Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/281399450
READS
43
3 AUTHORS:
Anastasios Drougkas
Luca Pel
4 PUBLICATIONS 2 CITATIONS
SEE PROFILE
SEE PROFILE
Pere Roca
Polytechnic University of Catalonia
86 PUBLICATIONS 572 CITATIONS
SEE PROFILE
1 INTRODUCTION
Unreinforced masonry buildings are common in many seismic prone countries. In most cases they
were built before the development of comprehensive research and the proposal of rational
engineering procedures for their design. The careful observation of damages and collapses produced
by recent earthquakes [1] has shown that the most vulnerable mechanisms are normally the out-ofplane failures of walls, mainly due to presence of insufficient connections between elements and lack
of rigid horizontal diaphragms. Once such drawbacks are overcome by introducing appropriate
devices able to ensure the monolithic behavior of the building, the shear walls provide to the building
the sufficient stability to lateral forces [2]. For this reason, research studies were devoted to the
interpretation of the failure modes observed in the experimental tests of this structural members. The
comprehensive experimental programs available in the literature present both empirical and analytical
approaches for the derivation of the strength of masonry shear walls [3,4,5]. Nowadays, the advent of
sophisticated numerical approaches has produced different strategies for the simulation of the
behavior of masonry structures [6,7,8]. In this context, it is considered that approaches combining
both experimental and numerical studies are necessary to consolidate the existing knowledge and to
obtain a better understanding of the complex behavior of masonry shear walls.
This research presents the numerical simulation of in-plane shear tests on walls under different
levels of vertical stress. The reference experimental program was carried out at UPC, Barcelona [9]
over walls made of solid clay bricks and cement mortar. Two different numerical approaches are used
to simulate the tests, i.e. meso- and micro-modeling. The comparison with analytical models proposed
by the current technical standards shows the need for improved simplified methods for the analysis of
masonry shear-walls.
2 CASE STUDY
In the present research, a series of in-plane shear tests on 15 scale walls, under different levels of
vertical stress, was considered [9]. The walls were composed of solid clay bricks and cement mortar
arranged in single leaf running bond. The walls were constructed lying vertically on a flat surface with
the mortar being poured for the filling of the bed and head joints.
The bricks were scaled to 1:4 of the full brick dimensions and the mortar was produced using M80
mortar with adjusted granulometry for the removal of the larger aggregates. As such, the bricks
measured 72.5 12.5 35 mm3 and the joints were 2.5 mm thick. The walls were capped using a stiff
but vertically unrestricted beam. The vertical pre-compression and the horizontal load were applied on
this beam.
The panels, consisting of 16 courses in height and four units in length, measured 297.5 mm in
length and 137.5 mm in height. The thickness of the wall measured 35 mm. The geometry of the walls
is shown in Figure 1, along with the setup for the shear wall experiments.
Table 1.
Material properties
Units
Mortar
Wallette - Vertical
Interface
fc [MPa]
35.0
8.34
17.54
ft [MPa]
0.55
ft [MPa]
3.40
1.36
c0 [MPa]
0.42
E [MPa]
4080
3500
[-]
390
Figure 2 shows the envelope of the maximum shear average stresses resisted by the set of
experimental walls for the different average compression stresses applied. As can be seen in the
diagram, one of the walls was subjected to increasing concentric compression, with no shear force
applied, up to failure. In this case, the obtained maximum compressive stress was lower than the one
produced by the standardized tests, implying the involvement of effects related to the larger size of
the tested walls.
(a)
(b)
(c)
(d)
(e)
(f)
Figure 3. Experimental failure mechanisms for different levels of vertical pre-compression: a) 1.19
MPa, b) 1.93 MPa, (c) 2.98 MPa, (d) 4.76 MPa, (e) 5.94 MPa and (f) 9.05 MPa.
3 MESO-MODEL
The experimental behavior of the shear walls has been simulated firstly by means of a FE mesomodel. This approach considers the heterogeneous and composite character of the material by
distinct modeling of the units and the joints. In this way, assuming that bricks have been
experimentally characterized, the values derived from the experimental tests can be directly assigned
to units. Joints are understood as a mechanical unit representing both the mortar and the unit-mortar
interface, thus overall mechanical properties have to be assumed. This characteristic approximation in
the approach is counterbalanced by a reasonable computational effort, if compared with that of
detailed micro-models which represent units, interfaces and mortar.
In the present study, standard plane-stress continuum finite elements were employed in the mesomodel discretization. This choice is motivated to get a result which is a good compromise between
accuracy and computational efficiency. The FE mesh, including units, mortar joints and the test
loading beam, is shown in Figure 4 and is composed of 18492 quadrangular finite elements.
Units
Mortar
fc [MPa]
ft [MPa]
E [MPa]
17.5
17.5
3.40
0.55
4080
3480
[-]
0.15
0.25
Gc [N/mm]
Gf [N/mm]
17.5
17.5
0.096
0.027
4 SIMPLIFIED MICRO-MODEL
In the context of simplified FE micro-modeling, the model features separate simulation of the units
using continuum elements and the mortar joints using interface elements [7]. This approach allows for
a detailed simulation of masonry under in-plane loads for case studies where the dominant failure
modes initiate and develop in the mortar joints. The continuum elements are used for the simulation of
the cracking of the units and the interface elements are used for the simulation of cracking, shearing
and crushing in the joints.
The benefits include the potential for direct simulation of all the usual failure modes encountered in
the units and the interface. The drawbacks include a higher level of involvement in the creation of the
model geometry and the slightly elevated computational cost. Additionally, several mechanical tests
are necessary for the characterization of the properties of masonry, most importantly of the
compressive strength of the composite.
In the present case study, the model was created under a plane stress assumption. The mesh is
illustrated in Figure 5 and is composed of 11520 continuum elements and 2136 interface elements.
The element size was thus 2.52.5mm2 in area, or equivalently numbered 306 elements per unit.
9th International Masonry Conference, Guimares 2014
The units were modeled using a multi-directional fixed crack model [16]. Exponential softening in
tension based on fracture energy was employed, along with zero stiffness retention in shear. The
interfaces were modeled using a combined cracking-shearing-crushing model [7]. The general
purpose finite element program DIANA [17] was used for the calculations.
(a)
(b)
Figure 5. Simplified micro-model mesh: (a) continuum elements and (b) interface elements.
The properties used for the simplified micro-model are presented in Table 3. As for the mesomodel, a standard value is used for the Poissons coefficient. In turn, the shear fracture energy of the
interface is chosen based on the value for the cohesion [13]. The dilatancy angle was considered
equal to 0. The elastic stiffness of the interfaces was adjusted in order to achieve a masonry elastic
modulus equal to that of the masonry composite [7].
Table 3.
Units
Interface
fc [MPa]
17.5
ft [MPa]
3.40
ft [MPa]
0.55
E [MPa]
4080
c0 [MPa]
0.42
[-]
0.15
[-]
450
[-]
00
Gf [N/mm]
0.096
GI [N/mm]
0.013
GII [N/mm]
0.042
5 RESULTS
Both the meso-models and the simplified micro-models provide accurate results in terms of
maximum shear, for all levels of vertical stress considered. Furthermore, the failure modes are
satisfactorily simulated and agree well with those observed experimentally. The simulated failure
models exhibit the predominant features characteristic of shear walls for different levels of vertical
load: mortar joint or interface cracking for low vertical load, combined cracking and sliding for medium
load and crushing of the compressed foot for high vertical load. The cracking patterns obtained in the
units and the mortar are illustrated in Figures 6-7.
Figure 6. Meso-model cracking patters for different levels of vertical prestress: (a) 0.893 MPa, (b)
3.333 MPa and (c) 5.936 MPa.
Figure 7. Micro-model cracking patters for different levels of vertical prestress: (a) 0.893 MPa, (b)
3.333 MPa and (c) 5.936 MPa.
Similarly, the cracking pattern in the mortar joints or in the interfaces exhibits a shift with the level
of vertical load applied. For a low pre-compression level, cracking or separation of the bed joints is
extensive due to the wall acting as a cantilever in bending. For higher levels of vertical load, the
cracking or separation is observed mostly in the head joint interfaces due to the influence of indirect
tension.
Figure 8 shows the comparison among the experimental maximum average shear stresses and the
numerical ones derived from micro- and meso-modeling. Both numerical strategies adopted attain a
remarkable agreement with the experimental evidence.
The figure also shows the failure envelope curves obtained by well-known analytical models
proposed by current technical standards [18,19]. The shear strength of masonry is provided by the
Eurocode 6 by a Mohr-Coulomb frictional law. According to the Eurocode 6, for the type of masonry
adopted in the experimental program examined, the initial shear strength of masonry should be taken
equal to 0.20 MPa. This simple formula agrees well with the experimental results for lower values of
pre-compression of the wall, even if the limitation of this formulation has been recognized in other
studies.
The second analytical model shown for the sake of comparison is a flexure domain, based on inplane equilibrium of the shear wall with a plastic compression stress distribution at the base crosssection and neglecting the tensile strength of bed joints. As can be seen in Figure 8, this approach
reproduces correctly the experimental behaviour of the walls for low vertical compression but
overestimates the shear strength for high initial compression.
Figure 8. Comparison between experimental results, numerical and analytical models predictions.
The last analytical failure criterion is based on the assumption that diagonal shear failure is
attained when the principal stress at the center of the wall reaches a critical value, which is defined as
a reference tensile strength of masonry according to Turnek and Cacovic [3]. This value is rather
difficult to be obtained experimentally. In this study, and in a simplified way, the masonry tensile
strength is approached by the mortar-unit tensile strength, which has been measured experimentally
as 0.55 MPa (Table 1), and represents a 3.1% of the masonry compressive strength. It must be
noted, however, that the tensile strength value to be considered for the Turnek and Cacovic criteria
is normally assumed to be a global value that does not coincide with the unit-mortar tensile strength.
Finally, Figure 9 presents the comparison between the stress-displacement curves obtained from
the two numerical approached. Overall, in addition to reaching approximately the same peak stress,
the curves show good agreement in terms of initial stiffness and ductility, especially in the low vertical
stress level cases. The softening of the structure is also in good agreement between the two models,
although the simplified micro-model yields a slightly lower stiffness near the peak stress for high
levels of vertical prestress, resulting in the attainment of the maximum stress for a slightly higher
displacement value.
Both models exhibit very similar results in terms of wall ductility. Low prestress values results in a
very ductile response, as the wall responds in a rocking mode. Increasing the vertical stress level
results in higher maximum shear but a more brittle post peak behavior.
2.500
2.000
(MPa)
1.500
1.000
0.895
1.190
1.486
1.933
2.381
2.681
2.981
3.333
3.867
4.286
4.762
5.357
5.952
9.048
0.500
0.000
0.000
0.200
0.400
0.600
0.800
d (mm)
1.000
1.200
1.400
1.600
Figure 9. Comparison between stress-displacement curves from the two models for different values
of vertical stress: Meso-model in solid lines and simplified micro-model in dotted lines.
6 CONCLUSIONS
A series of experimental shear tests on walls has been numerically simulated using two different
modeling techniques corresponding to a meso-model and a simplified micro-model. Both techniques
have achieved good results on the maximum average shear forces resisted for different precompression levels. Good agreement is also found between the experimental failure modes and
those predicted numerically. The analytical models considered presented certain limitations in the
prediction of the maximum shear. For example, the EC6 equation overestimates the shear strength of
walls under high vertical load, while the flexural model underestimates it for low levels. The numerical
models, however, provide accurate results throughout the range of vertical pre-compression levels.
ACKNOWLEDGEMENTS
This research has received the financial support from the Ministerio de Economia y Competitividad
through the research projects SUBTIS (Study of the Sensitivity of Urban Buildings to Tunneling
Induced Settlements, ref. num. BIA2009-13233) and the research project MICROPAR (Identification
of mechanical and strength parameter of structural masonry by experimental methods and numerical
micro-modelling, ref num. BIA2012-32234) and from the ERDF (European Regional Development
Fund).
REFERENCES
[1] Penna A, Morandi P, Rota M, Manzini CF, da Porto F, Magenes G (2013) Performance of
masonry buildings during the Emilia 2012 earthquake. Bulletin of Earthquake Engineering, DOI:
10.1007/s10518-013-9496-6
[2] Magenes G., Calvi, (1997), In plane seismic response of brick masonry walls, Earthquake
Engineering and Structural Dynamics, vol. 26, pp. 1091-1112.
[3] Turnek V., Cacovic F., (1971): Some experimental results on the strength of brick masonry
walls, Proc. of the 2nd Intern. Brick Masonry Conference, Stoke-on-Trent, pp. 149-156.
[4] Yokel, F.Y., Fattal, S.G. (1976). Failure hypothesis for masonry shear walls., J. Of Structural
Division, ASCE,515-532.
[5] Ganz, H.R., Thurlimann, B. (1983). Strength of brick walls under normal force and shear. Proc.
8th Int. Symposium on load bearing brickwork, London, 27-29.
[6] Roca P, Cervera M, Gariup G, Pel L. Structural analysis of masonry historicalconstructions.
Classical and advanced approaches. Arch Comput Methods Eng 2010;17:299325.
[7] P. B. Loureno and J. G. Rots, Multisurface interface model for analysis of masonry structures,
J. Eng. Mech., vol. 123, no. 7, p. 660, 1997.
[8] Pel L, Cervera M, Roca P. An orthotropic damage model for the analysis of masonry
structures. Construction and Building Materials 2013; 41:957-967.
[9] Charry Ablanque J. A., Estudio Experimental del Comportamiento de Paredes de Obra de
Fbrica de Ladrillo ante la Accin de Cargas Laterales. Department of Strength of Materials and
Structural Engineering, Technical University of Catalonia, Barcelona Spain: PhD dissertation,
2010.
[10] CEN, BS-EN 1052-1 - Methods of test for masonry - Part 1: Determination of compressive
strength. 1999.
[11] Comisin Permanente del Hormign. Instruccin de Hormign Estructural EHE-08. Ministerio
10
11