You are on page 1of 13

Computers and Structures xxx (2018) xxx–xxx

Contents lists available at ScienceDirect

Computers and Structures


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

A 3D detailed micro-model for the in-plane and out-of-plane numerical


analysis of masonry panels
Antonio Maria D’Altri a,⇑, Stefano de Miranda a, Giovanni Castellazzi a, Vasilis Sarhosis b
a
Department of Civil, Chemical, Environmental, and Materials Engineering (DICAM), University of Bologna, Viale del Risorgimento 2, Bologna 40136, Italy
b
School of Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, UK

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a novel 3D detailed micro-model to analyse the mechanical response of masonry panels
Received 30 January 2018 under in-plane and out-of-plane loading conditions is proposed. The modelling approach is characterized
Accepted 11 June 2018 by textured units, consisting of one brick and few mortar layers, represented by 3D solid finite elements
Available online xxxx
obeying to plastic-damage constitutive laws. Textured units are assembled, accounting for any actual 3D
through-thickness arrangement of masonry, by means of zero-thickness rigid-cohesive-frictional inter-
Keywords: faces, based on the contact penalty method and governed by a Mohr-Coulomb failure surface with ten-
Masonry
sion cut-off. This novel approach can be fully characterized by the properties obtained on small-scale
Cohesive interfaces
Micro-modelling
experimental tests on brick and mortar and on small masonry assemblages. The interface behaviour
Plastic-damage model appears consistent with small-scale tests outcomes on masonry specimens. Experimental-numerical
Cracking comparisons are provided for the in-plane and out-of-plane behaviour of masonry panels. The accuracy,
Crushing the potentialities and the efficiency of the modelling approach are shown and discussed.
Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction strategies for a suitable description of the mechanical behaviour


of masonry is still nontrivial and an on-going process in the scien-
Masonry is one of the oldest building materials. It is composed tific research [5].
of masonry units (i.e. brick, blocks, etc.) usually bonded with mor- Generally, computational strategies for masonry structures are
tar. Due to its heterogeneous and composite nature, its mechanical classified in micro-modelling and macro-modelling [6]. In addition,
behaviour is extremely complex. The near-collapse mechanical homogenization and upscaling procedures represent a link
behaviour of masonry structures is generally deeply influenced between the two approaches [7–12]. The macro-modelling
by the failure of brick-mortar bonds, which act as planes of weak- approaches account for the masonry mechanical nonlinearity by
ness [1]. Indeed, the brick-masonry interface represent a disconti- means of a macroscopic continuum description of its behaviour,
nuity between two distinct and different materials and its strength, employing different formulations (e.g. phenomenological plasticity
which depends from a huge number of factors (e.g. brick pores [13], damage mechanics [14] and nonlocal damage-plasticity [15]).
dimensions, units moisture content, nature of micro-layer of Isotropic continuum nonlinear constitutive laws with softening
ettringite, compaction of the mortar, etc. [1]) and, therefore, is have been successfully used for the analysis of large-scale historic
extremely variable, is generally considerably smaller than the mor- structures [16], where, due to the chaotic and random texture of
tar and unit one [2]. Under extreme loading conditions (e.g. earth- historic masonry, the hypothesis of isotropic material generally
quakes), masonry structures can show cracking and/or crushing of appears suitable. Nonetheless, when dealing with masonries char-
the units too. acterized by well-organized and periodic masonry textures, the
Due to these features, as well as the difficulties in characterizing hypothesis of isotropic material is no longer suitable. To overcome
the masonry mechanical properties of existing structures (espe- this issue, few masonry macro-modelling approaches have been
cially if they are historic [3]), the evaluation of the vulnerability extended to orthotropic continua [17,18]. Furthermore, phe-
of masonry buildings by means of deterministic numerical models nomenological continuum models accounting for the micro-
is still challenging [4]. Indeed, although significant advances have structure of masonry have been recently developed (the so-called
been carried out in the last decades, the definition of numerical continuous micro-models, see for example [19]).
However, an account of the inelastic response over discontinu-
⇑ Corresponding author. ity surfaces at the brick-mortar bonds appears to be crucial in the
E-mail address: antoniomaria.daltri2@unibo.it (A.M. D’Altri). analysis of masonry structures. Indeed, the behaviour of masonry

https://doi.org/10.1016/j.compstruc.2018.06.007
0045-7949/Ó 2018 Elsevier Ltd. All rights reserved.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
2 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

walls is largely affected by the displacement discontinuities which behaviour is defined by an exponential coupled cohesive behaviour
are generated at the brick-mortar interfaces, as experimentally evi- in tension and a cohesive-frictional behaviour in shear, accounting
denced in [20]. for the brick-mortar bond failure both in tension and shear.
Although their larger computational demand, micro-models To the author knowledge, the coupling of contact-based rigid-
with interface elements can capture the complex patterns of dis- cohesive interfaces with 3D nonlinear-damaging textured units
continuities which characterize the damage evolution in masonry (which explicitly account for the mortar layers) to model masonry
with a higher degree of accuracy, and reproduce the main features is a novelty in the scientific literature. This novel modelling
of their response, such as, for example, the relative sliding of units. approach can, in fact, be fully characterized by the properties
For these reasons interface elements found broad application in the obtained on small-scale specimen tests on brick and mortar (stiff-
numerical analysis of masonry structures [21–30] and they are still ness, compressive and tensile responses) and on small masonry
currently object of investigation [31,32]. Discrete element models assemblages (tensile and shear responses of the mortar-brick
(DEM) represent a further numerical strategy, utilized to analyse bond).
the mechanical behaviour of systems made of particles, blocks or To reach this goal, this paper introduces an interface model. The
multiple bodies, which appears suitable for masonry structures interface behaviour is governed by an ad-hoc modification of the
[33–39]. standard surface-based contact behaviour implemented in Abaqus
Nevertheless, these micro-modelling approaches present some [41], a general-purpose FE software. Contextually, an automatic
criticalities. One the one hand, DEM approaches do not generally subroutine ad-hoc written by the authors is implemented to repro-
account for masonry crushing, making this modelling strategy duce a Mohr-Coulomb failure surface with tension cut-off.
more suitable for analysing dry-joint masonry or low bond The interfacial behaviour appears consistent with experimental
strength masonry, where failure occurs in the mortar or in the outcomes on small-scale masonry specimens. Experimental-
brick-mortar interface rather than in the units [40]. On the other numerical comparisons are provided for the in-plane and out-of-
hand, most of the existing micro-models in the literature concern plane behaviour of masonry panels. The direct characterization of
linear elastic units and joints which can simulate the sliding, crack- all the model mechanical properties from small-scale tests on
ing and crushing of masonry (e.g. all the models based on the mul- brick, mortar and brick-mortar bond and their clear mechanical
tisurface interface model proposed in [21]). Particularly, the meaning constitute an appealing quality of the model proposed.
crushing is usually accounted for by means of a cap in the joint fail- The paper is organized as follows. Section 2 illustrates the main
ure surface, i.e. through a phenomenological representation of the features of the modelling approach proposed. Section 3 describes
crushing. However, the characterization of the compressive nonlin- the brick-mortar interface nonlinear behaviour. Section 4 describes
ear behaviour of masonry is not an easy task. Indeed, it depends on the plastic-damage model utilized for brick and mortar. Section 5
the texture of masonry, on the direction of the compressive load collects experimental-numerical comparisons and their discussion
(e.g. perpendicular to the bed joints, parallel to the bed joints, for the in-plane and out-of-plane behaviour of masonry panels.
etc.), on the relative dimensions between bricks and mortar joints Finally, Section 6 highlights the conclusions of this research work.
[11], etc. Moreover, a reliable characterization of the compressive
behaviour of masonry should be based on tests on relatively large
specimens. Conversely, the characterization of the single materials 2. Modelling approach
(mortar and brick) in compression appear easier and dependent on
less variables. As already mentioned, several modelling strategies can be fol-
In this context, the development of a novel model whose lowed to analyse masonry structures, see Fig. 1. An accurate model
mechanical setting could be exclusively based on small-scale spec- for simulating the mechanical behaviour of masonry should
imen tests of masonry components (i.e. mortar and brick) and account for the main masonry failure mechanisms [21]. At a small
small masonry assemblages, without using spread mechanical scale, masonry failures are depicted in Fig. 2. In particular, brick-
properties, such as the masonry compressive strength, was consid- mortar interface tensile failure (Fig. 2a) and shear sliding
ered. Furthermore, the idea of developing a 3D solid model able to (Fig. 2b) are characterized by the failure of the bond between brick
account for, at the same time, the in-plane and out-of-plane and mortar. Masonry crushing (Fig. 2d), cracking (Fig. 2e) and diag-
response of masonry elements (since, in practice, they can be cou- onal cracking (Fig. 2c) are, instead, combined mechanisms involv-
pled) was also contemplated. ing bricks and mortar (Fig. 2d–e) and bricks, mortar and brick-
To pursue this goal, a novel numerical approach to model mortar interface (Fig. 2c).
masonry is conceived. In particular, a 3D detailed micro-model In the modelling approach herein proposed, the brick-mortar
for the in-plane and out-of-plane numerical analysis of masonry bond failures (Fig. 2a and b) are accounted for by brick-mortar
structures is proposed in this paper. In this modelling approach, nonlinear cohesive interfaces, whereas the combined mechanisms
textured units consisting of one brick and few mortar layers are involving also brick and mortar (Fig. 2c–e) are accounted for by the
explicitly modelled using 3D solid Finite Elements (FEs) obeying nonlinear behaviour of brick and mortar FEs, see Fig. 1b. Therefore,
to plastic-damage constitutive laws conceived in the framework brick and mortar crushing and cracking, although characterized by
of nonassociated plasticity. Particularly, two plastic-damage mod- a complex evolution of micro-cracks, are represented by the inelas-
els with distinct parameters are assumed for brick and for mortar, tic behaviour of brick and mortar FEs.
both in tension and compression regimes. This permits to repre- Textured units composed of 3D solid FEs (Fig. 3) with brick
sent the brick and mortar behaviour when cracking and/or crush- properties (red elements in Fig. 3) and mortar properties (grey ele-
ing occur. ments in Fig. 3) are conceived and they are assembled by means of
Textured Units are assembled, accounting for any actual 3D zero-thickness interfaces (green surfaces in Fig. 3). For single leaf
through-thickness arrangement of masonry, by means of masonry panels, the textured unit concerns one brick as well as
zero-thickness cohesive-frictional interfaces based on the contact one head joint and one bed joint (Fig. 3). Brick and mortar FEs
penalty method. In the pre-failure interfacial behaviour, all the sig- are characterized by distinct nonlinear plastic-damaging beha-
nificant linear elastic deformability of the system is addressed to viour, both in tension and compression regimes.
the 3D brick and mortar FEs, being negligible the interfacial defor- Each mortar layer is continuously linked to a brick and sepa-
mations. The interfaces are characterized by a Mohr-Coulomb fail- rated by an interface from other bricks. This reduces considerably
ure surface with tension cut-off. The post-failure interfacial the number of interfaces (instead of considering all the two

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 3

Fig. 1. Modelling strategies for masonry structures (following [6,19]): (a) masonry sample, (b) detailed micro-modelling, (c) continuous micro-modelling, (d) discrete micro-
modelling and (e) macro-modelling.

Fig. 2. Masonry failure mechanisms (following [21]): (a) brick-mortar interface tensile failure, (b) brick-mortar interface shear sliding, (c) diagonal masonry cracking, (d)
masonry crushing and (e) brick and mortar tensile cracking.

interfaces of a mortar layer), and therefore the computational cost micro-structure, geometrical imperfections, etc.). Experimental
of the model, without compromising the model accuracy. Indeed, characterizations of dilatancy by van der Pluijm et al. [46] show
the fact that a brick-mortar bond failure occurs in the upper or that the dilatancy ratio is significantly influenced by the type of
lower bond of a mortar layer does not affect the mechanical interface failure. Particularly, the magnitude of dilatancy turns
response of masonry. out to be substantially higher when the crack crosses mortar
Contact penalty method is enforced in the zero-thickness inter- (and/or units), compared to the dilatancy measured when detach-
faces between the textured units. Traditional point-against-surface ment of the brick-mortar interfaces occurs (bond failure), which is
contact method is considered [42]. The penalty stiffness is assumed considerably smaller.
to keep insignificant the penetration of the elements and to guar- In the modelling approach herein proposed, zero-thickness
antee good convergence rates of simulations (compared, for exam- interfaces are conceived without a dilatant behaviour, whereas
ple, with Lagrange multipliers methods [42]). In this study, penalty dilatancy is considered in the 3D nonlinear FEs in the framework
stiffness is assumed to be equal to 500 times the representative of nonassociated plasticity [47]. This approach, although simpli-
stiffness of underlying elements. In the pre-failure of interfaces, fied, appears to be consistent with the experimental outcomes
all the significant deformability of the system is addressed to the pointed out in [46], i.e. significant dilatant behaviour only occurs
3D FE part. when mortar (and/or units) undergoes failure.
Dilatancy plays an important role in the mechanical behaviour The main idea at the base of the setting of the parameters is that
of masonry [43], although it is still currently object of investigation the properties of the interface are based on brick-mortar bond tests
and debate [44,45], and its characterization is complex as it is (tensile failure and shear sliding), whereas the properties of the
influenced by several mechanical factors (e.g. materials mortar and brick FEs are based on tests on the single components.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
4 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

Fig. 3. Detailed micro-modelling approach. An example of textured unit mesh is given in the picture.

Although the experimental data available makes non-trivial the sile strength f t of the interface is reached, see Fig. 4a. As can be
separation of the two problems, this assumption, in the Authors noted in Fig. 4a, penetration can occur between elements. However,
opinion, appears reasonable and leads to a rationally easy setting although no procedures to remove penetration have been imple-
of the parameters. mented, by using quite high penalty stiffnesses (i.e. equal to 500
times the stiffness of the underlying elements) the penetration
3. Brick-mortar interface behaviour between elements has been found negligible. Furthermore, the pen-
alty stiffness adopted has been found a good compromise between
In the normal direction, the contact stress r is computed by convergence and accuracy (i.e. negligible penetration).
means of the linear relationship: In the shear direction, the tangential slip d is linearly related to
the interface shear stress with the relation:
r ¼ knpenalty u; ð1Þ
n
s ¼ kspenalty d; ð2Þ
where kpenalty
is the penalty stiffness in normal direction and u is the
s
normal displacement. Through the contact penalty method, this where kpenalty is the penalty stiffness in shear. This relation is valid
relation is assumed to be valid also for tensile stresses until the ten- until the shear stress equals the shear strength f s , see Fig. 4b. The

Fig. 4. Interfacial pre-failure behaviour: (a) normal behaviour and (b) shear behaviour.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 5

uMAX
shear strength f s of the interface is assumed to be dependent on the f
1  e uk
contact stress: Q¼ ; ð6Þ
1  ef
f s ðrÞ ¼  tan /r þ c; ð3Þ
being f a non-dimensional brittleness parameter and uMAX the max-
where tan / and c are parameters experimentally defined. imum separation ever experienced by the contact point. The cohe-
Interface failure occurs, i.e. the process of degradation begins, sive behaviour is only activated for tension, whereas for pure
when the contact stresses at a point satisfy a failure criterion. Par- compression stress states no failure is considered at the interfacial
ticularly, failure is supposed when the maximum contact stress level (see Fig. 5).
ratio intersects a Mohr-Coulomb failure surface with tension cut- Concerning the shear behaviour, when the shear stress s
off. This simple criterion can be expressed as: reaches the shear strength f s ðrÞ, a simplified cohesive-frictional
  behaviour is activated, and the contacting surfaces start sliding.
hri s
max ; ¼ 1; ð4Þ After failure the shear stress is composed of a cohesive term
f t f s ðrÞ ð1  HÞf s ðrÞ and a frictional one Hlhri (Fig. 6b), according to
where the symbol hxi ¼ ðjxj þ xÞ=2 denotes the Macaulay bracket the relationship:
function. The Macaulay brackets are used to signify that a purely 
ð1  HÞf s ðrÞ þ Hlhri; d < dk
compressive stress state does not induce interfacial failure. A sketch s¼ ; ð7Þ
of the failure surface adopted for the interfacial behaviour is shown lhri; d P dk
in Fig. 5. Once failure of the interface is reached, cohesive behaviour where dk is the ultimate slip of the cohesive behaviour, l is the fric-
in tension and cohesive-frictional behaviour in shear is activated. tional coefficient and H is an exponential scaling function defined
After reaching tensile strength f t , an interfacial cohesive beha- as:
viour is activated in normal direction and the stress r decreases
with an increasing separation u, while at u ¼ uk stress ends to be
dMAX
n
1  e dk
transmitted, see Fig. 6a. The stress follows the relationship: H¼ ; ð8Þ
1  en

ð1  Q Þf t ; u < uk
r¼ ; ð5Þ being n a non-dimensional brittleness parameter and dMAX the max-
0; u P uk imum slip ever experienced by the contact point.
where Q is an exponential scaling function defined as: It has to be pointed out that the two variables Q and H are
forced to assume the same value at any step of the analysis
(Q ¼ H). This means that the damage evolution of Mode I and
Mode II are fully coupled. Therefore, the degradation of cohesion
in tension degrades the cohesion in shear and vice versa. Although
this adoption can be considered approximated, it is, however, more
realistic than considering independent the two phenomena. In par-
ticular, the two variables Q and H can increase from 0 to 1 only.
Indeed, the degradation of the cohesion is an irreversible process.
This model is, in general, not restricted to the monotonic beha-
viour. The degradation of cohesion is an irreversible process and
once the maximum degradation has been reached, the cohesive
contribution to the tensile and shear stresses is zero, and the only
contribution to the shear stresses is from the frictional term.
The interface behaviour is based on large displacements. In par-
ticular, the finite-sliding tracking approach implemented in Aba-
qus [41], which allows for arbitrary separation, sliding, and
rotation of the surfaces, is adopted.

3.1. Comparison between experimental and numerical results for


small-scale masonry specimens

Experimental tests conducted by van der Pluijm in [2,43] on


Fig. 5. Interfacial failure surface: Morh-Coulomb surface with tension cut-off (s1 small-scale masonry specimens, composed of two bricks jointed
and s2 are the shear stress components along two orthogonal directions in the plane
together by a mortar joint, were used as reference to compare with
of the interface).
numerical outcomes and to tune the brittleness parameters f and n.

Fig. 6. Interfacial post-failure behaviour: (a) tensile response and (b) shear response.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
6 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

As in [2,43] the tensile and shear failures were only observed in the Two independent scalar damage variables, one for the tensile
brick-mortar interfaces, linear elastic behaviour for brick and mor- regime (0 6 dt < 1) and one for the compressive regime
tar has been assumed. The mechanical properties adopted in the (0 6 dc < 1), are supposed. Accordingly, the stress-strain relations
numerical simulations are collected in Table 1. Fig. 7 shows the under uniaxial tension, rt , and compression, rc , are:
comparison between experimental and numerical results for small
scale masonry specimens subjected to tension (Fig. 7a) and shear rt ¼ ð1  dt ÞE0 ðet  ept Þ; rc ¼ ð1  dc ÞE0 ðec  epc Þ; ð9Þ
(Fig. 7b).
The tensile properties of the interface are assumed to be consis- where E0 is the initial Young’s modulus of the material, et and ec are
tent with the fracture energy of the brick-mortar interface in ten- the uniaxial tensile and compressive strains, and ept and epc are the
sion (Mode I), which in [2] is equal to Gint uniaxial tensile and compressive plastic strains (Fig. 8). Particularly,
I ¼ 12:0 N/m. Indeed, once
the tensile strength f t and the displacement uk are fixed, which can the curves depicted in Fig. 8 represent the main input data of the
be defined directly from the experimental envelope (Fig. 7a), the model.
brittleness parameter f is chosen so that the area under the curve Mesh objectivity in the softening branch passes through an indi-
rect definition of the fracture energy, i.e. the model is local, and reg-
in Fig. 6a equals Gint
I .
ularization occurs scaling the fracture energies by means of the
Analogously, the shear properties of the interface are assumed pffiffiffiffiffiffi P P P 
nq nn ng
to be consistent with the Mode II-fracture energy of the brick- equivalent length leq ¼ ah V e ¼ ah q¼1 n¼1 g¼1 detJwq wn wg
mortar interface, which, in [43], follows the relation where wq ; wn and wg are the weight factors of the Gaussian integra-
Gint
II ¼ 130r þ 58 N/m (with r in MPa). In this case, tan /, c; dk ,
tion scheme, J the Jacobian of the transformation, V e the element
and l are defined directly from the experimental outcomes [43], area and ah a modification factor that depends on the typology of
whereas the brittleness parameter n is chosen to be the best the finite element used. In this way, the mesh size does not signif-
approximation of Gint icantly influence the material response.
II for the three experimental curves in Fig. 6b.
Finally, as can be observed in Fig. 7, the tensile (Fig. 7a) and Additionally, to control the dilatancy in the quasi-brittle mate-
shear (Fig. 7b) interfacial behaviours here proposed appear in good rial response, a nonassociative flow rule is considered to define the
agreement with the experimental results obtained in [2,43]. It has plastic strain rate. It is obtained by a flow rule generated by a
to be pointed out that the shear stiffness which can be read in Drucker-Prager type plastic potential. In particular, it is defined
Fig. 7b is given by the deformability of the 3D FEs (in this case by the dilatancy angle w, typically assumed equal to 10° in agree-
mainly to the mortar FEs) and not by the deformability of the inter- ment with experimental evidences [48] and previous numerical
faces, which can be considered rigid-cohesive. models [49,50], and a smoothing constant  generally assumed
equal to 0.1 [49].
As regard as the yield surface, a multiple-hardening Drucker-
4. Brick and mortar nonlinear behaviour Prager type surface is assumed. It is characterized by the ratio
f b0 =f c0 between the biaxial initial compressive strength f b0 and
Tensile and compressive plastic-damage nonlinear behaviour is the uniaxial initial compressive strength f c0 and a constant q,
assumed for brick and mortar, based on the plastic-damage model which represents the ratio of the second stress invariant on the
developed by Lee and Fenves [47] for quasi-brittle materials. In the tensile meridian to that on the compressive meridian at initial
following, the main features of the model are recalled. yield. Typically, f b0 =f c0 ¼ 1:16 and q ¼ 2=3 for quasi-brittle materi-

Table 1
Mechanical properties for small-scale masonry specimens.

Mortar properties Interfacial properties


Young’s modulus [MPa] 2970 Tensile behaviour Shear behaviour
Poisson’s ratio [n] 0.15 f t [MPa] 0.28 tan / [n] 1.01
uk [mm] 0.20 c [MPa] 0.87
Brick properties f [n] 4.38 dk [mm] 0.4
Young’s modulus [MPa] 16,700 n [n] 1.1
Poisson’s ratio [n] 0.15 l [n] 0.73

Fig. 7. Comparison between experimental and numerical results for small-scale masonry specimens: (a) tensile behaviour (experimental envelope (grey area) and numerical
response (red line)) and (b) shear behaviour (experimental envelopes (grey areas) and numerical responses (blue, green and orange lines) for three different levels of initial
compression: 0.1, 0.5 and 1.0 MPa).

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 7

Fig. 8. Plastic-damaging behaviour of brick and mortar: (a) tensile and (b) compression uniaxial nonlinear curves.

Table 2 Experimental tests conducted by Vermeltfoort and Raijmakers


General parameters for quasi-brittle materials (brick and mortar). [52] and by Chee Liang [53] are considered for the in-plane and
 [n] w [n] f b0 =f c0 [n] q [n] out-of-plane response of masonry panels, respectively. Mechanical
0.1 10° 1.16 2/3 properties utilized for the in-plane and out-of-plane benchmarks
are collected in Table 3. When more than one value is given in
the same cell of the table, the first value refers to the in-plane
benchmark, whereas the second one refers to the out-of-plane
als [51]. The general parameters adopted for quasi-brittle materi-
benchmark. In general, the tensile response of masonry joints is
als, such as brick and mortar, are collected in Table 2.
defined in terms of the tensile strength and fracture energy in ten-
sion (Mode I), whereas the shear response of masonry joints is
5. Numerical examples defined in terms of friction, cohesion, residual friction and Mode
II-fracture energy. It appears clear that uk and f will be derived
Experimental-numerical comparisons for the in-plane and out- from the value of fracture energy in tension (Mode I), whereas dk
of-plane behaviours of masonry panels are here provided to show and n will be derived from the value of Mode II-fracture energy.
the effectiveness and the accuracy of the model proposed. The To this aim, the brittleness parameters f and n have been kept
detailed micro-model herein proposed has been implemented in equal to the ones of Section 3.1, and the values uk and dk have been
Abaqus Standard [41]. Geometric nonlinearity is considered in all chosen so that the fracture energy values were satisfied. Reference
the analyses to account for large-displacement effects. to [54] has been made to define the uniaxial inelastic stress-strain

Table 3
Mechanical properties utilized for the in-plane and out-of-plane benchmarks. When more than one value is given in the same cell, the first value refers to the in-plane benchmark,
whereas the second one refers to the out-of-plane benchmark.

Interfacial mechanical properties


Tensile behaviour Shear behaviour

f t [MPa] 0.20, 0.12 tan / [n] 0.75, 0.58


uk [mm] 0.36 c [MPa] 0.22
f [n] 4.38 dk [mm] 0.4
n [n] 1.1
l [n] 0.75, 0.58
Mortar mechanical properties
Young’s modulus [MPa] 850, 2300
Poisson’s ratio [n] 0.15
Tensile uniaxial nonlinear behaviour Compressive uniaxial nonlinear behaviour
Stress [MPa] Inelastic strain dt [n] Stress [MPa] Inelastic strain dc [n]
1.5 0 0 7.8 0 0
0.1 0.002 0.95 8.2 0.002 0
0.4 0.015 0.95
Brick mechanical properties
Young’s modulus [MPa] 16,700
Poisson’s ratio 0.15
Tensile uniaxial nonlinear behaviour Compressive uniaxial nonlinear behaviour
Stress [MPa] Inelastic strain dt [n] Stress [MPa] Inelastic strain dc [n]
3.5 0 0 11.0 0 0
0.3 0.002 0.95 11.5 0.001 0
0.6 0.007 0.95

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
8 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

Fig. 9. In-plane response of masonry wall panels [52]: (a) boundary conditions and (b) assembly of textured units employed in the numerical model.

relationships. The evolution of the degradation damage scalar vari- Fig. 10 provides experimental-numerical comparisons: the
ables dt and dc has been kept substantially proportional to the experimental load-displacement curves for J4D, J5D and J7D walls
decay of the uniaxial stresses, as successfully experienced is sev- are compared with the numerical results carried out using a tex-
eral numerical campaigns [49,16,55]. tured unit mesh composed of 20 hexahedral 8-nodes FEs. In this
Concerning the in-plane benchmark, the mechanical properties figure, the numerical predictions reported by Lourenço & Rots
for brick, mortar and brick–mortar interfaces employed in the [21] and by Macorini & Izzuddin [30] are also shown. A good agree-
analyses (Table 3) were reported in previous research [19,21,30]. ment between experimental and numerical results can be
In addition, the tensile strength of mortar has been assumed with observed up to collapse, including initial stiffness, maximum
reference to the results on mortar prisms obtained in the experi- capacity and the post-peak response of the panels. Also, the predic-
mental campaign carried out in the TU Delft laboratories in 1991 tions of the proposed modelling approach are generally close to
[2]. those reported in [21,30] for all the considered walls, with the cur-
Concerning the out-of-plane benchmark, the material parame- rent predictions of the post-peak response for wall J7D better than
ters used for the interfaces elements (Table 3) are equivalent to the one obtained in [21].
the values used in [30] for the same wall. The elastic stiffness The discretization of the textured units is explicitly chosen by
of brick and mortar were not investigated by Chee Liang [53]. the user. The role of the mesh size is shown in Fig. 11a, in which
Therefore, the Young’s modulus of mortar has been assumed the influence of mesh refinement on the load-displacement curves
according to [54], whereas the Young’s modulus of brick has been is collected. The results obtained using a textured unit mesh con-
kept the same as that shown in [2], being the materials of the sisting of 20 hexahedral 8-nodes FEs (coarse mesh) and a textured
same type. The other properties are the same to the in-plane unit mesh consisting of 108 hexahedral 8-nodes FEs (fine mesh)
benchmark. are compared. As can be noted, very small discrepancies emerged.
Thereby, mesh dependency appears negligible, also thanks to the
5.1. In-plane response regularization of the fracture energy in the continuum plastic-
damage model. This aspect is particularly appealing as the analyses
Results obtained by Vermeltfoort and Raijmakers [52] in shear with the coarse mesh presented a computational cost considerably
tests on single-leaf panels are here considered. The identical wall smaller than the fine mesh.
specimens, named J4D, J5D and J7D in [52], with a length (990 Fig. 11b shows the influence of the nonlinear behaviour of tex-
mm) to height (1000 mm) ratio of approximately 1 were consid- tured units on the load-displacement curves. As can be noted, the
ered (Fig. 9). They are characterized by 18 brick layers of which 2 fact of accounting for the cracking and crushing of textured units
were fixed to steel beams so as to keep the top and bottom edges
of the element straight during the test (green zones in Fig. 9a). Each
brick is 204 mm  98 mm  50 mm, whereas the bed and head
mortar joints are 12.5 mm thick. Particularly, the masonry panels
were initially preloaded with a vertical top pressure, Pv = 0.3 MPa
for J4D and J5D and Pv = 2.12 MPa for J7D. Then a horizontal load
was then applied in the plane of the walls at the top edge under
displacement control up to collapse, see Fig. 9a.
During the tests, first, horizontal cracks appeared at the top and
bottom of the walls. Then, cracks started to develop diagonally
along the bed and head mortar joints and through the bricks, up
to failure. The experimental response was characterized by a soft-
ening branch that started when diagonal cracks appeared in the
centre of the specimens.
The wall is modelled here using the detailed micro-modelling
approach presented in the previous sections. The analyses followed
the two-step boundary conditions depicted in Fig. 9a. The assem-
bly of textured units employed in the numerical model is high- Fig. 10. Experimental – numerical comparisons of the load – displacement curves
lighted in Fig. 9b. for the masonry wall panels loaded in plane.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 9

Fig. 11. Load – displacement curves for Pv = 2.12 MPa: (a) investigation of the mesh dependency and (b) influence of the nonlinear behaviour of the textured units.

Fig. 12. Comparison of the panel’s crack pattern: (a) tensile damage contour plot, (b) compressive damage contour plot, (c) interfaces which exhibited failure and (d)
experimental crack pattern for the specimen with Pv = 2.12 MPa (J7D in [52]).

significantly affects the post-peak behaviour (Fig. 11b), whereas in considering or not the nonlinear behaviour of textured units
the hypothesis of linear elastic textured units slightly overesti- would increase by increasing the vertical pressure as well as the
mates the peak load. Basically, it is expected that the differences interlocking of the masonry texture (e.g. for multi-leaf walls).

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
10 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

Fig. 13. Experimental crack pattern: (a) photos of the failure of Wall 8 and Wall 12 from [53] and (b) sketch of the crack pattern of Wall 12.

5.2. Out-of-plane response

Numerical analyses are also carried out to assess the effective-


ness of the detailed micro-modelling approach developed to inves-
tigate the out-of-plane behaviour of masonry panels. Comparisons
are carried out against the experiments performed by Chee Liang
[53].
The out-of-plane behaviour of a solid wall, simply supported
along its four edges and subjected to bi-axial bending, is consid-
ered, and reference is made to experiments on two identical spec-
imens: wall 8 and wall 12 in [53]. The single-leaf masonry wall
panels were 1190 mm high, 795 mm wide and 53 mm thick. The
dimensions of the brick were 112 mm  53 mm  36 mm and
the thickness of the mortar joints were 10 mm. The two specimens
were loaded up to collapse by applying a uniform out-of-plane
pressure through an air-bag sandwiched between the wall and a
stiff reacting frame. Another stiff steel frame was connected to
the wall on the other side, so as to prevent out-of-plane displace-
ments and provide fixed supports along the four edges. The crack
pattern experienced by the two wall specimens [53] is shown in
Fig. 14. Comparison of the lateral pressure – out-of-plane displacement curves. Fig. 13.
To compute the solution up to the collapse of the panel (also in
case of softening), a quasi-static direct-integration dynamic analy-
sis procedure has been adopted [41]. This algorithm permits to
Finally, Fig. 12 shows the deformed shape and crack pattern in study quasi-static responses in which inertia effects are introduced
the masonry wall panel obtained from the numerical model, in primarily to regularize unstable behaviours. The Authors experi-
terms of tensile damage contour plot (Fig. 12a), compressive dam- enced a better performance of this algorithm, specifically in the
age contour plot (Fig. 12b), and interfaces which exhibited failure softening regime, with respect to more common arc length
(Fig. 12c). Also, numerical results are compared with the experi- procedures.
mental crack pattern experienced in [52] (Fig. 12d). As can be Fig. 14 provides the numerical-experimental comparisons in
noted in Fig. 12, these predictions are in good agreement with terms of lateral pressure-transversal displacement curves, where
the actual crack pattern. Particularly, the interfaces which exhib- the textured unit mesh composed of 20 hexahedral 8-nodes FEs,
ited failure are placed along the panel diagonal. Furthermore, few shown in Fig. 11a, has been implemented. Although the through-
textured units experienced tensile failure in the central part of this thickness discretization may play a certain role, especially in the
diagonal (Fig. 12a), representing brick and mortar cracking. In out-of-plane analysis of multi-leaf walls [56], the utilization of
addition, few textured units also showed crushing in the two two 8-nodes hexahedral FEs through-thickness appears sufficiently
extremities of the diagonal (Fig. 12b). These features have also accurate for the case under study. The experimental results
been experienced by the experimental tests [52], see for example reported in [53] consist of a partial load-displacement curve for
(Fig. 12d), confirming the good accuracy of the model proposed. wall 8 and the maximum capacity for the walls 8 and 12. Good
Finally, these predictions are also in good agreement with the main agreement between the numerical and experimental results can
crack paths and with the numerical results reported in [21,30]. be observed. The maximum lateral pressure obtained with the

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 11

Fig. 15. Comparison between experimental and numerical out-of-plane deflections when the lateral pressure is equal to 20 kN/m2, see the green and magenta points in
Fig. 14.

Fig. 16. Crack pattern obtained from the proposed model: (a) deformed shape, (b) out-of-plane displacements contour plot and (c) tensile and (d) compressive damage
contour plots at the end of the simulation.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
12 A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx

Table 4
Times required to conduct the analyses.

Simulation Time requireda (hh:mm:ss)


In-plane coarse mesh (Pv = 0.30 MPa) 00:06:33
In-plane coarse mesh (Pv = 2.12 MPa) 00:07:18
In-plane fine mesh (Pv = 2.12 MPa) 00:23:20
Out-of-plane 00:09:11
a
Utilizing a commercial laptop equipped with a processor IntelÒCoreTM i7-6500U
CPU @ 2.50 GHz and 16 GB RAM.

According to the modelling approach proposed, masonry is rep-


resented by textured units consisting of one brick and few mortar
layers composed of 3D solid FEs obeying to plastic-damage consti-
tutive laws. This permits to represent the brick and mortar
mechanical behaviour when cracking and/or crushing occur. Tex-
tured units are assembled, accounting for any actual 3D through-
thickness arrangement of masonry (including walls with openings,
Fig. 17. Crack pattern obtained by consolidated numerical models: (a) Milani [57] multi-leaf walls, etc.), by means of zero-thickness cohesive-
and (b) Macorini and Izzuddin [30]. frictional interfaces based on the contact penalty method. This per-
mits to account for the brick-mortar bond failures both in tension
proposed model appears very close to the experimental capacity and shear.
[53], to the collapse pressure determined in [57] through a 3D limit To reach this goal, this paper introduced an interface model.
analysis approach and to the numerical curve obtained in [30]. Par- Indeed, the interface behaviour assumed in the 3D detailed
ticularly, the curve obtained with the proposed approach very well micro-model is governed by an ad-hoc modification of the stan-
fits the partial load-displacement curve for wall 8. Additionally, dard surface-based contact behaviour implemented in Abaqus.
Fig. 15 provides the comparison between the experimental and Contextually, an automatic subroutine ad-hoc written by the
numerical out-of-plane deflections at the instant, shown in authors has been implemented to reproduce a Mohr-Coulomb fail-
Fig. 14 by means of a green point and a magenta point, with lateral ure surface with tension cut-off.
pressure equal to 20 kN/m2, i.e. at an instant slightly prior to fail- The interfacial behaviour appeared to be consistent with exper-
ure. Here again, a good numerical-experimental agreement is imental outcomes on small-scale masonry specimens. The results
achieved in terms of out-of-plane deflections. of numerical analyses carried out to investigate both the in-plane
Finally, Fig. 16 shows the crack pattern obtained by means of and the out-of-plane responses of brick-masonry panels up to col-
the proposed model, in terms of deformed shape at collapse lapse has been presented and compared with experimental out-
(Fig. 16a), out-of-plane displacement contour plot (Fig. 16b), ten- comes. From this comparison, it was shown that the use of the
sile damage contour plot (Fig. 16c) and compressive damage con- proposed modelling approach allows the accurate representation
tour plot (Fig. 16d). By comparing the numerical crack pattern of of the masonry behaviour both in the in-plane and out-of-plane
Fig. 16 with the experimental one (Fig. 13), it can be noted that responses. The results achieved demonstrate the significant poten-
the actual failure mechanism, although slightly different in the tial of the proposed approach.
two walls, is qualitatively represented by the numerical model Additionally, although this model accounts for a very detailed
proposed. Particularly, the large vertical crack that runs in the description of masonry constituents and is characterized by a lar-
middle of the panel crossing head mortar joints and bricks as ger complexity with respect to existing numerical models, its com-
well as the diagonal cracks observed in the tests are well repre- putational demand appears reasonably acceptable. Indeed, as
sented. Indeed, as can be noted in Fig. 16c, tensile damage is shown in Table 4, the computational time needed in the simula-
experienced in the central part of the textured units which are tions are, after all, moderate. Even, the 3D detailed micro-model
placed in the central vertical part of the wall, in agreement with proposed appears faster than other more standard 2D micro-
the actual vertical cracks experienced by both walls (Fig. 13) modelling approaches, see in [19] the time needed for the same
which alternatively crosses the bricks. For the sake of compar- in-plane benchmark, based on well-known interface elements
ison, the crack pattern obtained by numerical models consoli- [21]. Therefore, the contact-based formulation proposed appears
dated in the scientific community [57,30] is reported in Fig. 17. preliminarily efficient. The Authors are currently testing this model
As can be noted, the crack pattern computed by the model here on large-scale masonry structures using parallelization techniques
proposed (Fig. 16), is in good agreement with the ones depicted to reduce the computational time. From the first attempts, stan-
in Fig. 17. dard workstations appear sufficient to supply this task.
Finally, considering the accuracy of the model proposed, its
6. Conclusions application to simulate the behaviour of masonry panels under cer-
tain loading conditions can be used to help laboratory experi-
In this paper, a novel numerical approach to model masonry has menters in designing new or optimizing experimental set-ups, in
been proposed. The 3D detailed micro-model presented consists of predicting the crack pattern, the maximum load and the ultimate
the coupling of contact-based rigid-cohesive interfaces with 3D displacement of scheduled tests.
nonlinear-damaging textured units (which explicitly account for
the mortar layers), which is a novelty in the scientific literature. Acknowledgements
This novel modelling approach can, in fact, be fully characterized
by the properties obtained on small-scale specimen tests on brick Mauro Parodi, Massimo Damasio and Claudio Cavallero (http://
and mortar (stiffness, compressive and tensile responses) and on www.exemplar.com/) are gratefully acknowledged for their tech-
small masonry assemblages (tensile and shear responses of the nical support. Financial support by the Italian Ministry of Educa-
mortar-brick bond). tion, Universities and Research MIUR is gratefully acknowledged

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007
A.M. D’Altri et al. / Computers and Structures xxx (2018) xxx–xxx 13

(PRIN2015 ‘‘Advanced mechanical modeling of new materials and [28] Sacco E, Toti J. Interface elements for the analysis of masonry structures. Int J
Comput Methods Eng Sci Mech 2010;11(6):354–73.
structures for the solution of 2020 Horizon challenges” prot.
[29] Rekik A, Lebon F. Identification of the representative crack length evolution in
2015JW9NJT_018). a multi-level interface model for quasi-brittle masonry. Int J Solids Struct
2010;47(22–23):3011–21.
References [30] Macorini L, Izzuddin BA. A non-linear interface element for 3D mesoscale
analysis of brick-masonry structures. Int J Numer Meth Eng 2011;85
(12):1584–608.
[1] Hendry AW. Structural masonry. UK: Palgrave MacMillan; 1998. [31] Minga E, Macorini L, Izzuddin BA. Enhanced mesoscale partitioned modelling
[2] van der Pluijm R. Material properties of masonry and its components under of heterogeneous masonry structures. Int J Numer Meth Eng 2018;113
tension and shear. In: 6th Canadian Masonry Symposium, 15–17 June 1992, (13):1950–71.
Saskatoon, Canada, 1992. [32] Minga E, Macorini L, Izzuddin BA. A 3D mesoscale damage-plasticity approach
[3] Mazzotti C, Sassoni E, Pagliai G. Determination of shear strength of historic for masonry structures under cyclic loading. Meccanica 2017;53(7):1591–611.
masonries by moderately destructive testing of masonry cores. Constr Build [33] Baraldi D, Cecchi A. Discrete approaches for the nonlinear analysis of in plane
Mater 2014;54:421–31. loaded masonry walls: molecular dynamic and static algorithm solutions. Eur J
[4] Formisano A, Marzo A. Simplified and refined methods for seismic Mech A Solids 2016;57:165–77.
vulnerability assessment and retrofitting of an Italian cultural heritage [34] Formica G, Sansalone V, Casciaro R. A mixed solution strategy for the nonlinear
masonry building. Comput Struct 2017;180:13–26. analysis of brick masonry walls. Comput Methods Appl Mech Eng 2002;191
[5] Sacco E, Addessi D, Sab K. New trends in mechanics of masonry. Meccanica (51–52):5847–76.
2018;53(7):1565–9. [35] Lemos JV. Discrete element modeling of masonry structures. Int J Archit
[6] Lourenço PB. Computations on historic masonry structures. Prog Struct Mat Heritage 2007;1(2):190–213.
Eng 2002;4(3):301–19. [36] Casolo S. Modelling in-plane micro-structure of masonry walls by rigid
[7] Marfia S, Sacco E. Multiscale damage contact-friction model for periodic elements. Int J Solids Struct 2004;41(13):3625–41.
masonry walls. Comput Methods Appl Mech Eng 2012;205–208:189–203. [37] Smoljanović H, Nikolić Ž, Živaljić N. A combined finite–discrete numerical
[8] Giambanco G, La Malfa Ribolla E, Spada A. Meshless meso-modeling of model for analysis of masonry structures. Eng Fract Mech 2015;136:1–14.
masonry in the computational homogenization framework. Meccanica [38] Beatini V, Royer-Carfagni G, Tasora A. A regularized non-smooth contact
2017;53(7):1673–97. dynamics approach for architectural masonry structures. Comput Struct
[9] Milani G, Tralli A. Simple SQP approach for out-of-plane loaded homogenized 2017;187:88–100.
brickwork panels, accounting for softening. Comput Struct 2011;89(1– [39] Baraldi D, Cecchi A. Discrete and continuous models for static and modal
2):201–15. analysis of out of plane loaded masonry. Comput Struct 2018 (In press).
[10] Godio M, Stefanou I, Sab K, Sulem J, Sakji S. A limit analysis approach based on [40] Bui T, Limam A, Sarhosis V, Hjiaj M. Discrete element modelling of the in-plane
Cosserat continuum for the evaluation of the in-plane strength of discrete and out-of-plane behaviour of dry-joint masonry wall constructions. Eng
media: application to masonry. Eur J Mech A Solids 2017;66:168–92. Struct 2017;136:277–94.
[11] Stefanou I, Sab K, Heck J. Three dimensional homogenization of masonry [41] AbaqusÒ. Theory manual, version 6.14; 2014.
structures with building blocks of finite strength: a closed form strength [42] Weyler R, Oliver J, Sain T, Cante J. On the contact domain method: a
domain. Int J Solids Struct 2015;54:258–70. comparison of penalty and Lagrange multiplier implementations. Comput
[12] Massart TJ, Peerlings RHJ, Geers MGD. An enhanced multi-scale approach for Methods Appl Mech Eng 2012;205–208:68–82.
masonry wall computations with localization of damage. Int J Numer Meth [43] van der Pluijm R. Shear behaviour of bed joints. In: 6th North American
Eng 2007;69(5):1022–59. Masonry Conference, 6–9 June 1993, Philadelphia, Pennysylvania, USA; 1993.
[13] Brasile S, Casciaro R, Formica G. Finite Element formulation for nonlinear [44] Serpieri R, Albarella M, Sacco E. A 3D microstructured cohesive–frictional
analysis of masonry walls. Comput Struct 2010;88(3–4):135–43. interface model and its rational calibration for the analysis of masonry panels.
[14] Pegon P, Anthoine A. Numerical Strategies for solving continuum damage Int J Solids Struct 2017;122–123:110–27.
problems with softening: application to the homogenization of masonry. [45] Godio M, Stefanou I, Sab K. Effects of the dilatancy of joints and of the size of
Comput Struct 1997;64(1–4):623–42. the building blocks on the mechanical behavior of masonry structures.
[15] Toti J, Gattulli V, Sacco E. Nonlocal damage propagation in the dynamics of Meccanica 2017;53(7):1629–43.
masonry elements. Comput Struct 2015;152:215–27. [46] van der Pluijm R, Rutten H, Ceelen M. Shear behaviour of bed joints. In:
[16] Castellazzi G, D’Altri AM, de Miranda S, Chiozzi A, Tralli A. Numerical insights Proceedings of the twelfth international brick/block masonry conference;
on the seismic behavior of a non-isolated historical masonry tower. Bull 2000.
Earthq Eng 2018;16(2):933–61. [47] Lee J, Fenves GL. Plastic-damage model for cyclic loading of concrete
[17] Pelà L, Cervera M, Roca P. An orthotropic damage model for the analysis of structures. J Eng Mech 1998;124(8):892–900.
masonry structures. Constr Build Mater 2013;41:957–67. [48] Mirmiran A, Shahawy M. Dilation characteristics of confined concrete. Mech
[18] Berto L, Saetta A, Scotta R, Vitaliani R. An orthotropic damage model for Cohesive-frictional Mater 1997;2(3):237–49.
masonry structures. Int J Numer Meth Eng 2002;55(2):127–57. [49] Milani G, Valente M, Alessandri C. The narthex of the church of the nativity in
[19] Petracca M, Pelà L, Rossi R, Zaghi S, Camata G, Spacone E. Micro-scale Bethlehem: a non-linear finite element approach to predict the structural
continuous and discrete numerical models for nonlinear analysis of masonry damage. Comput Struct 2018 (In press).
shear walls. Constr Build Mater 2017;149:296–314. [50] Castellazzi G, D’Altri AM, de Miranda S, Ubertini F. An innovative numerical
[20] Vasconcelos G, Lourenço PB. In-plane experimental behavior of stone masonry modeling strategy for the structural analysis of historical monumental
walls under cyclic loading. J Struct Eng 2009;135(10):1269–77. buildings. Eng Struct 2017;132:229–48.
[21] Lourenço PB, Rots JG. Multisurface interface model for analysis of masonry [51] Lubliner J, Oliver J, Oller S, Oñate E. A plastic-damage model for concrete. Int J
structures. J Eng Mech 1997;123(7):660–8. Solids Struct 1989;25(3):299–326.
[22] Gambarotta L, Lagomarsino S. Damage models for the seismic response of [52] Vermeltfoort AT, Raijmakers TMJ. Deformation controlled meso shear tests on
brick masonry shear walls. Part I: the mortar joint model and its applications. masonry piers - part 2. Draft Report, Department of BKO, TU Eindhoven; 1993.
Earthq Eng Struct Dynam 1997;26(4):423–39. [53] Chee Liang N.G. Experimental and theoretical investigation of the behavior of
[23] Gambarotta L, Lagomarsino S. Damage models for the seismic response of brickwork cladding panel subjected to lateral loading. Ph.D. thesis University
brick masonry shear walls. Part II: the continuum model and its applications. of Edinburgh; 1996.
Earthq Eng Struct Dynam 1997;26(4):441–62. [54] Kaushik HB, Rai DC, Jain SK. Stress-strain characteristics of clay brick masonry
[24] Giambanco G, Rizzo S, Spallino R. Numerical analysis of masonry structures via under uniaxial compression. J Mater Civ Eng 2007;19(9):728–39.
interface models. Comput Methods Appl Mech Eng 2001;190(49– [55] D’Altri AM, Castellazzi G, de Miranda S, Tralli A. Seismic-induced damage in
50):6493–511. historical masonry vaults: a case-study in the 2012 Emilia earthquake-stricken
[25] Alfano G, Sacco E. Combining interface damage and friction in a cohesive-zone area. J Build Eng 2017;13:224–43.
model. Int J Numer Meth Eng 2006;68(5):542–82. [56] Casolo S, Milani G. Simplified out-of-plane modelling of three-leaf masonry
[26] Parrinello F, Failla B, Borino G. Cohesive–frictional interface constitutive walls accounting for the material texture. Constr Build Mater 2013;40:330–51.
model. Int J Solids Struct 2009;46(13):2680–92. [57] Milani G. 3D upper bound limit analysis of multi-leaf masonry walls. Int J
[27] Fouchal F, Lebon F, Titeux I. Contribution to the modelling of interfaces in Mech Sci 2008;50(4):817–36.
masonry construction. Constr Build Mater 2009;23(6):2428–41.

Please cite this article in press as: D’Altri AM et al. A 3D detailed micro-model for the in-plane and out-of-plane numerical analysis of masonry panels.
Comput Struct (2018), https://doi.org/10.1016/j.compstruc.2018.06.007