Professional Documents
Culture Documents
net/publication/275970264
CITATIONS READS
0 200
3 authors:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Bahman Ghiassi on 08 May 2015.
Abstract
Vulnerability of masonry buildings to seismic activities has been confirmed by the past earthquakes. A large
number of these structures in Iran are residential buildings, hospitals and schools, and their damage would
end to the human being causalities. Therefore, seismic evaluation and strengthening of the masonry buildings
are very crucial. Coating the walls with thin reinforced mortar layers is an ordinary technique in retrofitting
the masonry buildings in Iran. However, due to the lack of appropriate design guidelines, the rehabilitation
procedures are mostly carried out for the purpose of practically being experienced. In this study, the elastic
properties of bare and retrofitted masonry walls, used in linear evaluation procedures, are derived from
applying the developed homogenization method. This approximate homogenization method can be used
easily in any kind of bond patterns, the solution is in closed-form and the results show good agreement with
the homogenization procedures used by other researchers. The homogenized model is also used to simply
develop the anisotropic yield surfaces of bare and retrofitted masonry elements, assuming an elastic-brittle
Department of Civil Engineering, Tarbiat Modares University, Jalaale-al Ahmad Ave., Tehran, Iran
While more than 70% of the buildings in Iran are masonry, most of them are residential buildings, hospitals
and schools In which many people and children are accommodated The experience of past earthquakes
revealed that masonry buildings are mostly vulnerable to the seismic activities, and can be destroyed even by
moderate earthquakes leading to massive losses of human beings` lives and widespread damage extensive
loses. Their vulnerability is mostly due to: 1-some structures are constructed in the period of existing no
available seismic code, 2- while the seismic codes had been published others are designed and constructed
without following them, 3- the third group are designed and constructed according to the seismic codes, but
their rules are not safety enough because of the complexity and lack of information on the behavior of the
masonry structures. In general, high seismic vulnerability of these structures can be associated with their
There are a large number of methods for retrofitting masonry structures that are intended to improve both
their in-plane and out-of-plane integrities and behaviors. Some conventional methods are surface treatments
like Ferrocement (Abrams and Lynch 2001), FRP or Shotcrete layers (Kahn 1984), grout and epoxy
injection (Calvi and Magenes 1994), external reinforcement (Chuang et al. 2004), confining masonry
walls and post-tensioning (Elgawady et al. 2004). Coating the walls with reinforced concrete layers is one
of the most common methods for strengthening the masonry structures in Iran. However, because of the lack
of information, experimental and analytical, on this method and lack of appropriate design guideline,
Numerical modeling of masonry structures can be applied with a macro or micro approach. Micro modeling
represents brick, mortar and brick-mortar interfaces separately. Although this method may be the most
accurate one, it is much more expensive in terms of computational costs and the corresponding high number
of degrees of freedom, limits its applicability. A large number of micro-models were developed by researches
such as Page 1978 and Lourenco and Rots 1997. In macro modeling approach, there is no distinction
between brick, mortar and brick-mortar interfaces and the masonry structures are modeled as isotropic or
anisotropic homogenous continuum media and also it doesn’t take much time to model and analyze even
are needed. These properties can be achieved through the homogenization techniques. The homogenization
theory for periodic media allows derivation of global behavior of masonry from the behavior of constitutive
materials (brick and mortar). In other words homogenization is the technique of describing the composite
behavior of masonry in terms of macro or average stresses and strains. Homogenization techniques can be
applied in two main ways called experimental homogenization and analytical homogenization. The former
needs so many costly tests and the obtained results are limited to the tests conditions (Zucchini and
Lourenco 2002). The latter method has been used by many authors in different frameworks (elastic, plastic,
limit states, etc.) and with a variety of methods like single-step homogenization (Anthoine 1995), two-step
homogenization and Engineering methods (Pande 1989) and in most cases only in plane behavior in a two
Single-step homogenization, used by many authors like Anthoine 1995 and Urbanski et al. 1995 needs
In the two-step homogenization, the masonry is homogenized first horizontally, and then vertically or vice
versa. This simplified approach has been used by many authors, but there are two sources of errors in the
results obtained by this method: 1- the results depend on the order of the fulfilling steps; 2- different bond
patterns may lead to similar results because the geometrical arrangements are not entirely taken into account
(Anthoine 1995).
Engineering methods are used with some simplifying assumptions and in many cases lead to a closed-form
Zucchini and Lourenco 2002 proposed a new micro-mechanical model to overcome the limitations of the
two step homogenization and showed good results but not in a closed-form.
This paper presents a new three dimensional homogenization method based on energy concepts. The
advantages of this method are: 1-simple usage, 2-few calculations and 3-closed-form solution usable in
masonry walls modeling for linear evaluation procedures. The results of proposed method also show very
Masonry is a composite material consisting brick and mortar that its components are usually are arranged
periodically. In this section a basic cell, having a periodic pattern in the whole wall, is selected (Fig. 1) and
the average stresses and strains are computed by a developed homogenization method. This homogenization
technique is based upon the strain energy of the hyper elastic materials. There is a relation between total
energy and the stress and strain tensors in the hyper elastic materials as:
u
ij (1)
eij
where, ij is the stress in i-th plane and along j-th direction of the basic cell, u is total volumetric strain energy
of the basic cell and eij is the strain in i-th plane and along j-th direction of the basic cell.
Here, it is assumed that the components of masonry are hyper elastic. Therefore, according to the equation
(1), in case of computing the total strain energy of basic cell in terms of average stresses and strains, the
average stresses of basic cell can be obtained in terms of average strains and then the elasticity tensor will be
achieved.
Fig. 1
Total volumetric strain energy can be calculated by the summation of the volumetric energy of each
constituent as follows:
U E V E U b V b U m1 V m1 U m2 V m2 (2)
where, U E is the total volumetric strain energy, V E is the total volume of basic cell, U b is the total strain
energy of bricks, V b is the total bricks volume, U m1 is the total strain energy of horizontal mortar, V m1 is the
horizontal mortar volume, U m2 is the total strain energy of vertical mortar and V m2 is the vertical mortar
volume.
V b 2abtz (4)
V m1 2t f (b t v )t z (5)
V m2 2t v at z (6)
Fig. 2
The strain energy in hyper elastic materials can be calculated as:
1
u ij ij eij (7)
2
where,
ij Cijkl eij (8)
1 1 1
uij ij eij Cijkl eij elk u eT Ce (9)
2 2 2
where, C is the elasticity tensor and e is strain tensor. Assuming that the brick and mortar are isotropic
where
C11 2 , C22 (11)
E E
, (12)
(1 )(1 2 ) 2(1 )
C11 C12
G12 G13 G23 G (13)
2
Having calculated the strain energy of each component and putting them in equation (2), the total strain
energy of the basic cell will be obtained in terms of masonry components strains. In order to calculate the
strains of masonry components in terms of average strains, the basic cell should be subjected to the
compression along x, y and z axes, and to the shear on xz, xy and yz planes (axes are shown in Fig. 2). Here
The stresses in opposite faces of each masonry components are the same, like differential surfaces.
Normal and pure shear stresses produce only corresponding stresses in each masonry component.
As an example the assumed stress distribution in masonry components under average tension stress along x
Fig. 3
The stiffness differences between units and mortar cause complex interactions between the masonry
components if the masonry is deformed. These differences also cause heterogeneous distribution of
deformations over the units and mortar in comparison with the average deformation of homogenous masonry.
For better understanding of internal deformational behavior of masonry components and their relations,
detailed finite element calculations have been carried out on the selected basic cell for different loading
conditions (Fig. 4). Fig. 5 shows the deformed shape of the basic cell subjected to compression along x and y
axes. Zucchini and Lourenco 2002 had shown that for compression along axis z, all basic cell components
are subjected to the real homogenous state of normal stress. Further information about stress states on the
Fig. 4
Fig. 5
The displacement of each nodes on the faces normal to x and y directions, subjected to compression along x
and y axes, are shown in Fig. 6 for different ratios of mortar/brick stiffness (Em/Eb). The node numbers are
shown in Fig. 4.
Fig. 6
Fig. 6 illustrates that with increasing the mortar/brick stiffness ratio the relative nodal displacements will
Fig. 6 shows that with increasing the mortar/brick stiffness ratio the relative nodal displacements will
Considering the mentioned assumptions, the relation between masonry components strains and average
11b f1 11, 11m1 11, 11m2 nf1 11 , 22b f 2 22 , 22m1 nf 2 22 , 22m2 22 (14)
in which:
Eb b tv a tf n b tv a tf Gb b tv
n , f1 , f2 , n2 1 , f3 , nG ,z
Em b ntv a nt f b tv a n2t f Gm nGb tv
where b, tv, a and tf are shown in Fig. 2, ij is the average strain and k ij is the strain of k-th component of
masonry, Eb is the elastic modulus of brick, Em is the elastic modulus of mortar, Gb is the brick shear modulus
and Gm is the mortar shear modulus. It is assumed that the elastic properties of horizontal and vertical mortars
are the same, however considering different elastic properties for them has no effect on the homogenization
procedure. So the total volumetric energy of the basic cell will be as follows:
U E A11
2
B 22
2
C 33
2
D11 22 E11 33 F 22 33 G12
2
H13
2
I 23
2
(17)
where,
1 1
A (C11b f12Vb C11m1Vm1 C11m2n2 f12Vm2 ) , B (C11b f12Vb C11m1n2 f 22Vm1 C11m2 f 22Vm2 )
2V E 2V E
1 1
C (C11bVb C11m1Vm1 C11m2Vm2 ) , D (C12b f1 f 2Vb C12m1nf2Vm1 C12m2nf1 f 2Vm2 )
2V E 2V E
1 1
E E
(C12b f1Vb C12m1Vm1 C12m2nf1Vm2 ) , F (C12b f 2Vb C12m1nf2Vm1 C12m2 f 2Vm2 )
V VE
2 2
G E
(Gb f32Vb Gm1n22 f32Vm1 Gm2 f32Vm2 ) , H E
(GbVb nG z 2 Gm1Vm1 Gm2Vm2 z 2 )
V V
2
I (Gb f32Vb Gm1n22 f32Vm1 Gm2 f32Vm2 )
VE
2 A D E 0 0 0
D 2B F 0 0 0
E F 2C 0 0 0
C (18)
0 0 0 0.5G 0 0
0 0 0 0 0.5H 0
0 0 0 0 0 0.5I
The young’s modulus and shear modulus of homogenized masonry also will be computed as:
x x x
eq
E11 eq
, E22 eq
, E33 (19)
4 BC F 2
4 AC E 2
4 AB D 2
eq
G12 0.5G, G13
eq
0.5H , G23
eq
0.5I (20)
2 DC EF 2 DC EF 2 EB DF eq 2 EB DF 2 AF DE eq 2 AF DE
12
eq
, 21
eq
, 13
eq
,
2 31
, 23
eq
, 32 (21)
4 BC F 2 4 BC E 2 4 BC F 2
4 AB D 2
4 AB D
2
4 AC E
where,
eq eq eq
In the above equations, E11 , E22 and E33 are continuum elastic modulus in X, Y and Z directions,
eq eq eq
respectively and G12 , G13 and G23 are continuum shear modulus in XY, XZ and YZ planes, respectively.
The proposed relations are still simplified with the aim of regression methods as follows:
0.2 0.36
0.028T f b E 0.021b E
E11 ( 1.14) Eb m , E22 ( 0.41T f 2.23) Eb m (22)
a Eb a Eb
0.07
0.03bTf E
E33 ( 1.14) Eb m (23)
a Eb
0.4 0.2
0.2b E 0.027bTf E
G12 G23 ( 0.44T f 2.25) Gb m , G13 ( 1.15) Gb m (24)
a Eb a Eb
3. Results verification
Zucchini and Lourenco 2002 used micro-mechanical model for masonry homogenization considering
actual deformations of basic cells, including additional internal deformation modes with respect to the
standard two step homogenization procedure. They showed how good their model is in agreement with finite
element results. In this section, to check the accuracy of adopted procedure, the results are compared with
Zucchini and Lourenco’s homogenization model. The basic cell properties are chosen as Zucchini and
Lourenco’s, i.e. the unit dimensions and mortar thickness are 210x100x52 and 10 mm, respectively, for the
unit, the Young’s modulus (Eb) is 20 MPa and the poison ratio ( b ) is 0.15, for the mortar the Young’s
modulus is varied to yield a ratio, Em/Eb ranging from 1 to 1000 and the poison ratio ( m ) is kept 0.15. The
same properties are adopted for bed, head and cross joints. The computed elastic properties of the
homogenized cell are compared with the Zucchini and Lourenco’s, Fig. 7. Considering different stiffness
ratios between mortar and brick allows assessing the model performance for inelastic behavior (Zucchini
and Lourenco 2002). In nonlinear behavior, the material stiffness is degraded and the homogenization of
nonlinear processes leads to the large differences in the stiffness of mortar and unit.
Fig. 7
The very good agreement between the proposed model and Zucchini and Lourenco’s one is shown in Fig. 7.
In this section the results accuracy of the proposed homogenization procedure is controlled through modeling
some structural walls, each one with two different approaches, using a finite element software. The first
approach is modeling in a micro phase and the other is macro modeling by using the continuum properties of
In the first sample, a brick masonry wall by the length and height of 2720 and 2090 mm, respectively, is
modeled in both macro and micro phases (Fig. 8). The wall thickness is kept 10 cm. In micro model the brick
dimensions are 200x100x50 mm and the thickness of the mortar is 1 cm. The mechanical properties of brick
and mortar, used in the modeling, are summarized in Table 1, and the continuum mechanical properties,
Table 1
Table 2
The wall is subjected to in-plane load. The displacement of top right corner of the wall is shown in Table 3
Fig. 8
Table 3
In the second sample, a 1040x890 mm masonry panel is subjected to out of plane loading with different
supporting conditions of at 4 sides. Table 4 comprises the central point displacement of micro and macro
Table 4
In this section the retrofitted masonry is homogenized and the continuum elastic properties are computed. The
basic cell, selected for the homogenization process, is shown in Fig. 9. The calculations are made in such a
way that the thickness of wall and concrete layer are inserted in the relations and therefore, the proposed
Fig. 9
Masonry is assumed as continuum orthotropic material and concrete layer as isotropic. The calculations are
done as in the previous sections. The relations between components strains and average strains are obtained
as follows:
tm tc t t
f 23 , f13 m c (25)
n23tm tc n13tm tc
11c 11m 11 , 22c 22m 22 , 33c f3 33, 33m n3 f3 33 (26)
23c f 23 23 , 23m n23 f 23 23 , 12c 12m 12 , 13c f1313 , 13m n13 f1313 (27)
in which,
Ec t t G Gc
n3 , f3 m c , n13 c , n23
Em3 n3tm tc Gm13 Gm23
where, Ec and Em3 are concrete and masonry Young’s modulus, respectively, in z direction; tm is the thickness
of masonry wall; tc is the thickness of concrete layer; Gc is the concrete shear modulus; Gm13 and Gm23 are
masonry shear modulus in xz and yz directions, respectively. The average elasticity tensor of the
2 A D E 0 0 0
D 2 B F 0 0 0
E F 2C 0 0 0
C (28)
0 0 0 0.5G 0 0
0 0 0 0 0.5H 0
0 0 0 0 0 0.5I
where,
1 1 1
A (C11wtw C11ctc ) , B (C22wtw C11ctc ) , C (C33wtwn32 f32 C11ctc f32 )
2(tm tc ) 2(tm tc ) 2(tm tc )
1 1 1
D (C12wtw C12ctc ) , E (n3 f3C13wtw f3C12ctc ) , F (n3 f3C23wtw f3C12ctc )
(tm tc ) (tm tc ) (tm tc )
2 2 2
G (G12wtw Gctc ) , H 2 2
(n13 f13G13wtw f13
2
Gctc ) , I 2 2
(n23 f 23G23wtw f 23
2
Gctc
(tm tc ) (tm tc ) (tm tc )
x x x
eq
E11 eq
, E22 eq
, E33 (29)
4 BC F 2 4 AC E 2 4 AB D 2
eq
G12 0.5G, G13
eq
0.5H , G23
eq
0.5I (30)
2 DC EF 2 DC EF 2 EB DF
12
eq
, 21
eq
, 13
eq
(31)
4 BC F 2
4 AC E 2
4 BC F
2
2EB DF 2 AF DE 2 AF DE
31
eq
, 23
eq
, 32
eq
(32)
4 AB D 2
4 AB D 2
4 AC E 2
the proposed relations are also simplified with the aim of regression methods as follows:
tm E E
eq
E11 E22
eq
(0.068 1.068)(0.243 m 0.45 b 0.335) Ec (33)
tc Ec Ec
tm E E
eq
E33 (0.077 0.924)(0.124 m 0.785 b 0.126) Ec (34)
tc Ec Ec
tm E E E
eq
G12 (0.062 1.062) (0.074 b ln( m ) 0.282 b 0.145) Ec (35)
tc Ec Eb Ec
tm E E E
eq
G13 G23
eq
(0.066 1.066) (0.052 b ln( m ) 0.413 b 0.04) Ec (36)
tc Ec Eb Ec
6. Structural comparisons of retrofitted masonry
In this section a retrofitted masonry wall, subjected to in-plane and out-of-plane loading, is modeled with
macro and micro approaches. The wall displacements in each condition are shown in Table 5 and 6; the
Table 5
Table 6
The obvious good agreement between the results of both models indicates the accuracy of the
homogenization process used in this research. For better verifying the results of the homogenization process,
a complete 2 story retrofitted masonry building (Fig. 10) are modeled, first with a micro approach -modeling
the brick and mortar separately- and then with a macro one using shell elements with the elastic properties
Fig. 10
This masonry building is subjected to concurrent lateral static loads applied at structural mass center,
applying 100% of design forces in x direction plus 30% of design forces in y direction and vice versa. The
deformed shape and displacement of the roof are shown in Fig. 11.
Fig. 11
7. Parametric study
The effect of various values of Em/Eb on the orthotropic properties of retrofitted wall is shown in Fig. 12. The
effect of different stiffness ratios of unit and mortar (Em/Eb) and also different thickness ratios of masonry
wall and concrete layer (tc/tw) are presented in the Fig. 12.
Fig. 12
The objective of this section is to derive the yield surfaces of homogenized bare and retrofitted masonry.
These surfaces distinguish the linear regions of the homogenized material. Considering the assumptions,
made to develop the homogenization method, the stresses of each masonry component can be computed in
terms of homogenized material stresses (average stresses). Then, in case of assuming an elastic-brittle
behavior for mortar and unit, the yield surface is defined by the stresses in which any components reaches
their failure criteria. Both the brick and mortar are assumed isotropic. Zucchini and Lourenco 2002 used
classic Von Mises criterion for compression behavior and Rankine criterion for describing tensile behavior of
masonry components. As the masonry is a frictional material, using independent pressure failure criteria leads
to no acceptable results and applying pressure dependent failure criteria is preferred. Therefore, in this study
Fig. 13 shows the resulting yield surfaces of homogenized bare and retrofitted masonry in which the principle
stresses coincide with the material axes happening in the absence of shear stresses. The retrofitted masonry is
layered with a concrete layer of the thickness equal to that of masonry wall. The inner surface, formed by the
intersection of components yield surfaces, is homogenized masonry’s yield surface. The assumed parameters
are as follows:
Em
f cb 9.81, ftb 4, fcm 3.33, ftm 0.196 ( MPA), 0.1,m b 0.2
Eb
Fig. 13
Retrofitting the masonry with concrete layer has widen the yield surface of the material and also increased the
strength of material in both x and y directions. These yield surfaces are usable in computing the continuum
9. Conclusions
This paper presents a new approximate homogenization method of masonry. This method is simple and
usable in all bond patterns of brick masonry walls, its results are in closed-form and also show good
agreement with the other exact methods. A brick masonry wall is homogenized with proposed method in the
elastic range, the relations are derived and their simplified forms are proposed. Then, the masonry walls,
retrofitted by concrete layer, are homogenized and the relations for computing the elastic properties are
proposed. Finally, the anisotropic failure surfaces of bare brick masonry and retrofitted brick masonry with
[1] Abrams, D.P. and Lynch, M.J. (2001). “Flexural behavior of retrofitted masonry piers”,
KEERC-MAE Joint Seminar on Risk Mitigation for Regions of Moderate Seismicity, Illinois,
USA.
[2] Anthoine, A. (1995). “Derivation of in-plane elastic characteristics of masonry through
homogenization theory”, International Journal of Solids and Structures, Vol. 32, No. 2, pp.
137-163.
[3] Calvi, G. and Magenes, G. (1994). “Experimental results on unreinforced masonry shear walls
damaged and repaired”, 10th IB2MaC, Calgary, USA.
[4] Chuang, S. et al. (2004) ,“Sesmic retrofitting of unreinforced masonry by cable system” , 13th
WCEE, August, Vancour, Paper No. 3228
[5] Elgawady, M. and Lestuzzi, P. and Badoux, M. (2004). “A review of convetional seismic
retrofitting techniques for URM”, 13th IB2MaC, Amesterdam.
[6] Kahn, L. (1984) “Shotcrete retrofit for unreinforced brick masonry”, 8th WCEE, USA, 583-
590.
[7] Kappos, A.J. and Penelis, G.G. and Drakopoulos, C.G. (2002). “Evaluation of simplified
models for lateral loadd analysis of unreinforced masonry buildings”, Journal of Structural
Engineering, Vol. 128, pp. 890-897.
[8] Lourenco, P.B. and Rots, G. (1997). “Multisurface interface model for analysis of masonry
structures”, Journal of Engineering Mechanics, Vol. 123, No. 7, July, pp. 660-668.
[9] Mistler, M. et al. (2007). “In-plane and out-of-plane homogenisation of masonry”, Comput
struct, doi:10.1016/j.compstruc.2006.08.087.
[10] Page, A.W. (1978). “Finite element model for masonry”, J. Struc. Div., Vol. 104(8), pp. 1267-
1285.
[11] Pande, G.N. and Liang, J.X. and Middleton, J. (1989). “Equivalen elastic moduli for unit
masonry”, Computers and Geotechnics 8, pp. 243-265
[12] Urbanski, A. and Szarlinski, J. and Kordecki, Z. (1995). “Finite element modeling of the
behavior of the masonry walls and coloumns by homogenization approach”, Computer
methods in structural masonry, pp.32-41.
[13] Zucchini, A. and Lourenco, P.B. (2002). “A micro-mechanical model for homogenisation of
masonry”, International Journal of Solids and Structures, Vol. 39, pp. 3233–3255
Fig. 7. Comparison between shear and Young’s modulus of the proposed model and Lourenco model
Fig. 7. Comparison between shear and Young’s modulus of the proposed model and Lourenco model
Fig. 8. The deformed shape of the micro model under in-plane loading
Fig .12- The effect of different values of Em/Eb on orthotropic properties of retrofitted masonry
List of Tables:
Fig. 7. Comparison between shear and Young’s modulus of the proposed model and Lourenco model
Fig. 8. The deformed shape of the micro model under in-plane loading
Brick Mortar
Young's Modulus(N/mm2) 9810 981
Poison Ratio 0.15 0.15