Beruflich Dokumente
Kultur Dokumente
Abstract: In this study, a discrete element approach is used to simulate brickwork. Bricks are represented by blocks with Mohr-Coulomb
plasticity and strain softening. The mortar consists of elastic Voronoi blocks connected by elastoplastic contacts. First, the parameters for
brick and mortar are calibrated through a series of bending and compression tests conducted on brick and mortar samples, respectively. Then,
Downloaded from ascelibrary.org by Tufts University on 07/11/18. Copyright ASCE. For personal use only; all rights reserved.
shear tests on triplets are performed and Brazilian tests on disc samples are investigated by the numerical approach. Compared with results
from other researchers, the numerical model can not only reproduce accurate strength values of the samples but is also able to display the
damage development and fracture patterns of the brick as well as mortar. DOI: 10.1061/(ASCE)MT.1943-5533.0002431. © 2018 American
Society of Civil Engineers.
Author keywords: Brickwork; Shear test; Fracture and strength of mortar; Numerical simulation.
The generality of micromodeling of masonry structures for these All presented simulations within this paper are conducted with the
numerical models is that the mortar and brick-mortar interface were 2-dimensional DEM code UDEC (Itasca 2011). The constitutive
simply considered to be the interface represented by the contact or model for the bricks is a Mohr-Coulomb elastoplastic model with
the simple interface element (dashpot-spring element), and compres- strain softening. The bricks are represented by rectangular blocks
sive or tensile strength characteristics and the thickness of the mortar with the same dimensions as tested samples. The mortar is com-
were ignored. For example, the researchers (Sarhosis et al. 2015; posed of thousands of elastic Voronoi blocks (the average edge
Wang et al. 2016; Giamundo et al. 2014; Sarhosis and Sheng length of Voronoi blocks is about 2.5 mm, as shown in Fig. 1),
2014; Forgács et al. 2017; Tóth et al. 2009) who used the DEM which are connected by contacts (dark gray lines in Fig. 1). Blocks
software UDEC to simulate masonry structures chose zero-thickness for bricks and Voronoi cells for mortar are internally meshed by
interfaces to represent the microstructure of the mortar joints. small triangular elements with length of about 2 mm (light gray
However, due to the simplifications, these models are unable to sim- line in Fig. 1). Therefore, there are two types of contacts: one be-
ulate the detailed fracture process, failure patterns, and accurate tween brick and mortar and the other between Voronoi blocks
strength of mortar. Pina-Henriques and Lourenço (2006) proposed inside the mortar. A Mohr-Coulomb failure criterion with tension
an approach to simulate compressive damage of the masonry struc- cut-off and shear softening [Eqs. (1) and (2)] is used to describe
ture based on the consideration that the brick and mortar are both the behavior of the contacts. The stress-displacement relation of
composed of elastic Voronoi blocks. The advantage of this method the contacts is shown in Fig. 1. Below the ultimate tensile and shear
(Pina-Henriques and Lourenço 2006) is that the damage character- strength, the stress-displacement relations in the normal and shear
istics of the brick and mortar are exhibited. However, if the scale of directions are linear, governed by normal stiffness kn and shear stiff-
the model is big, the number of Voronoi blocks is quite large. ness ks , respectively. The damage process associated with unrecov-
The purpose of this study is not only to present a discrete erable deformations and propagation of cracks inside the mortar or
element–based numerical simulation approach to analyze shear tests between brick and mortar is realized by the breakdown of contacts
of brickwork considering the fracture and strength characteristics of and relative movement along or across the contacts
Fig. 1. Numerical modeling concept and stress-displacement behavior of contacts. Bricks are represented by rectangular blocks with small
triangular internal mesh; mortar is represented by Voronoi blocks, which are connected by Contact Type 1 and interfaces between brick and mortar
represented by Contact Type 2.
Fig. 4. Simulated compression test for mortar: (a) fracture pattern, and
model is selected to describe the contacts (mortar-mortar interfaces) (b) stress-strain relation.
between the Voronoi blocks. The Voronoi block and Contact Type 1
parameters (Fig. 1) are listed in Table 3. These parameters are the
result of a calibration procedure based on bending and compression
tests for the mortar. In the postfailure stage, the compressive stress dramatically de-
The simulation result of a compression test for mortar is shown creases with ongoing strain.
in Fig. 4. The size of the 2-dimensional model is 40 × 40 mm, The simulation result of the bending test is illustrated in Fig. 5.
which is the same as in the corresponding lab tests. As shown in The sample size for mortar is the same as for the brick. An obvious
Fig. 4(a), shear and tensile cracks are distributed inside the mortar main tensile crack is observed [see magnified view of fracture area
sample, resulting in final failure. The curve of compressive stress in Fig. 5(b)]. The fracture pattern is the same as the one for the brick
versus vertical strain shows typical brittle failure characteristics. bending test. The crack path is not a straight line, but a curve with
Fig. 5. Simulated bending test for mortar: (a) fracture pattern; (b) magnified view indicating crack propagation; and (c) bending stress versus vertical
displacement of midspan.
small deviations like that observed during the lab tests. The curve Table 4. Comparison between lab test and simulation results for mortar
of bending stress versus vertical displacement of the midspan Mortar Bending Compressive
[Fig. 5(c)] also shows a dramatic decrease of bending stress with sample strength (MPa) strength (MPa)
growing displacement (brittle failure).
1 0.48 1.70
Comparison of simulation and lab test results for mortar shows
1.54
(Table 4) that the relative errors between the simulation result and 2 0.52 1.47
the average value of the lab tests are 5.5 and 8.4% for bending and 1.72
compressive strength, respectively. 3 0.60 1.63
1.72
4 0.60 1.68
Numerical Simulations for Shear Tests on Triplets 1.58
Average 0.55 1.63
The dimensions of the numerical model and the boundary condi- Simulation result 0.52 1.78
tions of the shear tests on triplets (three bricks connected by two Source: Data from Pelà et al. (2017).
layers of mortar) are illustrated in Fig. 6. During the shear tests,
the normal pressure Pn is first loaded on the right and left sides
of the triplet. Then the bottom of the right and left bricks are fixed Fracture Pattern
in the vertical direction, and the vertical load P applied at the top
of middle brick increases slowly. As shown in Fig. 6(c), Contact The simulated fracture pattern of shear tests on triplets is illustrated
Type 2 exists between brick and mortar, which represents the in Fig. 7. The average edge length of the Voronoi blocks is 2.5 mm.
interface between them. In order to reproduce the fracture process The simulation results are used for comparison with lab test results
and the strength value during shear tests on triplets, the parameters (Pelà et al. 2017). As shown in Fig. 7(a), three types of cracks are
for Contact Type 2 were determined by calibration according to the observed in the numerical model: Tensile Crack 1 and Shear Crack
lab test results (Table 5). An Intel i7 CPU with 3.60 GHz was used 3 are inside the mortar, whereas Shear Crack 2 propagates along the
for simulation and it took about 10 h to simulate a shear test of the boundary between brick and mortar. It can be concluded that shear
triplets. failure is the dominant pattern in the whole sample, but both tensile
Fig. 6. Numerical model of shear test on triplets: (a) model set-up; and (b and c) contact types.
Table 5. Contact Type 2 parameters for interface between mortar and brick Numerical Simulation of Disc Sample under
Parameter Symbol Value Unit Diametrical Loading
Normal stiffness kn 680 GPa=m
Shear stiffness ks 340 GPa=m Fracture Pattern
Tensile strength JT 0.2 MPa
The shear test on triplets is a standard test. However, it is practically
Residual tensile strength J Tr 0 MPa
Cohesive strength JC 0.3 MPa
difficult to extract this type of sample from existing masonry struc-
Residual cohesive strength J Cr 0 MPa tures. For this reason, disc samples based on in-situ core drilling
Friction angle ϕ 46 Degrees were proposed to do the shear tests (Pelà et al. 2017). For the
Residual friction angle φr 18.4 Degrees simulation of the shear failure of disc samples under diametrical
Contact dilation angle ψ2 15 Degrees loading—similar to the well-known Brazilian test (Fig. 9)—the
parameters for Contact Type 2 (interface between brick and mortar)
are the same as those given in Table 5. The diameter of the sample
is 90 mm and the compressive load is applied on the top of the
and shear cracks propagate inside the mortar. Fracture patterns of trip- sample in the vertical direction. The fracture pattern for α ¼ 40°
lets documented by other researchers are shown in Figs. 7(b and c). (α is the inclination angle of the mortar layer with respect to
Obviously, the simulated detailed fracture pattern and fracture loca- the horizontal) is illustrated in Fig. 9. Shear Crack 2 develops be-
tions in the mortar and along the interface are very similar to those tween the brick and the mortar, Tensile Crack 1 propagates inside
observed during lab tests. the mortar, and Shear Crack 3 is observed inside the mortar. It can
be concluded that shear failure dominates the whole sample and a
shear band develops in the mortar with increasing load. The simu-
Shear Strength lated fracture pattern almost duplicates the failure pattern observed
The simulated stress-displacement behavior under different normal during lab testing.
pressures is shown in Fig. 8(a). It can be concluded that the shear
strength increases with increasing normal pressure. Significant Shear Strength
residual strength is observed because frictional resistance is acti-
vated in the postfailure stage. A comparison between simulation The simulated curves of shear stress-loading plate displacement for
and lab test results (Pelà et al. 2017) with respect to the maximum different α values are shown in Fig. 10(a). The maximum shear
shear stress (shear strength) is illustrated in Fig. 8(b). The average stress increases with decreasing α. A residual shear strength is also
shear strength values obtained by lab tests under normal pressure of observed due to activation of frictional resistance. As shown in
0.3, 0.6, and 1.0 MPa are 0.26, 0.45, and 0.68 MPa, respectively. Fig. 10(b), under even the same α, the lab test data show a large
The corresponding simulation results for shear strength are 0.29, scatter. This means that the efficiency and accuracy of this type of
0.45, and 0.62 MPa, respectively. The relative errors between testing are not better than those of the shear tests on triplets. Never-
lab test and simulation results are 11.5, 0, and 8.8%, respectively. theless, the simulation results are inside the range of the test data,
Fig. 12. Fracture pattern of shear tests on triplets with 1.0 MPa normal
pressure considering shape effect. Average edge length of the Voronoi
blocks is 2.5 mm and the Voronoi structure is regenerated comparing
with Fig. 7(a).
Fig. 10. Simulation results for disc under diametrical loading: (a) shear
stress versus vertical displacement of loading plate for different; and Fig. 13. Fracture pattern of shear tests on triplets with 1.0 MPa normal
(b) comparison between simulation and lab test results. (Data from pressure considering size effect. Average edge length of the Voronoi
Pelà et al. 2017.) blocks is 5.0 mm.
corresponding size of Voronoi blocks. Messali, F., G. Metelli, and G. Plizzari. 2017. “Experimental results on the
retrofitting of hollow brick masonry walls with reinforced high perfor-
mance mortar coatings.” Constr. Build. Mater. 141: 619–630. https://
doi.org/10.1016/j.conbuildmat.2017.03.112.
Acknowledgments Milani, E., G. Milani, and A. Tralli. 2008. “Limit analysis of masonry
vaults by means of curved shell Finite Elements and homogenization.”
This work was supported by the National Natural Science Int. J. Solids Struct. 45 (20): 5258–5288. https://doi.org/10.1016/j
Foundation of China (NSFC, Grant Nos. 51608537 and 51538009), .ijsolstr.2008.05.019.
China Postdoctoral Science Foundation (Grant No. 2017M610508), Milani, G. 2008. “3D upper bound limit analysis of multi-leaf masonry
and Postdoctoral Foundation of Central South University. The walls.” Int. J. Mech. Sci. 50 (4): 817–836. https://doi.org/10.1016/j
constructive comments by anonymous reviewers are appreciated. .ijmecsci.2007.11.003.
Milani, G. 2015. “Upper bound sequential linear programming mesh adap-
tation scheme for collapse analysis of masonry vaults.” Adv. Eng. Soft-
ware 79: 91–110. https://doi.org/10.1016/j.advengsoft.2014.09.004.
References Milani, G., and P. B. Lourenço. 2009. “A discontinuous quasi-upper
bound limit analysis approach with sequential linear programming
Baraldi, D., and A. Cecchi. 2017. “Discrete and continuous models for
mesh adaptation.” Int. J. Mech. Sci. 51 (1): 89–104. https://doi.org/10
static and modal analysis of out of plane loaded masonry.” Comput. .1016/j.ijmecsci.2008.10.010.
Struct, in press. https://doi.org/10.1016/j.compstruc.2017.03.015.
Milani, G., and A. Taliercio. 2015. “In-plane failure surfaces for masonry
Beatini, V., G. Royer-Carfagni, and A. Tasora. 2017. “A regularized with joints of finite thickness estimated by a method of cells-type
non-smooth contact dynamics approach for architectural masonry approach.” Comput. Struct. 150: 34–51. https://doi.org/10.1016/j
structures.” Comput. Struct. 187: 88–100. https://doi.org/10.1016/j .compstruc.2014.12.007.
.compstruc.2017.02.002. Minghini, F., G. Milani, and A. Tralli. 2014. “Seismic risk assessment
Bednarz, L., A. Górski, J. Jasieńko, and E. Rusiński. 2011. “Simulations of a 50 m high masonry chimney using advanced analysis techniques.”
and analyses of arched brick structures.” Autom. Constr. 20 (7): Eng. Struct. 69: 255–270. https://doi.org/10.1016/j.engstruct.2014.03
741–754. https://doi.org/10.1016/j.autcon.2011.01.005. .028.
Çaktı, E., Ö. Saygılı, J. V. Lemos, and C. S. Oliveira. 2016. “Discrete element Pelà, L., K. Kasioumi, and P. Roc. 2017. “Experimental evaluation of
modeling of a scaled masonry structure and its validation.” Eng. Struct. the shear strength of aerial lime mortar brickwork by standard tests
126: 224–236. https://doi.org/10.1016/j.engstruct.2016.07.044. on triplets and non-standard tests on core samples.” Eng. Struct. 136:
Caliò, I., M. Marletta, and B. Pantò. 2012. “A new discrete element 441–453. https://doi.org/10.1016/j.engstruct.2017.01.028.
model for the evaluation of the seismic behaviour of unreinforced Pina-Henriques, J., and P. B. Lourenço. 2006. “Masonry compression:
masonry buildings.” Eng. Struct. 40: 327–338. https://doi.org/10 A numerical investigation at the meso-level.” Eng. Comput. 23 (4):
.1016/j.engstruct.2012.02.039. 382–407. https://doi.org/10.1108/02644400610661163.
Casolo, S., and G. Milani. 2010. “A simplified homogenization-discrete Rafiee, A., and M. Vinches. 2013. “Mechanical behaviour of a stone ma-
element model for the non-linear static analysis of masonry walls sonry bridge assessed using an implicit discrete element method.” Eng.
out-of-plane loaded.” Eng. Struct. 32 (8): 2352–2366. https://doi.org/10 Struct. 48: 739–749. https://doi.org/10.1016/j.engstruct.2012.11.035.
.1016/j.engstruct.2010.04.010. Sarhosis, V., K. Bagi, J. V. Lemos, and G. Milani. 2016. Computational
Casolo, S., and G. Milani. 2013. “Simplified out-of-plane modeling of modeling of masonry structures using the discrete element method.
three-leaf masonry walls accounting for the material texture.” Constr. Hershey, PA: IGI Global.
Build. Mater. 40: 330–351. https://doi.org/10.1016/j.conbuildmat.2012 Sarhosis, V., S. W. Garrity, and Y. Sheng. 2015. “Influence of brick-mortar
.09.090. interface on the mechanical behaviour of low bond strength masonry
Cecchi, A., and K. Sab. 2002. “A multi-parameter homogenization study brickwork lintels.” Eng. Struct. 88: 1–11. https://doi.org/10.1016/j
for modeling elastic masonry.” Eur. J. Mech. A. Solids 21 (2): 249–268. .engstruct.2014.12.014.
https://doi.org/10.1016/S0997-7538(01)01195-0. Sarhosis, V., D. Oliveira, J. Lemos, and P. Lourenco. 2014. “The effect of
Chen, W., H. Konietzky, C. Liu, and X. Tan. 2018. “Hydraulic fracturing skew angle on the mechanical behaviour of masonry arches.” Mech.
simulation for heterogeneous granite by discrete element method.” Com- Res. Commun. 61: 53–59. https://doi.org/10.1016/j.mechrescom.2014
put. Geotech. 95: 1–15. https://doi.org/10.1016/j.compgeo.2017.11.016. .07.008.
Chen, W., H. Konietzky, X. Tan, and T. Frühwirt. 2016. “Pre-failure Sarhosis, V., and Y. Sheng. 2014. “Identification of material parameters for
damage analysis for brittle rocks under triaxial compression.” Comput. low bond strength masonry.” Eng. Struct. 60: 100–110. https://doi.org
Geotech. 74: 45–55. https://doi.org/10.1016/j.compgeo.2015.11.018. /10.1016/j.engstruct.2013.12.013.
Cundall, P. A. 1971. “A computer model for simulating progressive large Shermi, C., and R. N. Dubey. 2017. “Study on out-of-plane behaviour of
scale movements in blocky rock systems.” In Proc., Symp. of the Int. unreinforced masonry strengthened with welded wire mesh and
Society of Rock Mechanics. Nancy, France: International Society for mortar.” Const. Build. Mater. 143: 104–120. https://doi.org/10.1016/j
Rock Mechanics. .conbuildmat.2017.03.002.