Professional Documents
Culture Documents
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
1. Introduction and background factors causing their formation is important, in order to propose
remedial measures.
Cracking is the most common and visible defect found in Masonry is a composite, heterogeneous, non-linear material
masonry. It is usually considered a symptom of distress within that exhibits distinct directional properties, because the mortar
the fabric of the building. The appearance of cracking can affect joints act as planes of weakness. Various attempts have been
the value of the structure, its insurability and saleability, and can made to model this complicated nature of masonry, where
give rise to litigation. Also it must be controlled to maintain the its inherent discontinuities (units, joints and interfaces) are
moisture resistance of the wall. Therefore, even though cracking of recognised. Depending on the level of accuracy and simplicity
masonry is not a normal structural design consideration, regardless desired, different modelling strategies have been used by masonry
of the reason for cracking it is unacceptable where good finish is researchers. Micro-modelling and macro-modelling techniques
desired. are the main methods among them [2].
Concrete slabs exposed to direct sunlight experience tempera- The micro-modelling strategy is preferred for understanding
ture related horizontal movements. In addition, temperatures on the local behaviour of masonry structures since both masonry
the top surface will be higher than those on the underside of the and mortar can be modelled in detail and many attempts
slab, causing an upward deflection of the slab during heating. In have been made by researchers [3–8]. The continuum units
a typical building, masonry and concrete elements are connected have been modelled by 2D plane stress or shell elements and
to each other at their respective interfaces. Therefore, significant joints have been modelled by interface elements. However, in
movements may be generated on the masonry walls due to the most cases, the collective response of the mortar joints with
movement of the roof slab. These movements can result in over- brick–mortar interface has been considered [4,6–8]. The failure
stressing and cracking in masonry. Even though these cracks lead of masonry is governed by complex phenomena, i.e. failure of
to considerable problems with respect to the performance and ap- each of the constituents and of the interface between them.
pearance of a building, there is very little in the literature regard- This level of complexity has been used mostly for numerical
ing this, e.g. Ref. [1]. Therefore a study about these cracks and the simulation of laboratory experiments that have been conducted
to investigate the failure of masonry by cracking and crushing
under different static and dynamic loading conditions. However,
∗ Corresponding author. Tel.: +94 11 2650422; fax: +94 11 2650622. such micro-modelling is only suitable for modelling of small
E-mail address: priyan@civil.mrt.ac.lk (W.P.S. Dias). structural elements, because of the large amount of data and
0141-0296/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2010.01.019
1412 K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422
variables involved. In large and practice-oriented analysis the released by the opening of the crack. Then the global structural
interaction between units and mortar is usually negligible for the behaviour will be the same when the strain energy is redistributed.
global structural behaviour. In these cases the macro-modelling The method is widely used to model the opening or development
technique is widely used [2]. of a crack when the crack has been identified beforehand [23].
In the case of macro-modelling, the stress–strain behaviour This paper describes the numerical model developed to under-
of masonry has been derived by a procedure of homogenization stand the propagation of cracks in masonry walls under a time
or ‘averaging’ of the stress–strain relationships of mortar joints varying thermal load on the roof slab. A commercially available
and masonry units. Different homogenization techniques have finite element code ANSYS (Version 11.0) was used to construct
been developed by masonry researchers [9–12]. However in most the model. The modelling was done as a structural–thermal cou-
cases average properties obtained from experimental techniques pled non-linear analysis. Mode 1 tensile cracking with linear soft-
have been used in the literature [1,4,13–15]. In cases where ening of masonry was assumed. The smeared crack approach was
representation of damage induced anisotropy effects is a required used to obtain the crack propagation. These results were compared
feature, a classical macroscopic continuum description has been with the information obtained from a survey of buildings having
enhanced with embedded discrete bands to represent the failure thermally induced cracks and also with results from a few physical
behaviour of masonry [9]. models that were constructed to 1/3 scale of the prototype.
Tensile weakness of masonry and the cracking that results
from it is the most significant factor contributing to the non- 2. Approach and methodology
linear behaviour of masonry structures. Therefore crack modelling
is undoubtedly the most important factor to be considered. In 2.1. Objective
crack modelling, the selection of a particular cracking model from
the various alternatives available in the literature depends on The main objective of the study was to identify the contribution
the purpose of the finite element study, the nature of the output of factors that cause cracking, such as the structural form of
desired from the study and available computing facilities. Usually the wall (i.e. whether the wall is load bearing or within a
a crack model must have three components [16]: a definition of concrete frame) and the effect of different geometric and structural
crack initiation, a method of crack representation, and a criterion features such as openings and lintels. The relative effectiveness
for crack propagation. of numerical modelling as coupled non-linear analysis to predict
Crack initiation has been represented in many ways based on locations and directions of cracks was also established.
the state of stress or strain at a point of a material element. Two
approaches have been adopted as the basis of failure theories [17]. 2.2. Related studies
The first assumes homogeneous isotropic behaviour, while the
The numerical modelling described in this paper was supple-
second considers the behaviour of the unit and joint material
mented by other studies, reported elsewhere. A building survey
separately under the action of biaxial or triaxial stress. A number
was carried out to identify factors associated with the type of
of investigators have proposed failure criteria for masonry using
cracking being investigated. The specific nature of the cracks was
experimental [18,19] and theoretical [20–22] methods.
carefully studied — i.e. whether they increased monotonically with
Different crack representation approaches have been used by
time or whether their widths were cyclic with diurnal temperature
researchers in crack modelling. The discrete crack approach and
variations. The latter is good evidence for the cracks being thermal,
smeared crack approach are the main methods that have been
as opposed to settlement cracks. In some cases soil investigation
used. In the discrete crack approach, cracking of materials is
reports were used to eliminate settlement as a cause for cracking.
modelled by separation of the appropriate nodes of adjoining
Care was also taken to distinguish between cracking due to shrink-
elements. When the stress or strain at a node or the average age of masonry and that due to thermal effects. The most common
in adjacent elements exceeds the allowable value, the node is type of shrinkage crack is where infill masonry walls separate away
redefined as two nodes and elements on either side are allowed from the concrete frame. In load bearing walls, a prominent verti-
to separate. This requires monitoring the response and modifying cal crack through the height of the wall will be a shrinkage crack. It
the topology of the finite element mesh corresponding to the crack was observed that cracks due to thermal movements of the overly-
configurations at each state of loading. ing slab do not depend on the building orientation. In most build-
In the smeared crack approach, the cracked material is assumed ings cracks were visible on both internal and external walls. These
to remain as a continuum. Rather than representing a single crack, observations indicate that the formation of cracks is not due to the
the crack is represented as an infinite number of parallel fissures direct solar radiation on walls. Considering all the cases as a whole,
across the cracked element: i.e. the effect of a crack or sheared the behaviour of these cracks and their patterns in relation to the
plane is spread out over the area that belongs to an integration structural form – i.e. whether the wall was load bearing or concrete
point. Although a fixed finite element mesh is used (no re-meshing) framed – was identified [24].
and new degrees of freedom are not involved during the analysis, Three physical models were constructed (1/3 scale) to identify
a large number of material parameters are required. However the the effect of structural form (using models representing 6 m long
use of a smeared crack approach is convenient when the crack load bearing and concrete framed walls) and the effect of aspect
orientation is not known beforehand [17]. A smeared crack can ratio of the wall (using models representing 3 m long and 6 m
occur in the element mesh at any location and in any direction. long load bearing walls) on the above phenomenon. Only the
Crack propagation will depend on the state of stress in the overlying slabs were exposed to solar radiation, while the walls
region ahead of the crack. Two methods can be used to predict were shielded. Strain and temperature variations of the models
potential crack growth. The first method uses a criterion based were monitored [25]. The objective of the physical modelling was
on the inherent strength of the material and the local state of not to establish an equivalence between model and prototype
stress. The second uses a criterion based on the fracture toughness but rather to investigate the general pattern of movements and
of the material [16]. In the first method, the development of cracking in the walls.
discontinuities is incorporated into the stress–strain constitutive Finally, prior to carrying out the non-linear finite element
laws of the material that will be non-linear with softening of analysis described here, linear elastic analysis using SAP2000 was
material. In the second method, the strain energy released by the also carried out, in part to highlight the conditions under which
softening of material is assumed as equal to the strain energy such an approach would be particularly deficient.
K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422 1413
Table 2
Description of assemblies.
Assembly Structural type Window opening Lintel Separation from top (m)
Table 3
Thermal boundary conditions.
Time (s) Temperature (K)
Top surface of roof slab Bottom surface of roof slab Bottom surface of concrete beam
0 (6:00 a.m.) 296 (23 °C) 296 (23 °C) 296 (23 °C)
28 800 (2:00 p.m.) 333 (60 °C) 308 (35 °C) 306 (33 °C)
32 400 (3:00 p.m.) 333 (60 °C) 308 (35 °C) 306 (33 °C)
43 200 (6:00 p.m.) 300 (27 °C) 300 (27 °C) 300 (27 °C)
64 800 (12:00 p.m.) 296 (23 °C) 296 (23 °C) 296 (23 °C)
86 400 (6:00 a.m.) 296 (23 °C) 296 (23 °C) 296 (23 °C)
125 mm
Level
3000 mm
1000 mm
1000 mm
750 mm 3000 mm
2.6. Loading and analysis temperature across the roof slab obtained from the analysis is
shown in Fig. 4. The 3.125 m and 3.000 m levels (from the base
In addition to the initially applied self-load of the structure, of the wall) represent the top and bottom of the slab respectively.
the temperature variation of the structure over a period of 24 h The zero value of the time axis corresponds to 6 a.m.
(86 400 s) was applied to the structure. The self-load was applied After reaching its peak value, the temperature of the top surface
at the first load step and the thermal load was then applied to the decreases as shown in Fig. 4. However, the results show that the
structure at a number of substeps. The analysis was performed as a rate of decrease of temperature within the slab is less than that of
structural–thermal coupled analysis. The normal substep interval the outer surfaces and hence the internal temperature of the slab
was set to 30 s while the minimum and the maximum substep sizes remains at higher values in the evenings. Also, the temperature
were set to 1 s and 300 s respectively. The variation of temperature across the slab reached a steady equilibrium state during the night.
from one substep to the other was considered as a ramped load. The The results of this thermal analysis were input as the thermal loads
analysis was performed at each substep and results of each substep for the structural analysis of the model at successive time steps.
were recorded.
3.1. Heat flow analysis The concrete–masonry interface of these walls was studied to
understand the displacements generated in both load bearing and
The temperature values of nodes at intervals of 25 mm across concrete framed walls. The horizontal and vertical displacement of
the thickness of the slab were obtained. The variation of the the nodes from the outer edge of wall (0.0 m) to the centre (3.0 m)
K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422 1415
σzp
-
fc
r2
r1 σxp = σyp = σzp
r2
r1
r1 η
r2
σyp
-
fc
σxp
- Octahedral Plane Distance (m)
fc
Fig. 5. Displacement at the wall–slab interface of load bearing wall.
Fig. 2. The failure surface described in 3D principal stress space (ANSYS Theory
reference).
ft
Tcft
E Rt
1 1
ε
εck 6εck
Table 4
Details about the first crack initiation (Note: MA9 to MA15 display only marginal cracking).
Assembly Location of the first crack Temperature of the roof slab
Top surface (°C) Bottom surface (°C)
MA1 Between 1.375 m to 1.5 m from the outer edge close to the wall–slab interface 57.7 34.6
MA2 Between 1.0 m to 1.25 m from the outer edge close to the wall–beam interface 58.0 34.2
MA3 Between 0.875 m to 1.125 m from the outer edge close to the wall–beam interface 53.0 32.7
MA4 Between 0.875 m to 1.125 m from the outer edge close to the wall–beam interface 52.7 32.6
MA5 Top corner of the right vertical boundary of the opening 45.3 30.0
MA6 Top corner of the right vertical boundary of the opening 48.6 31.3
MA7 Top corner of the right vertical boundary of the opening 48.6 31.3
MA8 Top corner of the right vertical boundary of the opening 37.7 27.8
MA9 Top of left and right wall–column interfaces 37.9 27.5
MA10 Bottom of the vertical separation at outer columns 45.3 30.1
MA11 Bottom of the vertical separation at outer columns 38.5 28.0
MA12 Bottom of the vertical separation at outer columns 34.2 26.7
MA13 Bottom of the vertical separation at the right outer column 43.2 29.5
MA14 Bottom of the vertical separation at the right outer column 44.7 30.0
MA15 Bottom left corner of the opening 59.8 34.9
1
1 2
3
Starting point
Fig. 10. Vector plot of total strains in 6 m load bearing wall (MA1).
Fig. 13. Vector plot of total strains in 6 m concrete framed wall (MA2).
location and the thermal load required for the formation of first
cracking was also similar to MA3. Therefore, we can conclude that
a lintel is not able to avoid or divert the cracking. Hence tying
the columns by a lintel is not a solution for this type of thermal
cracking.
3 7 8 8
2
9
6
1
Fig. 15. Cracking of 6 m two bay framed wall with opening and short lintel (MA5).
Fig. 16. Possibility of cracking identified by uncoupled linear elastic analysis using SAP2000.
slab top was at 57.5 °C another horizontal crack started under significant influence of stress relaxation after the first crack on the
the beam in the same right-hand panel (‘‘8’’) with a downward formation of subsequent cracks.
inclined diagonal part at the panel edge (‘‘9’’). At the time the slab Since the analysis done using SAP2000 as an uncoupled linear
reached its maximum gradient condition, it was seen that some of elastic analysis is not able to model stress relaxation and its
the elements consisted of cracks perpendicular to the second and effect on the formation of subsequent cracks, the results do not
third principal stresses as well (Fig. 15). give a clear picture about the location and pattern of cracking
It should be noted that this assembly was also subjected to in walls with openings and lintels. Therefore, especially when
an uncoupled linear elastic analysis using SAP2000 [23]. The analysing structures with complicated structural features, coupled
possibility of cracking (i.e. areas in the masonry and interface non-linear analysis will be essential with respect to predicting
elements where stresses exceeded failure boundaries) identified in actual phenomena.
the 6 m long two bay wall with an opening is shown in Fig. 16. The The effect of a lintel was further studied using MA6 (lintel from
survey observation of cracking in a wall with a similar structural left hand to centre column) and MA7 (lintel from left-hand to
arrangement is shown in Fig. 17, where care was taken to establish right-hand column). The results from both configurations are very
that the cause of the cracking observed was in fact due to thermal similar (see Fig. 18 for MA7). The lintels have been able to increase
movements in the roof slab—i.e. according to the orientation of the thermal load requirement for crack initiation. The temperature
the building, the wall was not subject to direct thermal radiation; of the top surface is 48.6 °C in both MA6 and MA7, compared to the
the crack had reappeared after repair; diurnal variation of the 45 °C in the previous case (i.e. MA5). Also in both MA6 and MA7
crack width was observed; and soil investigation reports of the site the formation of vertical cracking close to the central column at
showed that the building was founded on firm soil, with settlement the right-hand panel has been avoided. However, cracking around
very unlikely [25]. the opening seems to be more severe than in the previous case (i.e.
According to the results of our non-linear numerical modelling, MA 5).
cracking of the wall starts at the top right corner of the opening. The cracking in MA7 started at the top right corner of
After the formation of the first crack the stress relaxation in the the opening and propagated horizontally under the lintel (‘‘1’’).
wall has been captured by the non-linear modelling. This stress Meanwhile another crack started at the bottom left corner of
relaxation has reduced the stresses generated under the beam in the opening (slab top at 49.6 °C) and propagated horizontally
the left-hand panel. As a result the possibility of cracking in that towards the panel edge (‘‘2’’). Then the cracking in the right-hand
area has disappeared. Also there is a crack at the bottom left corner panel started when the slab top reached 54.2 °C. The cracking
of the opening which has not been identified by the linear elastic started under the beam and propagated horizontally (‘‘3’’) with
modelling, but observed in the building survey too. The first crack a downward diagonal part at the panel edge (‘‘4’’). This diagonal
has therefore increased the stresses in that area while reducing crack has turned to a vertically downward direction (‘‘5’’) when
stresses under the beam. These observations clearly indicate the it met the wall–column interface. After that another crack, which
1420 K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422
Fig. 17. Crack observations at the shopping complex of West Bus Park, Awissawella, Sri Lanka (Cracks have been enhanced for clarity).
3
6 3
4
8 7 5
Fig. 18. Cracking of 6 m two bay framed wall with opening and lintel from the left column to the right column (MA7).
started close to the top left corner of the internal column (‘‘6’’) panel edge. When the thermal load of the roof slab corresponded
when the slab top reached 57 °C propagated downward. When to 45 °C at the top surface of the slab the cracking at the bottom
the crack reached almost the lintel level, it turned diagonally (‘‘7’’) left corner of the opening started and developed horizontally. The
toward the opening and propagated horizontally (‘‘8’’) along the cracking in the right-hand panel started at the top left corner near
top of the lintel until it reached a point closed to the top left edge the central column when the slab top temperature reached 53.5 °C.
of the opening. After that the crack at the bottom left corner of In this case, even though the initial cracking started earlier than in
the opening started to propagate diagonally downwards (‘‘9’’). The MA6 and MA7 (see Table 4), the thermal load requirement for the
cracking perpendicular to the second and third principal stresses formation of subsequent cracks was higher.
could not be seen in this case.
The crack pattern on MA6 was also similar. The load required 3.6. Remedial measures for thermal cracking in walls
for forming a crack in the right-hand panel (when the slab top
reached 53.5 °C) was lower than in the case of MA7. However, It is necessary to discover methods to minimise this cracking.
crack development was not more extensive than in MA7. Cracking It was already established that the use of lintels does not seem
perpendicular to the second and third principal stresses was not to be effective. For the framed walls, separation of the masonry
evident in this case too. The cracks in the left-hand panel had a wall from the concrete beam and columns at its upper levels was
similar pattern to those in MA7, although the load requirement for considered a possible solution.
crack propagation was slightly lower. The effect was first studied on the two bay framed wall without
The pattern of cracking for the case where the height of lintels and openings. The separation was done at the wall–beam
the opening extends up to the level of concrete beam (MA8) is interface and the top part of the column–wall interface. The
somewhat similar to the case MA5 (having a small opening with a assembly MA9 was used to study the effect of separation at the
short lintel). In MA8 the cracking started very early at the top right wall–beam interface alone. In this case cracks were only developed
corner of the opening (when the slab top reached 37.7 °C). Then vertically (up to 1 m length) at the top part of the left and
this crack developed horizontally under the beam towards the right wall–column interfaces (Fig. 19). The height needing to be
K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422 1421
Fig. 19. Cracking of 6 m two bay framed wall with separation at wall–beam interface (MA9).
Fig. 20. Cracking of 6 m two bay framed wall with opening, short lintel and separation (MA14).
separated at the column–wall interface was studied using MA10 separations at masonry–concrete interfaces. Therefore, according
(separation at wall–column is 1.0 m, i.e. 1/3 of the wall height), to the above numerical simulations, the partial separation of the
MA11 (separation at wall–column is 0.75 m, i.e. 1/4 of the wall masonry–concrete interface in concrete framed walls is a very
height) and MA12 (separation at wall–column is 0.6 m, i.e. 1/5 effective remedial measure to control the formation of cracks due
of the wall height). In these cases cracks developed only at the to thermal movements of an overlying slab. Although this may
bottom ends of the vertical separations. According to the results, be in conflict somewhat with provisions for seismic regions, the
an appropriate height of separation at the wall–column interface fact that only a limited separation is required may be deemed
can be considered to be around 1/3 of the wall height, since smaller acceptable even in such regions, especially because the separation
separation heights and separation only at the wall–beam interface is required only at the uppermost floor, just below the roof
did give rise to cracking of some length. Also the thermal load level.
requirement for the first crack initiation in MA10 is much higher It should be noted in passing that we have not proposed any
than in the other cases (see Table 4). The 1 m height separation remedial measure for the prevention of cracking in load bearing
was possible only if continuous lintels were not employed. masonry walls. While this aspect is being currently investigated
Therefore a separation height of 0.875 m was selected for further (using intermittent vertical separations), it may in fact not be
studies to find its effect on walls with openings and lintels. possible to find a practically acceptable structural remedy. Hence
The separation extended vertically from beam soffit to lintel top for such walls, providing thermal insulation in the slab or avoiding
level. exposed roof slabs may be the only solution. Conversely, if
Assembly MA13 was used to study the effect of wall separation an exposed roof slab without thermal insulation is envisaged,
from beam and column (down to 0.875 m from top) of two prevention of cracking in the underlying masonry wall may be
possible only if a concrete frame is used, with separation of the
bay walls, when one span contained a window that extended
masonry–concrete interfaces in the upper areas.
up to the beam. In this situation only small cracks (of lengths
0.375 m and 0.25 m) were noted at the bottom ends of the left
and right wall–column separations respectively. In the presence 4. Conclusions
of a small opening with a short lintel (MA14), the crack can
only be seen at the right-hand end (Fig. 20). In the configuration The following conclusions can be made.
having an opening and a continuous lintel from left column to 1. The pattern (orientation and location) of cracking depends
right column (MA15), instead of cracking at the bottom ends of significantly on whether the wall is load bearing or framed by
the wall–column separation, a small crack was formed at the the reinforced concrete elements. The presence of structural
bottom left-hand corner of the opening. In all these cases the features such as lintels and openings also has a considerable
cracking is very small compared to the cases that did not have effect on the formation of these cracks.
1422 K.G.S. Dilrukshi et al. / Engineering Structures 32 (2010) 1411–1422
2. Both types of structural arrangements give diagonal cracking [7] Senthival R, Lourenco PB. Finite element modelling of deformation character-
near the ends of the wall panels. The diagonal cracks in concrete istics of historical stone masonry shear walls. Eng Struct 2009;31:1930–43.
[8] Chaimoon K, Attard MM. Experimental and numerical investigation of
framed walls usually have inclinations of around 45° to the masonry under three-point bending (in-plane). Eng Struct 2009;31:103–12.
horizontal and in load bearing walls it is generally around 60°. [9] Massart TJ, Peerlings RHJ, Geers MGD. Structural damage analysis of masonry
3. Cracking around openings is virtually inevitable irrespective of walls using computational homogenization. Int J Damage Mech 2007;16:
199–226.
the structural form of the wall. The cracks form at both top and
[10] Middleton J, Pande GN, Liang JX, Kralj B. Some recent advances in computer
bottom corners of window openings. methods in structural masonry. In: Middleton J, Pande GN, editors. Computer
4. Structural–thermal coupled non-linear numerical modelling methods in structural masonry. UK: Books and Journals International Ltd.;
is able to reproduce the observed phenomena even in the 1991. p. 1–21.
[11] Pande GN, Liang JX, Middleton J. Equivalent elastic moduli for brick masonry.
presence of openings and lintels in walls, because such Comput Geotech 1989;8:243–65.
modelling can capture the relaxation of stress after the [12] Gouwei Ma, Hong Hao, Yong Lu. Homogenization of masonry using numerical
formation of cracks. For solid walls however, a linear elastic simulations. J Eng Mech 2001;127:421–31.
[13] Sager V, Holla L. The numerical analysis of brick structures subject to mine
analysis can be used to identify likely locations and possible
subsidence. Aust Civ Eng Trans 1994;CE36: IEAus.
initial directions of cracking. [14] Paul JF, Thomas EB. Three dimensional modelling and full scale testing of stone
5. Reasonably good correspondence has been found among the arch bridges. Comput Struct 2001;79:2645–62.
relevant results from survey, experimental and numerical (lin- [15] Carpinteri A, Invernizzi S, Lacidogna G. In situ damage assessment and
nonlinear modelling of a historical masonry tower. Eng Struct 2005;27:
ear and non-linear) approaches regarding the likely locations 387–95.
and directions of cracking in masonry walls due to thermal [16] Ali S, Page AW. Cracking analysis of solid concrete masonry subjected to
movements of an overlying slab. The numerical modelling can concentrated loads. ACI Struct J 1989;86:367–75.
[17] Athanasios DT. Finite element modelling of cracks and joints in discontinuous
therefore be seen as having been qualitatively validated.
structural systems. In: Proceedings of 16th ASCE engineering mechanics
6. The results of the non-linear numerical modelling illustrate that conference. 2003.
the use of a continuous lintel in a concrete framed wall is not an [18] Page AW. An experimental investigation of the biaxial strength of masonry. In:
effective solution to the problem of thermal cracking in walls. Proceedings of the sixth international brick masonry conference. 1980.
[19] Samarasinghe W. The in-plane failure of brickwork. Ph.D. thesis. University of
However, separating the wall from the concrete frame at the Edinburgh; 1980.
wall–beam interface and wall–column interface (for a depth of [20] Syrmakezis CA, Asteris PG. Masonry failure criterion under biaxial stress state.
1/3 of the wall height from the roof beam soffit level) seems to J Mater Civil Eng 2001;13:58–64.
[21] Lourenco PB, Borst De R, Rots JG. A plane stress softening plasticity model for
be an effective solution.
orthotropic materials. Int J Numer Methods Eng 1997;40:4033–57.
[22] Hamid RL, Shing PB. Interface model applied to fracture of masonry structures.
Acknowledgements ASCE J Struct Eng 1994;120:63–80.
[23] Chaimoon K, Mario MA, Tin-Loi F. Crack propagation due to time-dependent
creep in quasi-brittle materials under sustained loading. Comput Methods
Financial assistance from the National Science Foundation Appl Mech Eng 2007;197:1938–52.
(NSF/SCH/2004/02) and the University of Moratuwa (Senate [24] Dilrukshi KGS, Dias WPS. Field survey and numerical modelling of cracking in
Research Fund) is gratefully acknowledged. masonry walls due to thermal movements of an overlying slab. J Natl Sci Found
Sri Lanka 2008;36(3):205–13.
[25] Dilrukshi KGS. Numerical and physical modelling of cracks in masonry walls
References due to thermal movements of an overlying slab. Ph.D. thesis. Sri Lanka:
University of Moratuwa; 2008.
[1] Ibraham KS, Suter GT. Finite element study of thermal stresses in low-rise [26] Jayasinghe MTR. Load bearing brickwork construction for Sri Lanka. Colombo
concrete masonry walls. In: Proceedings of fifth north american masonry (Sri Lanka): STRAD Consultants (Pvt) Ltd.; 1997.
conference. 1990. [27] Jayasinghe MTR, Jayatunga PTP. Guidelines for framed buildings braced with
[2] Lourenco PB. Historical structures: Models and modelling. 1999. brick wall infills. Transactions – Part 2 – Institution of Engineers. Sri Lanka;
http://www.civil.uminho.pt/masonry/publications/update-webpage/1999- 1998. p. 45–57.
lourenco.pdf. [28] Rahman AH, Suter GT. An analytical study of extreme temperatures in masonry
[3] Attard MM, Nappi A, Tin-Loi F. Modelling fracture in masonry. ASCE J Struct building walls. Masonry Int 1993;6:89–95.
Eng 2007;133:1385–92. [29] ANSYS Theory reference. Documentation for ANSYS 11.0. ANSYS Inc. 2007.
[4] Rots JG. Numerical models in Diana. In: Rots JG, editor. Structural masonry— [30] Asteris PG. Finite element micro-modeling of unfilled frames. Electron J Struct
An experimental/numerical basis for practical design rules. Rotterdam Eng 2008;8:1–11.
(Netherlands): A.A. Balkem publishers; 1997. p. 46–95. [31] ANSYS Thermal analysis guide. Documentation for ANSYS 11.0. ANSYS Inc.
[5] Rots JG. Computer simulation of masonry fracture: Continuum and discontin- 2007.
uum models. In: Middleton J, Pande GN, editors. Computer methods in struc- [32] Syrmakezis CA, Antonopoulos AK, Mavruli OA. Analysis of historical masonry
tural masonry. UK: Books and Journals International Ltd.; 1991. p. 14–123. structures using three-dimensional solid elements. In: Proceedings of
[6] Lourenco PB, Rots JG. Multisurface interface model for analysis of masonry the tenth international conference on civil, structural and environmental
structures. J Eng Mech 1997;123:660–8. engineering computing. 2005.