear
behavior of masonry in the tensile region in terms of an exponential
strain function, Eq. (5). In Eq. (7), fm0 is the compressive strength of a
masonry prism, 0:002 6 e0 6 0:004, 0:003 6 ecu 6 0:007 [3841],
em 10et , k is a correction parameters varying between 0.1 and
0.25, Ec is masonry elastic modulus, and e1 and et are shown in
Fig. 5. Fig. 5a shows graphically the uniaxial compression behavior
adopted for masonry, Eq. (7) from (a) to (c), whereas Fig. 5b shows
the uniaxial tensile behavior.
In Fig. 5c and d, Eq. (7)(ac) are used to construct the compressive behavior of masonry prisms with compressive strengths
fm0 3:0 and tensile strength equal to 0.1 MPa, respectively. Such
values for fm0 are chosen to be consistent with masonry prisms in
later applications.
When the uniaxial behavior of the numerical model is derived
from experimental test, it can be stated that the macroscopic
behavior of the homogeneous material is derived by means of a
so-called constructed procedure. The details of the procedure
to determine all model parameters are given elsewhere
[15,24,37]. Hereafter only the fundamental steps followed are
briey summarized.
2.3.2. Determination of parameters
Ultimate parameters c and b are determined by means of an
indirect procedure. In particular, parameter c represents the
asymptotic ultimate stress. The yield surface
p approaches a straight
p
line with a slope equal to c in the J 1  J 
2D space when the hardening parameter is a 0.
b controls the shape of the yield surface in the octahedral plane.

ct
n2

n
c

0:5
8

Hardening parameters, a1 and g1 are evaluated through the


hardening function. The proposed hardening function, a, is expressed in terms of a plastic strain trajectory or an accumulated
plastic strain. Taking the natural logarithm on both sides of Eq.
(2), we obtain:

lna lna1  g1 lnn

The DSC parameters are determined based on the data in Fig. 5.


Table 1 shows the parameters of the model for a clay bricks masonry and a stone brick masonry in compression and tension. An
ultimate disturbance Du = 1 with atmospheric pressure
pa 0:1013 MPa are assumed. Taking the natural logarithm twice
on both sides of Eq. (5) we obtain:

lnA Z  lnnD ln ln1  D

10

Therefore, by knowing A and Z, the best tting line through a


least squares procedure for the set of points ln ln1  D;
lnnD  may be determined, to evaluate parameters A and Z. Details
on how to determine such parameters are given by Desai [24],
Akhaveissy and Desai [15,37] and Akhaveissy et al. [43]. The
parameters of the model so obtained are summarized in Table 1.
2.3.3. Specimen level validations
The model is validated at specimen level using parameters summarized in Table 1 and Eqs. (3)(a) and (6). At the specimen level,
predictions are obtained by integrating the incremental constitutive relations, Eq. (3)(a). Fig. 6 shows a comparison between predictions from the model for clay brick masonry and the
numerical stressstrain curves from Fig. 5.
As it is possible to notice from Fig. 6, the results obtained by
means of the DSC/HISS-CT model match well the constructed
behavior, which bases on analytically deduced compressive and
tensile mono-dimensional curves.
The biaxial inelastic behavior in both tension and compression
has been also discussed in detail, in particular evaluating the possible orthotropy of the model under 2D membrane stress states. To
this aim, a 2D specimen with dimensions 800  800 mm is considered when subjected contemporary to horizontal and vertical actions and the corresponding failure surface provided by the DSC/
HISS-CT model is analyzed in detail. The compressive and tensile
strength are assumed equal to 3 MPa and 0.1 MPa, respectively.
The calibrated parameters of the model for such material are summarized in Table 1. Fig. 7a shows the failure surface obtained
through the DSC/HISS-CT approach. The panel is modeled by
means of 64 eight-noded isoparametric elements, with a regular
8  8 mesh.
The corresponding stressstrain curves for points AD indicated
in Fig. 7a are represented in Fig. 7be.
The following Fig. 8 shows a comparison between the predicted
failure surface (peak values) and available data from Thanoon et al.
[44] regarding a mortar-less grouted masonry exhibiting interlocking with similar mechanical properties. As can be noted, the comparison shows good correlation on previously presented results.
The adopted model is isotropic, but orthotropy may be accounted
for simply with a scale transformation along one of the axes.
3. Structural level applications
The DC/HISS model here specialized for masonry has been
implemented in a plane stress non-linear code, to perform

283

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

0.5

2 (MPa)
1 (MPa)

0
-4.5

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0.5

-0.5
-1
-1.5
-2
-2.5
-3
-3.5
-4

DSC/HISS-CT model by mesh 8*8


Kupfer envelope [44]

-4.5
Fig. 8. Comparison between failure surface found with the DSC/HISS-CT model and Kupfer envelope.

1
0.9

Acceleration (g)

0.8

Bedrock

0.7

Soft soil

0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.5

1.5

2.5

3.5

4.5

Period (sec)
Fig. 9. Site response spectrum for very high seismicity zone [48].

pushover analyses on entire masonry buildings. In particular, in


order to test the capabilities of the fully 2D numerical model proposed in the paper, two small and large scale walls already analyzed with an equivalent frame approach by Salonikios et al. [12]
and a ve story building studied within the so-called Catania Project by many authors [4547] are here re-considered and analyzed
with the full 2D DC/HISS model.
3.1. Acceleration spectrum
Iranian earthquake code [48] is utilized to discuss some safety
assessment considerations on the buildings under considerations.
Similarly to other national codes (as for instance the Italian one
[1]), such code contains a seismic hazard map that divides Iran into
a series of zones of equivalent seismicity. This map contains four
zones: (1) low seismicity, (2) moderate seismicity, (3) high seismicity, and (4) very high seismicity. The code uses the concept of
acceleration response spectra into the design process. The effective
peak ground acceleration for zone 1 is assumed 0.20 g, 0.25 g for
zone 2, 0.3 g for zone 3, and 0.35 g for zone 4. These levels of
ground motion correspond to an exceedance of the event roughly
one time every 475 years, having approximately a 10% probability
of exceedance in 50 years.

The code also considers site soil conditions in the resultant


ground motion estimation, dividing soils into four groups in terms
of seismic motion shear velocity Vs. For group 1 Vs is greater than
750 m/s, 375 < Vs < 750 for group 2, 175 < Vs < 375 for group 3,
and Vs < 175 for group 4. Groups 1 and 4 soils are the so-called bedrock and soft soil, respectively. Hence, the acceleration spectrum is
estimated basing on both seismic zones of equivalent seismicity
and site soil conditions. Fig. 9 shows a 5% damped elastic response
spectrum for very high seismicity zone with site soil conditions
corresponding to bedrock and soft soil.
The site response spectrums in Fig. 9 are used to estimate the
performance level of the unreinforced masonry buildings studied
in the following sub-section.
3.2. Two story buildings
The structures under consideration are respectively a one-bay
and a seven-bay wall with two stories. Fig. 10 shows the geometry
of the case studies, along with the lateral loads applied. Apart masonry self-weight, extra masses are considered at the oor levels,
in agreement with Salonikios et al. [12]. For the one-bay frame, a
uniformly distributed mass equal to 6 tons/m was assumed for
the rst oor, and 4 tons/m was assumed for the second oor

284

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

6 t/m
0.57*V

4 t/m

3 t/m

2 t/m

0.75

1.50
2.00

3.50
0.43*V

1.50
3.50

[m]

2.75

1.50

1.50

1.50

[m]

1.50

1.50

1.50

1.50

1.50

1.50

1.50

(a)

1.50

1.50

1.50

1.50

1.50

1.50

1.50

1.50

(b)
Fig. 10. (a) One-bay and (b) seven-bay two-story masonry buildings.

q= 40 N/mm

0.57V ,

q= 60 N/mm
0.43V

4500 mm

Fixed node

Fig. 11. Boundary conditions and mesh used for the analysis of the masonry wall.

[12,49,50]. The corresponding values for the seven-bay frame were


assumed to be 3 and 2 tons/m, respectively [12]. Mechanical properties assumed for masonry are the following: walls thickness
equal to 0.6 m, volumetric mass q 2 t=m3 , Young modulus
E = 1650 MPa, Poisson ratio m 0:2, tensile strength ft = 0.1 MPa
and compressive strength fm = 3.0 MPa. Joints cohesion and friction
angle are kept equal to 0.09 MPa and 30, respectively, in agreement with [12].
A triangular distribution of horizontal forces acting on the left
vertical edge and mimicking a rst mode distribution is applied,
in agreement with Salonikios et al. [12]. In Fig. 10, V represents
the value of the base shear force, incremented step-by-step within
the pushover approach adopted.
Both the two-bay and the seven-bay structures are analyzed by
means of the DSC/HISS-CT model using a macro-modeling

approach. For the numerical analyses, masonry is discretized by


means of eight-noded plane stress continuum elements. The
one-bay building is modeled by 768 eight-noded isoparametric
elements, 2646 nodes and 4924 DOFs, whereas the seven-bay
building is modeled with 3648 elements, 12732 nodes and 23058
DOFs. Maximum error allowed within a displacement based convergence criterion is kept equal to 1e-3.
Fig. 11 shows the discretizations adopted for the analyses. An
iterative-incremental method with an initial stiffness scheme
(modied Newton Raphson) is used in both cases. The frontal iterative method is adopted to solve the governing equilibrium equations for the structure. The maximum front width for solving the
equations related to the one-bay building is 408. The total number
of iterations required for this analysis is around 5300. A Cholesky
decomposition is used for matrices factorization, with all matrices

285

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

200
180

Point As

Base shear force (kN)

160

Point Ar

Point Bs

140
120

Point Br

100
Equivalent Frame [50]

80

Collapse prevention
level based on FEMA

60

Discrete element [12]

40

Equivalent Frame [12]

20

DSC/HISS-CT (present work)

0
0

10

20

30

40

50

60

70

80

90

100

110

120

130

140

Displacement at top (mm)

(a)
200

Base shear force (kN)

180
160
140

Point Bs

120

Point Br

100
80

Equivalent Frame [50]

60

Discrete element [12]

40

Equivalent Frame [12]

20

DSC/HISS-CT (present work)

0
0

10

15

20

25

30

35

40

Displacement at top (mm)

(b)
Fig. 12. Comparison between the different models and the predicted loaddisplacement curve for the one-bay unreinforced masonry building (a) the curve for displacement
at roof up to 140 mm and (b) a detail of the curve for displacement at roof up to 40 mm.

stored through a band method. The total time needed to analyze


respectively the one-bay and the seven bay structure is around
68 and 236 min on a PC equipped with 2 GB of RAM and a CPU Core
Duo Processor T2050.
Thanks to the relatively small dimensions of the walls under
consideration, a heterogeneous approach may be attempted at a
structural level. Therefore, in such examples, bricks are modeled
by means of quadrilateral elastic elements in plane stress, whereas
mortar joints are reduced to interfaces exhibiting elasticplastic
behavior [12]. Joints strength domain is constituted by a composite
yield surface, including a classic MohrCoulomb criterion and a
tension and compression cut-off.
In order to further evaluate the reliability of the results obtained, an equivalent frame approach is also adopted to analyze
both structures [12,50].
Fig. 12a shows a comparison among loaddisplacement curves
provided by the DSC/HISS-CT model and the other models (equivalent frames and heterogeneous approach) for the one-bay wall
previously described. End of simulations for all models is assumed
when lack of convergence is reached. A detail of the global behavior of the models up to 40 mm is shown in Fig. 12b. As it is possible
to notice, the agreement among all models is quite reasonable.
As it is possible to notice from the detail in Fig. 12b, the initial
slope of both the heterogeneous and the homogeneous DSC/HISSCT model of the global pushover curve is similar and sensibly lower
than equivalent frame approaches. This is rather usual when
equivalent frames with rigid offsets are utilized. Masonry elasticity
modulus assumed in the simulations is equal to 1650 MPa. Being

masonry compression strength fm0 around 3 MPa, Table 1, the ratio


between Young modulus and compressive strength fm0 kept in the
simulations is equal to 550. Here it is worth noting that such value
is in agreement with existing literature. For instance, FEMA 306
[51] indication is to keep such ratio equal to 550, the Canadian masonry code S304.1 [52] recommends E to be 850 times fm0 , whereas
International Building Code [53] and MJSC documents [54] suggest
a ratio equal to 700.
Fig. 12 shows that the DSC/HISS-CT predicted ultimate load is
about 187 kN, in very good agreement with ultimate loads provided by the heterogeneous approach (176.5 kN), the equivalent
frame methods by Salonikios et al. [12] (179.5 kN) and by
Akhaveissy [50] (195 kN).
In order to determine the performance level of the one-bayunreinforced masonry building, the demand spectrum and the
capacity spectrum are plotted within the acceleration-displacement response spectra (ADRS) format, Fig. 13.
Points A and B in Fig. 13 represent the so-called performance
points of the one-bay structure with a peak ground acceleration
equal to 0.35 g, assuming bedrock and soft soil conditions, respectively. The procedure A of ATC-40 guidelines [55] is adopted to
determine the performance points. Within this approach, the performance point with bedrock condition is represented by point A in
Fig. 13 and point Br in Fig. 12. As a consequence, the roof displacement and the base shear force are respectively equal to 13.94 mm
and 153.89 kN. Conversely, point B in Fig. 13 corresponds to the
performance point of the one-bay building for soft soil condition.
B corresponds to point Bs in the displacement-base shear diagram

286

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

0.5
PointB

Spectral acceleration [g]

0.45
0.4
0.35
Soft soil condition

Point A

0.3

Bedrock condition

0.25
0.2
0.15
0.1

Capacity curve

0.05

Demand curve

0
0

10

20

30

40

50

60

70

Spectral displacement [mm]


Fig. 13. Acceleration-displacement response spectra for the one-bay building.

[mm]

Point Bs

Point As

(b) roof displacement:


20.08 mm

(c) roof displacement:


85.30 mm

Point Br

(a) roof displacement: 13.94 mm

Fig. 14. One-bay wall. Disturbance parameter patch at different global deformation level in the pushover curve. (a): Point As. (b): Point Br. (c) Point Bs in Fig. 12.

of Fig. 12. For a soft soil condition, the displacement of the building
roof and the base shear are 20.08 mm and 162.03 kN, respectively.
The one-bay building disturbance parameter patches for both performance points, i.e. Br and Bs in Fig. 12, and the point corresponding to the ultimate displacement, point As , are sketched in Fig. 14.
As it is possible to notice, a signicant part of the wall reaches a
critical disturbance value greater than 0.90 in correspondence of
point As, but also for Br and Bs. Here it is worth mentioning that,
when disturbance is over 0.90, Dc , it can be stated that failure of
the wall is prone to occur, since the maximum disturbance possible
is equal to 1.00 [15,37]. It is worth noting that the slope of the line
connecting the points with abscissa Dc = 0.90 and Du = 1.00, Fig. 4a,
is near zero. It can be therefore deduced that the disturbance
parameter grows quickly from Dc to Du. As a consequence, it can
be supposed that, when the disturbance parameter is close to Dc,
a particle is reached to fully adjusted state. Consequently, for practical purposes, the disturbance parameter can be evaluated using
Dc instead of Du. It may occur in practice that the disturbance
parameter for a single Gauss point may be smaller than Dc, due
to boundary conditions and applied loads. Hence, it can be assumed that full failure has not been occurred for this particle.
Fig. 14a and b show a signicant damage, even at early stages of

the external load, in correspondence of the spandrel of the rst


oor, both for a bedrock and soft-soil condition. Roof displacements are respectively around 15 and 20 mm for bedrock and soft
soil conditions, whereas collapse may occurs at displacements sensibly higher, see point As in Fig. 12a, namely at around 85.3 mm. It
should be noted that the slope between points As and Ar in the
loaddisplacement curve, Fig. 13a, is negative, with the structure
showing instability. Therefore, point Ar cannot be kept as ultimate
displacement for the structure. According to FEMA 356 [56], the
value of drift respectively for the immediate occupancy criterion,
the life safety level and the collapse prevention level are equal to
0.1%, 0.3% and 0.4% for a bed-joint sliding mode and respectively
0.1%, 0.3(h/L)% and 0.4(h/L)% for a rocking mode. h and L here indicate respectively the height and width of a masonry shear wall. For
piers, evaluations based on FEMA 356 [56] and performed on the
one bay building show that a bed-joint sliding mode never occurs.
Therefore, the drift value to adopt for the immediate occupancy
criterion, the life safety level and the collapse prevention level
are 0.1%, 0.7% and 0.9%, respectively. The results obtained by
means of the DSC/HISS-CT model for the one bay building show
that the value of drift for point Br is 0.22% for ground oor and
0.16% for rst oor. Therefore, the performance level of the one

287

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

bay building on a bedrock condition, point Br, is smaller than the


life safety level provided by FEMA. For a soft soil condition, the
drift values corresponding to point Bs are 0.30% and 0.25% for
ground and rst oors respectively. Comparing such values with
FEMA 356 [56], it can be deduced that the performance level is
again lower than the life safety level. The roof displacement corresponding to the collapse prevention level provided by FEMA, 0.9%,
is equal to 67.5 mm. Such displacement level is shown in Fig. 13a
and appears close to the peak point As provided by DSC/HISS-CT
model. The good agreement between the ultimate displacement
As predicted by the model and the displacement corresponding
to the collapse prevention level by FEMA, indicates that quite reliable results, also in terms of displacements and total ductility, may
be obtained with the model proposed. On the other hand, the slope
between Bs and As in Fig. 13a is very small, indicating that a failure
mechanism activates between point Bs and As. The exact point of
activation is obviously unknown but certainly quite near the prediction made by FEMA 356 [56]. For practical purposed, point As
may be kept as the roof ultimate displacement provided by the
model in presence of the activation of the failure mechanism. More
detailed analyses conducted rening the load steps near this point
would provide the exact position where collapse occurs.
In Fig. 14, damage patches obtained numerically for points As
(a), Br (b) and Bs (c) are sketched for the sake of completeness.
Fig. 14c corresponds to the formation of the failure mechanism,
and suggests that failure occurs for the collapse of the corners of

Active failure
mechanism

Fig. 15. One-bay wall. Predicted deformation in correspondence of the performance


point for the bedrock condition, the roof displacement being equal to 13.94 mm.
Distorted elements suggest a concentration of damage exceeding 0.9, the displacements are exaggerated by a factor of 3.

1000
900

Base shear force (kN)

800

Point C

700
600

Point B

500

Point A

400

Equivalent Frame [50]

Collapse prevention
level based on FEMA

300

Discrete element [12]

200

Equivalent frame [12]

100

DSC/HISS-CT (present work)

0
0

20

40

60

80

100

120

140

160

Displacement at roof (mm)

(a)
1000

Base shear force (kN)

900
800
700
600
500

Point A

400

Equivalent Frame [50]

300

Discrete element [12]

200

Equivalent frame [12]

100

DSC/HISS-CT (present work)

0
0

10

Displacement at roof (mm)

(b)
Fig. 16. Comparison between the different models and the predicted loaddisplacement curve for the seven-bay unreinforced masonry building (a) pushover curve for roof
displacements up to 150 mm and (b) a detail of the curve for roof displacements up to 10 mm.

288

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

the window and the door. The presence of a single large strut that
circumvents the top window is also quite visible. The predicted deformed shape at the performance point for a bedrock condition,
point Br in Fig. 12a, is nally shown in Fig. 15. Displacements are
magnied by a factor of 3. The roof displacement at the performance point is equal to 13.94 mm. The typical instabilities of some
elements in the deformed shape indicate the activation of a failure
mechanism.
From simulation results, it can be concluded that the one-bay
unreinforced masonry building under consideration safely withstands a horizontal seismic action represented by a spectrum with
peak acceleration equal to 0.35 g, for both bedrock and soft soil
conditions. Conversely, it should be noted that common unreinforced masonry buildings, in both Europe and Middle-East are built
with sensibly lower thicknesses (around 350 mm), which, under
the same load conditions, may exhibit an inadequate behavior under expected earthquakes.
Akhaveissy and Desai [15] also proposed a close form solution
to estimate the base-shear force of simple, in-plane loaded masonry shear walls. Such approach may be easily generalized to
shear walls with different aspect ratios and openings, as the case
here discussed. While the reader is referred to Akhaveissy and
Desai [15] for further details on such model, here it is worth noting
that the approach bases essentially on the evaluation of the fracture length at the base of the walls. When DSC/HISS-CT model simulations are at disposal, such fracture length can be evaluated
directly from disturbance parameter patches at failure (point As)
or, alternatively, assuming typical values.
Akhaveissy and Desai assumed that the length of the tensile
crack for piers is equal to one half of the piers width [15]. Such approach is rather questionable albeit generally in good agreement
with experimental evidence. Therefore, hereafter a more sophisticated and less questionable approach based on formulas reported
on the Italian code [1], is used to draw some conclusions with
at hand calculations. In the Italian code, the length of the compressed zone is evaluated knowing the normal compression force
on the head of the pier and the bending moment, assuming a block
distribution of vertical stresses on the toe and imposing vertical
and rotational equilibrium of the wall. For the problem under consideration, it turns out that the compressed zone length is equal to
294 mm, with a corresponding tensile stress length equal
1206 mm.
When dealing with a masonry material with mechanical properties as in Table 1, and assuming that the one bay wall is constituted
by two independent piers (thus neglecting spandrels strength for
the sake of simplicity), it can be demonstrated that the lateral carrying capacity of each element is around F 1 = 94 kN, with a total
resistant base shear equal to F 2F 1  188 kN [15]. Such values
are very near to those obtained by means of the Italian code [1].
Comparing the result obtained manually (188 kN) with that
provided by the DSC/HISS-CT model (187 kN), it can be concluded
that the simplied approach furnishes very good estimates of the
failure base shear, with a negligible error. Such computation has
some technical relevance, because it allows a manual evaluation
of the ultimate shear resistance of a complex structure accounting
only for piers strength.
In order to check the ability of the proposed model in analyzing
more complex structures, a seven-bay two-story masonry wall
again originally studied by Salonikios et al. [12], is hereafter reanalyzed by means of the DSC/HISS-CT approach, modeling masonry within a macroscopic schematization.
Fig. 16 shows a comparison among roof displacement-base
shear curves provided by the present model, the aforementioned
heterogeneous model and two equivalent frame approaches.
A detail of the pushover curves so obtained up to 10 mm is
sketched in Fig. 16b. As it is possible to notice, the agreement

between present results and alternative models available in the literature is quite satisfactory. Point C in Fig. 16a corresponds to the
formation of a failure mechanism in the 2D model, whereas point A
roughly represents the elastic limit of the structure.
The roof displacement in correspondence of point A of the pushover curve is 2.94 mm, with a drift equal to 0.04% and 0.05% for
rst and ground oor respectively. The disturbance parameter
patch for point A is shown in Fig. 17a. The drift increases in correspondence of point B up to 0.4% and 0.8%. It is noticeable that signicant stiffness degradation is observed both in the present and
previous numerical simulations [49] for a drift equal to 0.4%.
Therefore, the deformation of the structure at point B, Fig. 16a,
may be regarded as the instant where a failure mechanism activates. Disturbance parameter patch, reported in Fig. 17b gives a
more precise insight into the mechanical phenomenon. All spandrels result cracked and the diagonal damage begins to spread inside ground oor piers. Point C may be regarded as the instant
where the failure mechanism is fully active. Here the drift is equal
to 2.3% and 2% respectively for the rst and ground oor. There,
ground and rst oor piers result completely cracked, see disturbance parameter patch reported in Fig. 17c.
Fig. 16a shows that the predicted ultimate load for the sevenbay wall is 872 kN, which has to be compared to values provided
by the discrete element method (820 kN) and the equivalent
frames by Salonikios et al. [12] (705 kN) and Akhaveissy [50]
(835 kN).
While the initial stiffness of the structure shows some variability between present model and previously presented approaches,
considered in any case acceptable due to the inaccuracies introduced by a discretization with equivalent frames, both peak
strength and stiffness degradation appear in very good agreement
with previously presented approaches.

[mm]

(a) roof displacement: 2.94 mm

(b) roof displacement: 42.42 mm

(c) roof displacement: 150.62 mm


Fig. 17. Seven-bay wall. Disturbance parameter patch at different global deformation level in the pushover curve. (a): Point A. (b): Point B. (c) Point C in Fig. 16.

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

289

Fig. 18. Building with ve stories. Geometry (in meters) of the inner wall under consideration.

Table 2
Mechanical properties of the masonry unit and joints [45].
c (MPa)

ft (MPa)

fm (MPa)

/ [deg]

0.15

0.1

3.00

26.56

In Fig. 16a, a comparison between the ultimate displacement


determined by FEMA and the model is provided for the seven
bay building. The ultimate displacement provided by FEMA stands
between points B and C. A comparison between the disturbance
parameter for points B and C, Fig. 17b and c, shows that the failure
patterns are very similar. On the other hand, the stiffness of the
global pushover curve is, albeit still positive, almost zero near point
B. This means that the wall, near point B, is prone to collapse. In
addition, the almost zero stiffness of the pushover curve is obviously related to a quick growth of the failure regions from point
B to point C. Basing on such considerations, point B in Fig. 16a
and the related damage pattern can be reasonably regarded respectively as the ultimate point of the seven-bay building and the corresponding failure mechanism. It is nally worth mentioning that
DSC/HISS-CT predicted results, analogously to what done for many
other models available, should be critically interpreted by users, to
properly identify the real behavior of the structure.
3.3. Application to an old masonry building
An inner wall of a ve-story building located in via Martoglio
(Catania, Italy) is analyzed by means of the previously presented
DSC/HISS-CT model. The geometry of the panel under consideration is shown in Fig. 18. The building has been analyzed through
several non-linear codes by many authors, including Brencich et al.

[45], Braga et al. [46], Liberatore [57], Gambarotta and Lagomarsino [58], Braga and Liberatore [59], Milani et al. [47], etc. The thickness of the last story is 160 mm while the thickness of the other
stories is 300 mm. Numerical data for joints reduced to interfaces
(cohesion c, tensile strength ft, compressive strength fm and friction
angle u) and used in the present analysis to characterize macroscopically masonry behavior are taken from Brencich et al. [45]
and are summarized in Table 2.
The discretization used to analyze the wall relies into 3168
eight-node isoparametric elements and 11919 nodes; the number
of DOFs is equal to 20274. Fig. 19 shows the mesh used for the
numerical analysis within the DSC/HISS-CT model.
The equivalent static forces at story levels applied to the model
are again available from Brencich et al. [45] and are summarized in
Table 3.
The schematization by means of an equivalent frame model and
the equivalent static horizontal loads applied are shown in Fig. 20.
Both displacement and force convergence criteria are used in the
non-linear analyses. Maximum error allowed within each iteration
is assumed equal to 0.2%. Here, L2 norms in terms of both residual
forces and displacements are adopted within the convergence criterion. The reference value for force control is the L2 norm of the
increment of the force vector, whereas the reference value for displacement control is the L2 norm of the sum of the iterative displacements at each increment [15].
The capacity curve provided by the present model is compared
with results obtained by means other models, namely equivalent
frame and a discrete element approach, in Fig. 21.
Fig. 21 shows good agreement among all models. In particular,
the ultimate base shears obtained with the DSC/HISS-CT model,
the equivalent frame and the discrete element methods are quite
similar, being respectively equal to 1480, 1430 and 1258 kN.

290

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

Fig. 19. Building with ve stories. Mesh utilized within the DSC/HISS-CT model.

Table 3
Equivalent static forces for the ve-story masonry building [45].
Level

Unit weight of masonry c 17 kN=m3


Thickness of wall (mm)
Masonry weight (kN)
Applied load on stories (kN)
Total weight (kN)
hi (m)
P
P
ci hi nj1 W j = nj1 W j hj

300
264.70
305.05
569.75
0.64
0.0704

300
480.55
305.05
785.60
4.52
0.4974

300
486.70
372.90
859.60
8.22
0.9045

300
486.70
372.90
859.60
11.92
1.2786

300
373.15
372.90
746.05
15.62
1.7187

160
129.80
53.70
183.50
19.12
2.1038

F h W  C  R  e  b  I  c I 1:
Equivalent static force Fh (kN)

16.05

311.00

439.65

512.90

154.40

e 1: R 1: b b1  b2 4: S 12 ); C 0:1

du
dy

11

2860

F5

2975

F4

2975

F3
F2

2975

Reasonable agreement is also found when dealing with the initial


slop of the loaddisplacement curve. Finally it is interesting to notice that the DSC/HISS-CT model presents a slightly visible brittle
behavior.
Finally, crack patterns provided by the DSC/HISS-CT model at
increasing roof displacements are shown in Fig. 22. As can be noticed, spandrels belonging to 3rd and 4th oor result almost totally
damaged, even at quite small global deformations (roof displacement: 9.4 mm). The lintel of the main entrance of the building results also severely cracked, the damage spreading progressively on
a quite large area over the lintel, up to about 2.2 m in correspondence of point B. From a detailed analysis of the crack patter, it
can be deduced that such zones should be properly strengthened.
From point C to D, the pushover curves slightly decreases, with
a quick increase of global displacements and damage, as can be deduced from Fig. 22c and d. A severe damage of the piers belonging
to the ground oor is particularly evident, clearly dening the failure mechanism near point D (74 mm). A global strengthening of
the building is strongly recommended to increase both ductility
and peak shear.
The pushover curve in Fig. 21 provides also important information on the ultimate drift of the walls, which is useful to determine
the pseudo ductility. In particular, the ultimate drift, du , is evaluated as the drift in correspondence of a shear strength degradation.
The yield drift, dy , is determined equating the areas under the
equivalent bilinear 1DOF model and the actual capacity curve, as
suggested by many codes of practice [1]. In the present case values
of the ultimate and yield drifts are equal to 73.8 mm and 5.8 mm,
respectively. The pseudo ductility l is hence determined as the ratio between ultimate and yield drift as:

F1
3710

156.30

3770

3285

3430

4245

3320 3450

3950

Fig. 20. Building with ve stories. Equivalent frame model and seismic loads
(dimensions in millimeters).

From simulations results, the pseudo ductility for the building is


12.63.
The deection amplication factor, Rl , can be determined
through Newmark and Hall method [61] and is equal to 4.93. The
over strength factor is also computed and is equal to 4.5. The
strength reduction factor due to the nonlinear hysteretic behavior
is dened as the ratio between the elastic and inelastic strength
demand, and is equal to 0.17. Finally, the seismic modication factor is determined as the product between the deection amplication factor, the over strength factor, and the strength reduction
factor. Consequently, in the present case, the seismic modication
factor is equal to 3.7.
It should be noted (1) that the seismic modication factor has
been determined based on capacity curve shown in Fig. 21,
whereas (2) the failure pattern corresponding to point C of the

291

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

Point C

1600

Base shear force (kN)

1400
1200
1000

Point B

Point D

800
performance level
predicted by FEMA
guidelines in [50]

600
400

Point A

DSC/HISS-CT
Equivalent frame by Akhaveissy [50]
Discrete element method [60]

200
0
0

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Roof displacement (mm)


Fig. 21. Building with ve stories. Loaddisplacement curve, unreinforced case. (See above-mentioned references for further information.)

[mm]

(a) roof displacement: 1.30 mm

(b) roof displacement: 9.39 mm

(c) roof displacement: 16.04 mm

(d) roof displacement: 73.90 mm

Fig. 22. Disturbance parameter patch at different global deformation level in the pushover curve. (a): Point A. (b): Point B. (c) Point C. (d) Point D in Fig. 21.

capacity curve, Fig. 22c, clearly shows a failure mechanism involving the ground oor. Therefore, it can be stated that the building
exhibits an inadequate behavior when dealing with the Life Safety
performance level. In addition, since the slope of capacity curve between points C and D is negative, it is recommended to consider
the capacity curve up to point C to determine the seismic modication factor. In this case, such factor is equal to 1.5, basing on
the capacity curve up to a displacement equal to 16 mm, point C
in Fig. 21. The seismic modication factor is hence equal to 1.5
and is related to the Life Safety performance level. The old masonry

building was also analyzed by an equivalent frame method [50].


The predicted performance level based on FEMA guidelines was
evaluated between the life safety and the collapse prevention level
[50]. This level corresponds to a displacement of the roof equal to
49.96 mm, Fig. 21. The drift values for the piers vary from 0.0033 to
0.0037 [50]. This means that the building can tolerate larger displacements up to a collapse prevention drift equal to 0.004. As a
consequence, it can be concluded that the collapse state predicted
by the DSC/HISS-CT model in Fig. 22c and d correlate well with the
life safety and the collapse prevention level of FEMA guidelines.

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

26515.0
27727.8
28911.2

25302.1

22750.0
23950.0

strengthened case. It is noticeable that Iranian earthquake code


recommends a seismic modication factor for a concrete frame
with low exibility equal to 5 [48]. No indications are given for
reinforced masonry structures. As a consequence, it can be stated
that such reinforcement, which is associated to a factor equal to
6.7, appears to be acceptable for a proper strengthening
intervention.
A detailed study of the crack patterns developing, in the reinforced case, at different roof displacements of the building is useful
to have an insight into the behavior of the building during a possible earthquake excitation.
They are depicted in Fig. 25, respectively for displacements
equal to 2.23, 9.84 and 79 mm. Such displacement values correspond to points A, B and C in Fig. 24, respectively and can be compared with the failure patterns of the unreinforced building
corresponding to similar displacements. Such comparisons show
that failure patterns of the reinforced and unreinforced building
are quite similar, meaning that the increase of the load bearing
capacity is exclusively due to the contribution of the steel bars
strength and that the failure mechanism remains essentially
unchanged.
In particular, Fig. 25c shows the disturbance parameter patch at
a displacement equal to 79 mm, i.e. near the formation of a failure
mechanism.

18456.2
19281.2
20518.7
21343.7

12741.0

11523.0

6272.5
7735.0
8905.0
10075.0

4967.5

205.5
1438.9
2672.2

The above considerations allow concluding that the building


has low seismic modication factor and hence great vulnerability.
For this reason, a strengthening intervention is recommended to
increase the resistance of the building. Different technologies
may be used at this aim. Hereafter, it is supposed to introduce vertical steel bars with variable distance as depicted in Fig. 23. Bars
are introduced in the homogeneous DSC/HISS-CT model by means
of elasto-plastic truss elements (yield strength equal to 300 MPa),
interconnected node by node to the 2D plane stress elements representing masonry. The diameter of the bars is equal to 10 mm,
with a distance between single bars that is globally in agreement
with the Iranian earthquake code [48], which requires the utilization of vertical bars at distance ranging from 600 to 1200 mm. No
horizontal bars are used.
The capacity curve of the strengthened masonry building is
compared to the corresponding capacity curve of the unreinforced
structure in Fig. 24.
Fig. 23 shows that the ultimate base shear obtained for the
strengthened structure is equal to 2680 kN. The utilization of 22
lines of vertical steel bars causes an increase of the ultimate base
shear of about 81%. In the reinforced case, the seismic modication
factor is equal to 6.7 and it is determined basing on the capacity
curve reported in Fig. 24. Hence, the seismic modication factor increases from 1.5 for the unreinforced building to 6.7 for the

16133.3
17026.7

292

Fig. 23. Building with ve stories. Location of hypothetical reinforcing steel bars (dimensions in millimeters).

2750
2500

Point C

Base shear force (kN)

2250
2000

Point B

1750
1500
1250
1000

RM (DSC/HISS-CT)

750

URM (DSC/HISS-CT)

500

URM by Discrete Element Method [60]

250

Point A

URM by Equivalent Frame [50]

0
0

10

20

30

40

50

60

70

80

Displacement at roof (mm)


Fig. 24. Comparison among the capacity curves of the unreinforced masonry building and the capacity curve of the strengthened building.

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

293

[mm]

(a) roof displacement: 2.23 mm

(b) roof displacement: 9.84 mm

(c) roof displacement: 79.00 mm


Fig. 25. Building with ve stories, reinforced case. Disturbance parameter patch at different global deformation levels in the pushover curve. (a) Point A. (b) Point B. (c) Point
C in Fig. 24.

294

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295

Fig. 26. Yielded steel bars at a roof displacement equal to 79 mm.

As can be noted, severe damage is present in almost all the piers


belonging to the ground oor, especially those positioned on the
right. Hence, it may be deduced that a failure mechanism activates
if the steel bars at the ground oor level are also yielded. In Fig. 26,
yielding of the bars at 79 mm of lateral roof displacement is represented with thick lines. As expected, Fig. 26 conrms that bars at
the ground oor are almost completely in plastic state, meaning
that the reinforced building is prone to collapse.

models or computationally expensive heterogeneous approaches,


for a fast evaluation of the safety level of complex masonry structures. Authors experienced a substantial reduction of the computational cost connected to the utilization of the DSC/HISS-CT model,
especially in comparison with full smeared crack 3D approaches
available in the most diffused commercial codes (e.g. SOLID65 in
ANSYS).
References

4. Conclusions
A simple, fully 2D macroscopic FE model for the pushover analysis of real scale masonry structures in-plane loaded has been
presented.
The so called disturbed state concept (DSC) combined with a
modied hierarchical single surface (HISS-CT) plasticity model
with associated ow rules were used to characterize the compressive and tensile behavior of the masonry material. The model can
easily account for micro-cracks in the masonry, which are responsible for softening and fracture. Two separate yield surfaces are
present in the model, for a proper independent description of the
compressive and tensile behavior. The continuous nature of the
proposed yield surface avoids computational difculties present
in available discontinuous or multiple surfaces models, such as
critical state and cap models [24]. The DSC model allows for the
generation of discontinuities in the material microstructure during
loading (unloading), and it does not require external enrichments
to allow coupling between continuous and discontinuous parts
within the deforming material [62]. Moreover, the same macroscopic model can be easily adapted to perform heterogeneous analyses at a structural level, modeling separately bricks (elastic) and
joints reduced to interfaces [24].
After a preliminary standard validation of the DSC/HISS-CT
model for masonry at a specimen level, the model has been embedded into a non-linear 2D FE code, to analyze real scale masonry
structures subjected up to failure to incremental horizontal loads.
Three structures of technical interest, namely two two-story and
one ve-story large scale walls have been analyzed in detail and
results have been compared with those provided by alternative
procedures, including equivalent frame approaches and heterogeneous 2D models.
For all the cases analyzed, pushover curves obtained with the
model proposed result in very good agreement with existing procedures, meaning that the fully 2D approach proposed may represent a valuable tool, alternative to simplistic equivalent frame

[1] NTC 2008, DM 2008. Nuove norme tecniche per le costruzioni. Ministero delle
Infrastrutture (GU n.29 04/02/2008), Rome, Italy, 14/01/2008 and Circolare 2
febbraio 2009, n. 617 Istruzioni per lapplicazione delle Nuove norme
tecniche per le costruzioni di cui al decreto ministeriale 14 gennaio; 2008.
[GU n. 47 del 262-2009 - Suppl. Ordinario no.27].
[2] OPCM, no.3274 Ordinanza del Presidente del Consiglio dei Ministri n.3274/
2003. Primi elementi in materia dicriteri generali per la classicazione sismica
del territorio nazionale e dinormative tecniche per le costruzioni in zona
sismica. Supplemento ordinario G. U. n. 105, 8 maggio; 2003 [in Italian].
[3] Loureno PB, Mendes L, Ramos LF, Oliveira DV. Analysis of masonry structures
without box behavior. Int J Architect Heritage 2011;5:36982.
[4] Grande E, Imbimbo M, Sacco E. Bond behaviour of CFRP laminates glued on
clay bricks: experimental and numerical study. Compos B Eng 2011;42(2):
33040.
[5] Milani G, Beyer K, Dazio A. Upper bound limit analysis of meso-mechanical
spandrel models for the pushover analysis of 2D masonry frames. Eng Struct
2009;31:2696710.
[6] Belmouden Y, Lestuzzi P. An equivalent frame model for seismic analysis of
masonry and reinforced concrete buildings. Constr Build Mater 2009;23:
4053.
[7] Roca P, Molins C, Mar AR. Strength capacity of masonry wall structures by the
equivalent frame method. J Struct Eng ASCE 2005;131(10):16019.
[8] Galasco A, Lagomarsino S, Penna A, Resemini S. Non-linear Seismic Analysis of
Masonry Structures. In Proceedings of 13th World Conference on Earthquake
Engineering, Vancouver 16 August, paper no. 843; 2004.
[9] Chen SY, Moon FL, Yi T. A macroelement for the nonlinear analysis of in-plane
unreinforced masonry piers. Eng Struct 2008;30(8):224252.
[10] Pasticier L, Amadio C, Fragiacomo M. Non-linear seismic analysis and
vulnerability evaluation of a masonry building by means of the SAP2000 V.
10 code. Earthquake Eng Struct Dynam 2008;37:46785.
[11] Magenes G, Della Fontana A. Simplied non-linear seismic analysis of masonry
buildings. Proc Br Masonry Soc 1998;8:1905.
[12] Salonikios T, Karakostas C, Lekidis V, Anthoine A. Comparative inelastic
pushover analysis of masonry frames. Eng Struct 2003;25:151523.
[13] Akhaveissy AH. Lateral strength force of URM structures based on a
constitutive model for interface element. Latin Am J Solids Struct 2011;8:
44561.
[14] Loureno PB. Computational strategies for masonry structures. Dissertation: Delft
University; 1996.
[15] Akhaveissy AH, Desai CS. Unreinforced masonry walls: nonlinear nite
element analysis with a unied constitutive model. Arch Comput Methods
Eng 2011;18(4):485502.
[16] Lourenco PB, de Borst R, Rots JG. A plane stress softening plasticity model for
orthotropic materials. Int J Numer Meth Eng 1997;40:403357.
[17] Desai CS, Somasundaram S, Frantziskonis G. A hierarchical approach for
constitutive modeling of geologic materials. Int J Numer Anal Methods
Geomech 1986;10:22557.

A.H. Akhaveissy, G. Milani / Construction and Building Materials 41 (2013) 276295


[18] Desai CS, Salami MR. A constitutive model for rocks. J Geotech Eng ASCE
1987;113:40723.
[19] Morbiducci R. Nonlinear parameter identication of models for masonry. Int J
Solids Struct 2003;40:407190.
[20] Feenstra PH, de Borst R. A plasticity model and algorithm for mode- I cracking
in concrete. Int J Numer Methods Eng 1995;38:250929.
[21] Loureno PB, Rots JG, Feenstra PH. A tensile Rankine type orthotropic model
for masonry. In: Computer methods in structural masonry 3. Pande GN,
Middleton J, editors.Books & Journals International, Swansea, UK; 1995.
[22] Desai CS, Toth J. Disturbed state constitutive modelling based on stressstrain
and nondestructive behavior. Int J Solids Struct 1996;33:161950.
[23] Salami MR, Desai CS. Constitutive modeling including multiaxial testing for
plain concrete under low conning pressure. J Mater Am Concr Inst
1990;87(3):22836.
[24] Desai CS. Mechanics of materials and interfaces: the disturbed state
concept. Boca Raton (Florida, USA): CRC Press; 2001.
[25] Desai CS, Hashmi QSE. Analysis, evaluation and implementation of a
nonassociative model for geologic materials. Int J Plasticit 1989;5:397420.
[26] Frantziskonis G, Desai CS, Somasundaram S. Constitutive model for
nonassociative behavior. J Eng Mech ASCE 1986;112:93246.
[27] Somasundaram S, Desai CS. Modelling and testing for anisotropic behavior of
soils. J Eng Mech ASCE 1988;114:147396.
[28] Desai CS, Galagoda HM. Earthquake analysis with generalized plasticity model
for saturated soils. J Earthquake Eng Struct Dyn 1989;18(6):90319.
[29] Wathugala GW, Desai CS. Constitutive model for cyclic behavior of cohesive
soils. I: Theory. J. Geotech. Eng., ASCE 1993; 119 (4): 71429.
[30] Desai CS, Zhang D. Viscoplastic model with generalized yield function. Int. J
Numer Methods Geomech 1987;11:60320.
[31] Desai CS, Samtani NC, Vulliet L. Constitutive Modeling and Analysis of
Creeping Slopes. J. Geotech. Eng., ASCE 1995; 121(1): 4356.
[32] Samtani NC, Desai CS, Vulliet L. An interface model to describe viscoplastic
behavior. Int J Numer Analyt Methods Geomech 1996;20(4):23152.
[33] Katti DR, Desai CS. Modelling and testing of cohesive soil using the disturbed
state concept. J Eng Mech ASCE 1994;121:64858.
[34] Milani G, Lourenco PB, Tralli A. Homogenised limit analysis of masonry walls,
Part I: Failure surfaces. Comput Struct 2006;84(34):16680.
[35] Desai CS. Unied DSC constitutive model for pavement materials with
numerical implementation. Int J Geomech ASCE 2007;7(2):83101.
[36] Zienkiewicz OC, Valliapan S, King IP. Stress analysis of rock as a no tension
material. Geotechnique XVIII(1); 1968.
[37] Akhaveissy AH, Desai CS. Application of DSC model for nonlinear analysis of
reinforced concrete frames. Finite Elem Anal Des 2012;50:98107.
[38] Kaushik HB, Rai DC, Jain SK. Stressstrain characteristics of clay brick masonry
under uniaxial compression. J Mater Civ Eng ASCE 2007;19(9):72839.
[39] Pandey BH, Meguro K. Simulation of Brick masonry wall behavior under in
plane lateral loading using applied element method. In Proceedings of 13th
world conference on earthquake engineering. Vancouver, BC, Canada, August
16, paper no. 1664; 2004.
[40] Weibull WA. A statistical distribution function of wide applicability. Appl
Mech 1951;18:2937.
[41] Kaushik HB, Rai DC, Jain SK. Uniaxial compressive stressstrain model for clay
brick masonry. Curr Sci 2007;92(4):497501.
[42] Yalcin C, Saatcioglu M. Inelastic analysis of reinforced concrete columns.
Comput Struct 2000;77:53955.
[43] Akhaveissy AH, Desai CS, Sadrnejad SA, Shakib H. Implementation and
comparison of generalized plasticity and disturbed state concept for load-

[44]

[45]

[46]

[47]
[48]

[49]

[50]
[51]

[52]
[53]
[54]

[55]
[56]
[57]

[58]

[59]

[60]

[61]

[62]

295

deformation behavior of foundation. Scientia Iranica J Trans A: Civ Eng


2009;16(3):18998.
Thanoon WA, Alwathaf AH, Noorzaei J, Saleh Jaafar M, Abdulkadir MR.
Nonlinear nite element analysis of grouted and ungrouted hollow
interlocking mortar less block masonry system. Eng Struct 2008;30:156072.
Brencich A, Gambarotta L, Lagomarsino S. A macroelement approach to the
three-dimensional seismic analysis of masonry buildings. In Proceeding of
11th European conference on earthquake engineering. Balkema, Rotterdam.
The Netherlands; 1998.
Braga F, Liberatore D, Spera G. Acomputer program for the seismic analysis of
complex masonry buildings. In Proceedings of 4th International Symposium
on computer methods in structural masonry. Pande GN, Middleton J, Kralj B,
editors. Florence, Italy, September 1997; 35: 309316.
Milani G, Lourenco PB, Tralli A. Homogenised limit analysis of masonry walls,
Part II: Structural examples. Comput Struct 2006;84(34):18195.
Iranian code of practice for seismic resistant design of buildings. Standard
number 2800-84 (2007). 3rd edition. Building and Housing Research Center.
Tehran, Iran.
Demirel IO. A nonlinear equivalent frame model for displacement based
analysis of unreinforced brick masonry buildings. PhD Thesis, Middle East
Technical University, Turkey; 2010.
Akhaveissy AH. Finite element nonlinear analysis of high-rise unreinforced
masonry building. Latin Am J Solids Struct 2012;9:54767.
Federal Emergency Management Agency (FEMA 306). Evaluation of
earthquake damage concrete and masonry wall buildings. basic procedure
manual, ATC-43, FEMA306. Applied Technology Council. California; 1999.
Canadian Standard Association (CSA). Design of masonry structures. S304.1.
Ontario, Canada; 2004.
International Code Council. International building code; 2003.
Masonry Standard Joint Committee (MSJC) Building code requirements for
masonry structures. ACI 53002/ASCE 502/TMS 4022, American Concrete
Institute, Structural Engineering Institute of the American Society of Civil
Engineers. The Masonry Society, Detroit; 2002.
ATC40. Seismic evaluation and retrot of concrete buildings, Volume 1,
Applied Technology Council (ATC); 1996
FEMA 356. NEHRP guidelines for the seismic rehabilitation of
buildings. Washington (DC): Federal Emergency Management Agency; 2000.
Liberatore D. Progetto Catania: Indagine sulla risposta sismica di due edici in
muratura (Catania project: investigation on the seismic response of two
masonry buildings). GNDT National Group for Seismic Protection: Rome,
Italy; 2000. [in Italian].
Gambarotta L, Lagomarsino S. Damage models for the seismic response of
brick masonry shear walls. Part II: The continuum model and its application.
Earthquake Eng Struct Dynam 1997;26(4):44162.
Braga F, Liberatore D. A nite element for the analysis of the response of
masonry buildings. In Proc.: 5th North American Masonry Conference, Urbana,
USA; 1990.
Brencich A, Gambarotta L, Lagomarsino S. Analysis of a masonry building in via
martoglio. pages 107143. Catania Project: Research on the seismic response
of two masonry buildings, [chapter 6], University of Genoa [in Italian], CNR
Gruppo Nazionale per la Difesa dei Terremoti; 2000.
Newmark NM, Hall WJ. Seismic design criteria for nuclear reactor facilities.
Report #46. Building Practices for Disaster Mitigation, National Bureau of
Standards, USA Department of Commerce, 1973; 209-236.
Muhlhaus HB. Continuum models for materials with microstructure.
Chichester, UK, John Wiley & Sons; 1995.

Вам также может понравиться