You are on page 1of 39

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/245303596

Nonlinear Dynamic Analysis of Unreinforced


Masonry

Article in Journal of Structural Engineering March 1998


DOI: 10.1061/(ASCE)0733-9445(1998)124:3(270)

CITATIONS READS

36 72

3 authors, including:

Yan Zhuge David Thambiratnam


University of South Australia Queensland University of Technology
69 PUBLICATIONS 584 CITATIONS 317 PUBLICATIONS 2,320 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Domeshell projects View project

Level Crossing Accidents Mitigator View project

All content following this page was uploaded by Yan Zhuge on 30 September 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
NONLINEAR DYNAMIC ANALYSIS OF UNREINFORCED MASONRY

By Yan Zhuge1, David Thambiratnam,2 Fellow, ASCE and John Corderoy3

Abstract: A comprehensive analytical model has been developed and presented in this paper to study the

response of unreinforced masonry to in-plane dynamic loads, including earthquake loads. The analysis has been

implemented in a nonlinear finite element program. Masonry is treated as a nonlinear homogeneous orthotropic

material. A failure envelope has also been developed which is capable of predicting both joint sliding and

cracking and/or crushing type of failure. The effect of bed joint orientation has been considered and this is

achieved through an ubiquitous joint model. The model is capable of performing both static and time history

analyses of masonry structures. Nonlinear dynamic analysis is carried out using the Modified Newton-Raphson

iteration scheme in conjunction with the Newmark time integration algorithm. In order to calibrate the model and

to demonstrate its applications, several numerical examples have been treated and the results have been

compared with those from full-scale tests on masonry shear walls under both cyclic and dynamic loads.

Reasonably good agreement has been found between the analytical and experimental results.

INTRODUCTION

Unreinforced masonry (URM) structures are widely used throughout Australia and other regions around the

world, even though they seem to have experienced the worst damage during earthquakes. However, it has been

found that even when unreinforced, masonry has a substantial deformation capacity after cracking, if it is

designed with suitable compressive loads and material properties. Therefore, it is necessary to study the dynamic

behaviour of masonry to provide a better understanding of its earthquake response.

1
Lecturer, School of Engrg, Univ. of South Australia, The Levels, SA 5095, Australia. Tel: 61-8-8302 3140
2
Prof., School of Civ. Engrg, Queensland Univ. of Tech., Queensland, 4001, Australia.
3
Prof. and Pro Vice Chancellor (R&A), Queensland Univ. of Tech., Queensland, 4001, Australia
Key words: masonry, nonlinear dynamic analysis, finite element, modelling, earthquake
The finite element method is a powerful numerical method for the analysis of structures and it has been

extensively used for analysing the behaviour of masonry structures in the last decade. Due to the nature of the

problem, masonry can be described as a two-phase material consisting of an elastic brick set in an inelastic mortar

joint matrix. A two-phase nonhomogeneous finite element model (micro-modelling of brick and mortar) of

masonry has been used by Page (1978), Lourenco and Rots(1994) and Lotfi and Shing (1994). However, such a

model undoubtedly made analysis more complex. To simplify the problem, a homogeneous one-phase material

(macro-) model has been used by some investigators, such as Samarasinghe et al. (1981), Dhanasekar et al.

(1984) and others.

In Engineering literature, reference to seismic/dynamic modelling and analysis of masonry has been rare. A

review of previous studies on dynamic behaviour of masonry has revealed that most research work has focused on

experimental investigations (Tomazevic and Lutman, 1996; Klopp and Griffith, 1994).

For finite element modelling of masonry under dynamic loads, a single degree of freedom model is still commonly

used (Jankulovski and Parsanejad, 1994). This model can be used to study the global behaviour of a building

system, but the local response of masonry walls, such as cracking, crack propagation, crushing etc. cannot be

represented by the model.

A two-dimensional finite element model has been employed by some investigators to study the cyclic behaviour

of masonry (LaRovere, 1990 and Vratsanou, 1991). However, the analyses were under pseudo dynamic loads,

which essentially means that a static analysis was performed with cyclic stress-strain relations.

It is arguable that the behaviour of URM is much more complex than that of concrete. Masonry is a two-phase

material and its properties are therefore dependent upon the properties of its constituents, the brick and the mortar.

The influence of mortar joint as a plane of weakness is a significant feature which is not present in concrete and

this makes the numerical modelling of URM very difficult.


In order to predict the complex behaviour of URM walls under in-plane dynamic loads, including earthquake

excitation, a comprehensive analytical model has been developed by the authors and is implemented in a nonlinear

finite element program (Zhuge, 1995). The model is capable of performing both static and time history analyses

of masonry structures and has been calibrated by using results from experimental testing of several masonry

panels.

A homogeneous finite element model has been adopted herein as the aim of the paper is to investigate the seismic

response of unreinforced masonry. First, an anisotropic material model which was originally derived for concrete

by Darwin and Pecknold (1977), is developed to analyse unreinforced masonry under various states of stresses

before and post-failure. The material model is then combined with a two-dimensional nonlinear ubiquitous joint

finite element model to analyse masonry subjected to in-plane static and dynamic loads. A failure criterion has

also been developed, which combines both biaxial and Coulomb shear failure models. The resultant failure model

is capable of predicting joint failure and this is achieved through the use of the ubiquitous joint model.

The results from the proposed analytical model are then compared with several experimental findings with

different failure modes. The development and application of this model are treated in this paper.

MATERIAL MODEL FOR MASONRY UNDER DYNAMIC LOADS

Cyclic Constitutive Relations for Masonry

The stress-strain relations of masonry depend on the state of stress. Previous experimental results have shown that

under biaxial compression-compression and uniaxial compression, masonry behaves in a nonlinear and

anisotropic manner. While under biaxial tension-compression or uniaxial tension, no such nonlinearity was

apparent and the walls might be regarded to be isotropic (Dhanasekar, 1984). Therefore, an anisotropic material

model for URM under biaxial compression-compression or uniaxial compression has been developed.
Considering a plane-stress state and referring to the principal stress axes, 1 and 2, the stress-strain relations in an

orthotropic form can be obtained in incremental form as

d 1 E1 E1 E 2 0 d1
1
d 2 = E2 0 d 2 (1)
d 1
2

12
sym
(1 2 )G d 12

where E1, 1 and E2, 2 are the elastic modulus and Poisson's ratio of masonry in the principal directions 1 and 2

respectively, the equivalent Poissons ratio is given by 2=12 and G is the in-plane shear modulus of

masonry.

When E1= E2= Em= the Youngs modulus of masonry, and G= Em/2(1+), Eq. 1 depicts the constitutive matrix

of a linear elastic isotropic material in a plane stress state.

If the shear modulus G is expressed as a function of E1, E2 and , as shown in Eq. 2, the constitutive matrix in Eq.

1 can be defined by these three properties (E1, E2 and ), which depend on the state of stress and strain.

1
(1 2 )G = ( E1 + E 2 2 E1 E 2 ) (2)
4

From Eq. 1, tangent moduli Et1, Et2 in the principal stress directions can be determined from the uniaxial stress

strain curve by applying an "equivalent uniaxial strain" concept which was first introduced by Darwin and

Pecknold (1977) for concrete. By using this technique, the actual biaxial stress-strain curves can be duplicated

from "uniaxial" curves developed in an experimental investigation, especially when the masonry structure is

under cyclic loads.

By using this method, the tangent modulus Eti, i=1,2, in the current principal stress direction i, is determined from

a family of i - iu curves, in which i is the principal stress and iu is the equivalent uniaxial strain in the i
direction and can be expressed as shown in Eq. 3. Therefore, the constitutive model can be developed in such a

way that the explicit relationship is determined completely by conventional parameters, i.e. the uniaxial

compressive strength f'c, uniaxial crushing strain cu, and the initial tangent modulus E0, see Figure 1.

i
iu = (3)

1 j
i

In the above equation i is the principal strain in the i direction.

Under uniaxial compression or under a biaxial compressive-compressive stress state of the material, nonlinear

material characteristics are presented and a more complicated stress-strain relation has to be derived. Naraine and

Sinha (1991) provided the first detailed information on the constitutive relation of unreinforced masonry panels

subjected to uniaxial and biaxial cyclic compression loading. Based on the experimental data, they proposed an

exponential equation:


= ( ) exp(1 ) (4)
c c c

where c and c are the compressive strength and corresponding strain respectively.

Eq. 4 can be expressed as follows in biaxial stress state:

i
= ( iu ) exp(1 iu ) (5)
ic ic ic

where i = the principal stress;

iu = the equivalent uniaxial strain in the i direction;

ic = maximum stress associated with direction i;

ic = corresponding strain in i direction.


This curve has initial slope E0 and passes through the point C (ic, ic) of maximum stress, which is a function of

the continuously changing principal stress ratio =i/j, the strength fc and strain cu. The Poisson effect is

contained in the parameter iu.

The initial tangent modulus E0 is given as:

d i ic
E0= iu = 0 = exp( ) (6)
d iu ic

For a given stress level, the value Eit for use in material stiffness matrix can be obtained by differentiating Eq. 5:

iu
d i (1 )
Eit = = exp ic ic (1 iu ) (7)
d iu ic ic

The peak stress ic is obtained from the failure envelope and will be discussed in the next section.

Failure Criteria for Brick Masonry

The failure of masonry walls has been investigated for many years. Unlike other materials (concrete or steel),

masonry is a non-homogeneous material. Therefore, it is not possible to attribute failure of masonry to a single

cause. Many types of failure are possible and the one that gives the lowest bound is the critical one. Furthermore,

masonry exhibits distinct directional properties due to mortar joints acting as planes of weakness. Depending

upon the stress state acting on the joints, failure may occur in the joints alone, or in some form of combined

mechanism involving the mortar and the masonry unit. Therefore, to develop a rational failure criteria, all the

above factors must be taken into account.


Multilaminate Model Formulation - Ubiquitous Joint Model

The ubiquitous joint model is an anisotropic plasticity model which assumes a series of weak planes embedded in

a Mohr-Coulomb solid. The model was first developed by Zienkiewicz and Pande (1977) for analysing rock

material with multiple planes of weakness, see Figure 2.

In this model, failure may occur in either the solid or along the slip (weak) plane, or both, depending on the

material properties of the solid and weak plane, the stress state and the angle of the weak planes. Herein the

"weak plane" signifies that parameters for failure along these planes are considerably smaller (weaker) compared

with those of the basic material matrix.

When applying this model to masonry, initially, the stresses may be computed assuming linear, elastic behaviour

everywhere. An examination of sliding, cracking and crushing conditions in any element is then possible if a

direction is specified for the joint plane (weak plane) in that element.

In order to predict the shear slip type of failure along the joints, the stresses can be transformed to directions

perpendicular and along the planes of weakness. For a weak plane k, the stresses may be expressed as:

x ( k )

k = y ( k ) (8)
( k )
xy

These stresses along the joint are then examined for shear sliding by Coulomb type failure criterion:

xy ( k ) < mxy ( k ) (9)


where mxy ( k ) = 0 + y ( k )

0 = shear bond strength, and

= friction coefficient at the brick mortar interface.

However, previous research had shown that Eq. 9 is applicable only when the normal stress n is less than

approximately 2 MPa.

The common method of predicting the failure of concrete and masonry is the biaxial failure criterion (defined in

terms of principal stresses 1 and 2). Since masonry is a brittle material similar in properties to concrete

(ignoring the effect of mortar joints acting as planes of weakness), conventional concrete failure criteria have been

adopted by many researchers and modified for masonry. However, biaxial cyclic compressive tests conducted by

Naraine and Sinha (1991) found that a failure surface described by Eq. 10 provides a better fit to the experimental

data. This failure surface has been adopted in the present study for crushing type of failure.

CJ2 + (1 - C)I1 + CI2 = 1 (10)

where J2, I1 and I2 are principal stress invariant defined as:

1 2
J2 = ( '
'
)2
f cn f cp

1 2
I1 = ( '
+ )
f cn fcp'
1 2
I2 = ( '
)
f cn fcp'

fcn and fcp = uniaxial compressive strength normal and parallel to the bed joint, respectively; C=1.6.

It should be noted that when C=1, Eq. 10 reduces to the Von Mises failure criterion.

Simplified Failure Envelopes For Brick Masonry


In order to illustrate the failure criteria discussed in this paper, the biaxial failure criterion (biaxial tension-

compression range) in terms of the principal stress system are transformed to direct stresses normal (n) and

parallel (p) to the bed joint and shear stress () along the bed joint. The failure envelope has been shown in

Figure 3 with various stress ratio (2/1= 5, 10, 15). As shown in the figure, the lowest failure curve defines the

governing failure condition for any stress combination. The uniaxial compressive strength fm is assumed equal to

10ft (uniaxial tensile strength) in this envelope. It is a simple matter to modify this for other compressive/tensile

strength ratios. The straight line describes the shear sliding and the curves describe cracking type of failure.

It can be seen from this figure that when 45 < 90 (low ratio of y/), failure is controlled by the shear failure

criterion (straight line, Eq. 9) which agrees well with the test results of Page (1982), showing that the crack

developed along the bed joints due to shear slips in this range. When the bed joint angle decreased, the shear

strength of the wall increased as n increased. If the shear strength exceeds the principal tensile stress ( < 45),

the failure will be controlled by the biaxial failure criterion (principal tensile stress).

NONLINEAR DYNAMIC ANALYSIS

In order to perform nonlinear dynamic analysis, the constitutive law developed for monotonic loads will be

extended to allow for fully reversed cyclic loads, such as seismic loads. The analytical model has been developed

for carrying out time history analysis of URM under seismic loads.

Solution of Dynamic Equilibrium Equations

The governing equation for the dynamic analysis of a structure is given by:

.. .
[ M ]{U } + [C]{U } + [ K ]{U } = {P} (11)
where [M], [C] and [K] are the mass, damping, and stiffness matrices; {P} is the external load vector and {U},

. ..
{U } and {U } are the displacement, velocity, and acceleration vectors.

For dynamic analysis, it is necessary to derive the element mass matrix in addition to the stiffness matrix. The

element mass matrix is a matrix of equivalent nodal masses that dynamically represent the actual distributed mass

of the element. The mass matrix and stiffness matrix can be derived by using Lagrange's equations. If the

.. ..
structure is only subjected to ground acceleration (u g , v g ) , the external load vector is (P1, P2,......,Pi,...., Pn),

where

u..
Pi = mgi ..
g
(12)

vg

and mgi is the mass matrix for the ith element.

Nonlinear dynamic analysis is carried out with the modified Newton-Raphson iteration scheme in conjunction

with the Newmark direct time integration algorithm. In direct integration, equation (11) is integrated using a

numerical step-by-step procedure, the term "direct" meaning that prior to the numerical integration, no

transformation of the equations into a different form is carried out.

Tensile Unloading/Reloading Curves for Masonry

Figure 4 shows two typical elastic and secant type of unloading or reloading options for masonry under cyclic

loading. For elastic unloading, the crack closes immediately upon a strain reversal, whereafter further strain-

decomposition is terminated and elastic behaviour is resumed (line BE). For secant unloading (line BO), the

stress follows a straight line back to the origin. The crack normal strain is reversible and upon reaching the origin

of the diagram, the crack truly closes and n=0, whereafter elastic behaviour is recovered (Rots, 1988).
The tensile unloading/reloading constitutive relations can be defined as follows: before the peak strain

( iu cr ), the elastic modulus of masonry is assumed for the linear unloading/reloading path; the secant

unloading/reloading model is used beyond the peak, as it provides a better approximation to reality. The secant

model accounts for the decrease of stiffness with increasing crack opening strain.

It is assumed that the static tensile stress-strain curve provides an envelope for the cyclic curve. If the strain

exceeds cr, the crack starts to open and the stress drops along the envelope curve. The unloading starts at point

B, where =cr and (1a) is defined as an unloading parameter which can be determined by calibrating the

finite element model with experimental results. Usually, a large amount of experimental testing will be required

to determine the value of this unloading parameter. The reloading path follows the same secant modulus as

unloading, back up to point B and proceeds along the envelope curve A-C up to the point of next unloading.

The smeared crack model has been used for crack modelling and the possibility of crack closing and opening and

the formation of secondary cracks have been considered. The cyclic tests carried out elsewhere (Zhuge, 1995)

have shown that under reversed loading, reopening and closing of cracks may take place repeatedly and this

feature is incorporated into the analytical model. When unloading starts after cracking, the usual constitutive

matrix [Df] can be replaced by:

Es 0 0
[D f ] = 0 E 0 (13)

0 0 ' G

where Es is the secant modulus of the unloading branch, as shown in Figure 4.

It is also assumed that (a) during unloading if 1 becomes zero, the crack is considered to be closed and elastic

behaviour is recovered and (b) during reloading a crack reopens and follows the same path as unloading, when the

stress in the direction normal to the crack becomes positive.


NUMERICAL EXAMPLES

The proposed analytical model was implemented into a finite element program and verified by comparing the

results with experimental data.

As a first example, the experimental results on a full scale unreinforced masonry wall tested by Calvi and

Magenes (1991) at the University of Pavia, Italy, under shear and compression, were considered. The test wall

panel has a width/height ratio of 0.75 (1500 x 2000 mm) and a thickness t = 380mm. The wall was subjected to a

vertical compression stress m = 1.2 MPa. The material properties are summarised in Table 1.

During the testing, the vertical compressive stress was kept constant whereas the horizontal load was increased

gradually in increments up to failure. When the horizontal load was increased up to 259.3KN, major diagonal

cracking occurred and the wall failed. The proposed analytical model was used to simulate this test

Comparisons are made for the failure load, the load-deflection behaviour and the failure patterns. The analytical

lateral load vs. displacement curve is plotted in Figure 5 and compared with the experimental one. It can be seen

that the results from the analytical model proposed in this paper agree well with those from the experiments. The

final failure was assumed in the program when the solution failed to converge. The ultimate load predicted by the

analysis is 250KN which agrees reasonably well with the experimental result of 259.3KN.

The final failure pattern of the wall is presented in Figure 6 and it is consistent with the experimental

observations. It can be seen from Figure 6 that the final crack is in diagonal shear connecting the toe and the

horizontally loaded point of the wall. The crushing type of failure did not occur in this case.

The experiments carried out by the authors (Zhuge et al, 1994) were simulated as the second example. In these

experiments, twelve full-scale unreinforced masonry walls were tested under combined compression and racking

(cyclic) loads. The wall specimens were divided into two groups: one group had an aspect ratio equal to 1.94
(930x480mm), while the other equal to 1.46 (1400x960mm). The vertical compressive stress varied from

0.34MPa to 0.45MPa. The experimental details can be found in Zhuge (1995). Analytical peak loads are

compared with the experimental peak loads from positive and negative cycles in Table 2.

Further analytical results are compared with the test results of wall 2 with an aspect ratio of 1.46 and compressive

stress equal to 0.45MPa. The material properties of the wall are shown in Table 3. The lateral load vs

displacement envelope obtained from the analytical model is shown in Figure 7 and compared with the

experimental hysteresis curves. It can be seen that there is reasonably good agreement in the results for the failure

load and deflection for this wall.

The experimental results have shown that the failure pattern of wall 2 is by rocking, where flexural cracks

initiated from both ends of the wall and developed along the interface of concrete base and mortar. At failure, the

base of the brick wall separated from the concrete base and crushing also occurred at the right hand corner. It was

found during the testing, the wall had large deformation capacity after cracking with considerable energy

absorption capacity. However, there were 5 walls tested under the same conditions as wall 2 and the failure

patterns were different. 3 walls (including wall 2) failed by rocking, and 2 walls failed by diagonal shear. The

different failure patterns of these walls may be attributed to the variation in workmanship. It is found that the

failure pattern of URM walls changes from flexural cracking to stair-stepped diagonal shear depending on the

vertical compressive load. The vertical compressive load applied to these 5 walls may lie within some critical

state of two failure patterns. Therefore, the crack developments predicted by the model and shown in Figure 8,

are in between those of walls which failed by diagonal shear (example 1) and those which failed by rocking

(example 3). In practice, the failure pattern may be determined by the local strength of mortar in the particular

case. In Figure 8, cracks developed along the vertical edge of the wall on the left hand side and these were caused

by the tensile bending stress generated by lateral load. However, in the real wall these cracking zones would be

likely to appear as joint failure and might not be seen with the naked eye. Toe crushing was also predicted near

the ultimate load although it looks longer than the experimental observation.
In the third example, the nonlinear analytical model developed in this paper was used to simulate the dynamic

response of URM single-story walls, which had been tested experimentally at the University of Adelaide (Klopp

and Griffith, 1994).

Experiments were conducted on double clay brick wall specimens, which were unreinforced. Sinusoidal base

motion was used to simulate earthquake excitation. During the experiments, for each test specimen a fixed

frequency was used and the amplitude of the sine wave was increased until the specimen was deemed to have

failed. In every test, failure occurred by rocking of the panels when the wall specimens separated from the

reinforced concrete bases. Testing was stopped at this point (Klopp and Griffith, 1994).

The material properties are summarised in Table 4. Detailed verification can be found in (Zhuge,1995) and here,

the results from the analytical model are only compared with those from one test result, viz, that of wall 5. The

input excitation time history for wall 5 is shown in Figure 9. The wall has an aspect ratio of 1.58 and a vertical

compressive stress equal to 0.015MPa. The fundamental period of vibration of the uncracked wall is 0.081sec.

Measurements on various types of URM on a shaking platform reported by (Pomonis et al., 1992) have shown

values for damping ratios to be between 7% and 8% of critical damping. The damping ratio in the present study is

assumed to be 7% of critical damping. A time step of 0.001sec was chosen for the time history analysis as the

fundamental period is equal to 0.081sec.

The comparisons are based on the maximum horizontal displacement at each cycle of acceleration (because no

displacement time history was recorded during the experiments), maximum acceleration at failure and the failure

pattern.

The predicted and experimental results for wall 5 are presented and compared in Table 5 for different values of

tensile strength and the unloading parameter .


It can be seen from Table 5 that the proposed model provides reasonably good results with the combination of

ft=0.4MPa and =2.

The horizontal displacement at the top of the wall with ft= 0.4MPa is plotted as a function of time in Figure 10.

For comparison, the results from the corresponding linear analysis are also presented in the same Figure.

It can be seen from Figure 10 that after cracking initiated at 0.59sec, the maximum top displacement increased

significantly for the nonlinear analysis (=2), however, when =4 was used, more cracks developed quickly and

thus caused the displacement divergence. The wall failed at 0.62 sec for =4 just after cracking at 0.59sec,

whereas, when =2 was used, the wall had substantial deformation capacity after cracking and did not fail until

1.08sec.

The horizontal displacements at the top of the wall with different values of tensile strength are shown in Figure

11. It can be seen, as expected, that the tensile strength has a significant effect on the structural behaviour of

masonry. This is because the failure is dominated by the cracking of the material. When ft= 0.3MPa, the wall

failed just after cracking. At ft= 0.4MPa and 0.5MPa, the wall had substantial deformation capacity after

cracking.

The development of cracks for wall 5 is presented in Figure 12 (ft =0.4MPa, =2). The cracks initiated at both

ends of the wall and developed along the interface between the bottom mortar and the concrete base. The cracks

grew gradually and the final failure pattern was controlled by rocking - the base of the masonry wall separated

from the foundation at both ends, and this pattern agrees well with the experimental results.

CONCLUSIONS

In this paper, an analytical model for studying the response of unreinforced masonry under in-plane dynamic loads

has been presented. The material model for the masonry is described. The nonlinear behaviour of brick masonry
is caused by two major effects: - progressive local failure (cracking of the mortar) and nonlinear deformation

characteristics (in the biaxial compression-compression and uniaxial compression stress state). All these effects

are considered in the orthotropic constitutive relations developed in this research. A failure envelope has been

developed which is capable of predicting both joint sliding and cracking and /or crushing type of failure for a

homogeneous material model. The effect of bed joint orientation has been considered. A simple secant type

unloading/reloading curve is adopted for masonry under tension and the unloading parameter was determined

through calibrating the finite element model against experimental results.

The analytical model was validated by comparing results with various experimental results and reasonably good

agreement has been found. However, further research could improve the model. More experimental results are

required to validate the proposed model, especially for non-linear time history analysis and for studying the effect

of unloading parameter .

APPENDIX I. REFERENCES

Calvi, G. M. and Magenes G. (1991). Experimental evaluation of seismic strength of old masonry structures,

Proc. 9th Int. Brick/Block Masonry Conf., Dublin, Ireland, pp. 490-497.

Darwin, D. and Pecknold, D.A. (1977). Nonlinear biaxial stress-strain law for concrete, J. Engrg. Mech. Div.,

ASCE, Vol. 103, 229-241.

Dhanasekar, M., Page, A. and Kleeman, P. A (1984). Finite element model for the in-plane behaviour of brick

masonry, Proc. 9th Australasian Conf. on the Mech. of Struct. and Materials, Sydney, 262-267.

Hamid, A. and Drysdale, R. G. (1981). Proposed failure criteria for concrete block masonry under biaxial

stresses, J. Struct. Div., ASCE, Vol. 107, No. ST8, 1675-1687.


Klopp, G. M. and Griffith, M. C. (1994). The earthquake design of unreinforced masonry structures in areas of

low seismic risk, 3rd National Masonry Seminar, Brisbane, Australia, 19.1-19.9.

LaRovere, H. (1990). "Nonlinear analysis of reinforced concrete masonry walls under simulated seismic

loadings", PhD Thesis, University of California, San Diego.

Lotfi, H. and Shing, P. (1994). Interface model applied to fracture of masonry structures, J. Struct. Engrg,

ASCE, Vol. 120, 63-80.

Lourenco, P. B. and Rots, J. G. (1994). Understanding the behaviour of shear walls: a numerical review, Proc.

10th Int. Brick and Block Masonry Conf., Calgary, Canada, Vol. 1, 11-20.

Jankulovski, E. and Parsanejad, S. (1994). Earthquake resistance assessment of masonry buildings, 3rd

National Masonry Seminar, Brisbane, Australia, 16.1-16.12.

Naraine, K. and Sinha, S. (1991). Cyclic behaviour of brick masonry under biaxial compression, J. of Struct.

Engrg., ASCE, 117(5), 1336-1355.

Page, A. (1978). "Finite element model for masonry", J. Struct. Div., ASCE, 104(8), 1367-1285.

Page, A. (1982). An experimental investigation of the biaxial strength of brick masonry, Proc. of the 6th Int.

Brick/Blcok Masonry Conf., Rome, Italy, 3-15.

Pomonis, A., Spence, R. J., Coburn, A. W. and Taylor, C. (1992). Shaking table test on various types of

unreinforced masonry, 10th World Conf on Earthquake Engrg., Madrid, Spain, 3533 -3538.
Riddington, J. and Ghazali, M. (1990). "Hypothesis for shear failure in masonry joints", Proc. of Inst. of Civil

Engineers, U.K., Part 2, Vol. 89, 89-102.

Rots, J. G. (1988). Computational modelling of concrete fracture, PhD thesis, Delft University of Technology,

The Netherlands.

Samarasinghe, W., Page, A. W. and Hendry, A. W. (1981). Behaviour of brick masonry shear walls, The Struct.

Engineer, U.K., 59B(3), 42-48.

Tomazevic, M., Lutman, M. and Petkovic, L. (1996). Seismic behaviour of masonry walls: Experimental

Simulation, J. of Struct. Engrg., ASCE, 122(9), 1040-1047.

Tomazevic, M. and Lutman, M. (1996). Seismic behaviour of masonry walls: Modeling of hysteretic rules, J. of

Struct. Engrg., ASCE, 122(9), 1048-1054.

Vratsanou, V. (1991). Determination of the behaviour factors for brick masonry panels subjected to earthquake

actions, Proc. Int. Conf. on Soil Dyn. and Earthquake Engrg., Germany, 23-26.

Zhuge, Y. (1995). Nonlinear dynamic response of unreinforced masonry under in-plane lateral loads, PhD

Thesis, Queensland University of Technology, Brisbane, Australia..

Zhuge, Y., Thambiratnam, D. and Corderoy, J. (1994). Experimental testing of masonry walls under in-plane

cyclic loads, Proc. 10th Int. Brick/Block Masonry Conf., Calgary, Canada, 313-322.

Zienkiewic, O. C. and Pande, G. N (1977). Time-dependent multilaminate model of rocks - a numerical study of

deformation and failure of rock masses, Int. J. Numer. and Analytical Methods in Geomec., Vol. 1, 219-247.
APPENDIX II. NOTATION

The following symbols are used in this paper:

a = strain softening parameter

C = constant

[C] = damping matrix

[Df] = material constitutive matrix after failure


Em = Youngs modulus for masonry
E1 = Youngs modulus in the principal direction 1
Es = secant modulus of unloading/reloading branch
E0 = initial tangent modulus in compression
t
E = tangent modulus of elasticity
fcn,fcp = uniaxial compressive strength normal and parallel to the bed joint respectively
ft = tensile strength of masonry
fm = compressive strength of masonry
G = shear modulus
I 1, I 2, J 2 = principal stress invariant
[K] = stiffness matrix
[M] = mass matrix
{P} = external load matrix
{U} = vectors of nodal displacement
.
{U } = vectors of nodal velocity

..
{U } = vectors of nodal acceleration
..
u
g = ground acceleration

= unloading parameter
= shear aggregate interlock factor
X, Y = normal stress
1, 2 = principal stresses
ic = maximum stress associated with direction i, i=1,2.
m = vertical compressive stress
n = normal stress
iu = equivalent uniaxial strain in ith direction, i=1,2.
ic = corresponding strain to ic in i direction
1, 2 = principal strains
cu = uniaxial crushing strain
cr = cracking strain
= mass density
= shear stress
0 = shear bond strength at zero precompression
u = ultimate shear strength
= shear strain
= friction coefficient at the brick mortar interface
= Poissons ratio

TABLE 1. Material Properties of Example 1

fm cu ft E0

(MPa) (MPa) (MPa)

(1) (2) (3) (4) (5)

7.92 4.60 x 10-3 0.34 2991 0.2


TABLE 3. Material Properties of Wall 2 (Example 2)

Fm ft E0 0

(MPa) (MPa) (MPa) (MPa)

Masonry Mortar Masonry Masonry Masonry

(1) (2) (3) (4) (5)

9.63 5.5 0.3 3950 0.73


TABLE 2. Comparison of Analytical and Experimental Ultimate Strengths

Experimental Analytical

Wall Cracking Positive Peak Negative Peak Cracking Peak Load


No. Load Load Load Load

(kN) (kN) (kN) (kN) (kN)

(1) (2) (3) (4) (5) (6)

2 50 +98.5 -90 45 105

3 30 +40 -35 30 40

7 40 +80 -78 45 65

8 40 +110 -100 45 105

9 40 +70.8 -71.8 45 65

10 35 +100.9 -99.8 45 105


TABLE 4. Material Properties of Wall 5 (Example 3)

fm ft E0

(MPa) (MPa) (MPa) (kg/m3)

Masonry Mortar Masonry Mortar Masonry Mortar Masonry

(1) (2) (3) (4) (5) (6) (7)

8.8 2.9 0.0a 0.0a 1404b 1000 1937

a
The tensile strengths were not provided by Klopp and Griffith (1994) - they will be assessed in the present study.
b
A dynamic Youngs modulus was determined based on the dynamic shear stiffness measured in the laboratory

with a lower value equal to 1065MPa.


TABLE 5. Comparison of Results for Wall 5

Absolute Max. top displacement

Max. base (mm)

acceleration Experiment Analysis

(g) ft=0.3MPa ft=0.4MPa ft=0.5MPa

=2 =4 =2 =4 =2 =4

(1) (2) (3) (4) (5) (6) (7) (8)

0.343 0.024 0.020 0.020 0.020 0.020 0.020 0.020

0.616a 0.103 0.422 0.673 0.137 0.221 0.143 0.509

failed failed failed

a
The wall failed immediately after the acceleration increased beyond 0.616g.





10

8 =67.5
Biaxial failure

criterion =45

6 Coulomb criterion
2 / 1 = 15
xy / ft

4 =22.5
10

2
5
=90 =0
0
2 0 -2 -4 -6 -8 -10 -12 -14
n / ft (Compression)


300

250
Horizontal Load (KN)

200

150

100
Experiment
50
Analysis
0
0 2 4 6 8 10
Horizontal Displacem ent (m m )
120

80
Lateral Force (KN)

40

0
-10 -5 0 5 10
-40
Experiment
-80
Analysis
-120
Lateral Deflection (m m )
Wall 5
0.7

0.5

0.3
Acceleration (g)

0.1

-0.1

-0.3

-0.5

-0.7
0 0.25 0.5 0.75 1
Tim e (s)
Wall 5 0.2

0.15

0.1

0.05
Displacement (mm)

-0.05

-0.1
divergence
-0.15 linear
=2
-0.2 =4

-0.25
0 0.2 0.4 0.6 0.8 1
Tim e (s)
Wall 5 0.15

0.1

0.05
Displacement (mm)

-0.05
f't=0.3MPa
f't=0.4MPa
-0.1
f't=0.5MPa

-0.15
0 0.2 0.4 0.6 0.8 1

Tim e (s)
(a) t = 0.59 sec (b) t = 0.84 sec

(c) t = 1.07 sec (d) t = 1.08 sec


FIGURE CAPTIONS

FIG. 1. Equivalent Uniaxial Stress-Strain Curve for Biaxial Stress State

FIG. 2. A Weak Plane Oriented at an Angle to the Global (1-2) Coordinate

FIG. 3. Simplified Failure Envelope for Masonry

FIG. 4. Tensile Cyclic Softening Model

FIG. 5. Comparison Between Analysis and Experiments

FIG. 6. Failure Pattern of Wall 2

FIG. 7. Comparison of Load-Deflection Curve

FIG. 8. Failure Pattern of the Wall

FIG. 9. Input Acceleration Time History of Wall 5

FIG. 10. Comparison Between Linear and Nonlinear Analysis

FIG. 11. Effect of Tensile Strength on Time History Analysis

FIG. 12. Cracking Sequences of Wall 5 (ft = 0.4MPa, = 2)

View publication stats

You might also like