Sei sulla pagina 1di 14

Materials and Structures (2009) 42:213226

DOI 10.1617/s11527-008-9379-5

ORIGINAL ARTICLE

A comparison of different failure criteria in a numerical


seismic assessment of an industrial brickwork chimney
F. J. Pallares A. Aguero S. Ivorra

Received: 19 January 2007 / Accepted: 14 April 2008 / Published online: 24 April 2008
 RILEM 2008

Abstract A theoretical analysis using three wellknown masonry analysis constitutive models is
performed on a masonry structure to simulate the
response of the structure to specific seismic forces.
The results of the three numerical approaches are
compared and a discussion is presented, mainly
intended for professionals, concerning the suitability
of the three models and the limitations of each
numerical approach. The aim of the study is to
evaluate the relative accuracy of the three different
models and their suitability for determining the
failure mode of the masonry chimney. The models
studied are: a linear elastic constitutive model, an
elastic-plastic Drucker-Pragers type model and a
model including cracking and/or crushing in the
material using Willam-Warnkes criterion. A macromodelling approach is used because of the great
number of elements forming the structure and the
computational demand. Seismic actions are
F. J. Pallares (&)
Department of Applied Physics, Polytechnic University of
Valencia, c/Camino de Vera s/n, Valencia 46022, Spain
e-mail: frapalru@fis.upv.es
A. Aguero
Department of Continuous Medium and Theory of
Structures, Polytechnic University of Valencia, c/Camino
de Vera s/n, Valencia 46022, Spain
S. Ivorra
Department of Construction Engineering, University of
Alicante, Apartado 99, Alicante 03090, Spain

synthetically generated and scaled until chimney


failure, in accordance with the present regulations on
seismic-proof constructions in Europe and Spain.
Conclusions for researchers and professionals are
obtained to determine the suitability of each model
according to the results required and the available
calculation capacity.
Keywords Accelerogram  Cracking 
Failure criteria  Masonry chimney 
Seismic analysis

1 Introduction
Industrial brickwork chimneys built during the industrial revolution towards the end of the 19th and early
20th century are common in many European countries.
They were built to get rid of smoke and to create the
necessary draught for industrial processes as, e.g., in
textile or paper manufacturing. Very few of them
remain in use, since they became obsolete when new
energy generation systems made their appearance in
the 20th century. In many cities these chimneys form a
characteristic landscape and are often protected by law
as part of the cultural heritage. Figure 1 shows one of
the many industrial chimneys that can be observed in
the city of Valencia (Spain).
In the existing scientific literature, there are few
references to this type of construction. In [1] the

214

Materials and Structures (2009) 42:213226

similar masonry structures nowadays involves a


considerable number of professionals, many of whom
apply simplified analysis, equivalent static analyses
or linear elastic dynamic analyses, since the use of
more refined models is normally reserved for research
purposes due to their complexity. However, the
currently available commercial software with commonly used masonry failure criteria for both
professionals and researchers can lead to more
realistic approaches.
The results presented here are obtained within the
framework of a theoretical study and show good
agreement with real failure modes observed in actual
earthquakes. The numerical failure modes obtained in
[5], which use the same criteria presented here, are in
accordance with the real failure modes observed
in actual earthquakes in [6], collapsing the chimneys
at the base or in the top third of the structure, mainly
depending on the height of the chimney. The practical
modelling guidelines and conclusions outlined here
may be of use to both professionals and those
involved in research.
Fig. 1 Industrial masonry chimney

2 Description
authors analyse the typology and structure of industrial chimneys built between 1870 and the first
decades of the 20th century in the Italian regions of
Piedemont and Veneto, and the problems associated
with their restoration.
In [2] the authors study the behaviour of three
typical chimneys in these areas using the finite
element method with a linear analysis, taking into
account the chimney self-weight, wind, temperature
differences and earthquakes as acting loads. Pallares
et al. [3] considers the behaviour of a chimney when
a seismic action is introduced, while Aoki and Sabia
[4] studies the structural characterization of a brick
chimney using experimental tests and model
updating.
This paper continues the investigation of this type
of construction. The aim is to evaluate the relative
accuracy of the different models and their suitability
for determining failure modes and crack patterns in a
seismic analysis. The different models are based on
the use of three different masonry failure criteria
usually proposed in the literature, and the study
focuses on the comparison of the criteria. The study
of the seismic vulnerability of these chimneys and

In Europe and America many of these chimneys were


built in seismic areas, so it is important to know their
behaviour in case of a low to moderate intensity
earthquake and also what criteria can lead to accurate
results. This study focuses on the Spanish Mediterranean area using the European code [7] and the
Spanish standard [8].
To achieve the aforementioned goals, the finite
element method (e.g. [9]) is used as a numerical tool to
study the response of the structure when subjected to
seismic tremors. There are many examples in which
masonry has been studied through this numerical
method in recent years. Advanced models for small
structures have been used to micro-model both bricks
and mortar, involving considerable computational
cost, unaffordable for large structures. Examples of
this micromodelling approach, usually employed to
compare with experimental results obtained in the
laboratory, can be found in [10] or [11].
Large structures are analysed through macromodelling approaches in which average stress states are
considered in the material. Examples of this kind of
modelling technique can be found in [12, 13].

Materials and Structures (2009) 42:213226

However, few works have been published relating


to the specific dynamic behaviour of masonry in
earthquakes, although researchers have recently
started focusing on this field, [14].
Different models can provide a fair approximation
of the response of the structure to earth tremors,
permitting the investigation of various aspects,
according to the objective of the study, and this is
one of the points of this paper.
In the present study, numerical analyses were
carried out in order to: (a) compare the three different
criteria usually employed when performing dynamic
analyses on masonry structures, (b) determine the
important aspects to be taken into account in seismic
analysis, (c) show the suitability of each criterion
according to the results required and the calculation
capacity available.
The criteria studied are commonly used in calculations for geo-materials such as masonry. These
criteria, used to separate the linear from the nonlinear behaviour of the material, are the DruckerPrager criterion and the Willam-Warncke criterion.
The third analysis carried out is a linear analysis
without any plastic or failure criterion. The results of
this elastic analysis were compared with the results
from the non-linear analyses and to evaluate the
quality of the selected mesh and the results obtained.
Examples and justification for the use of these models
is provided below in each case by references.
None of the models provides the stiffness degradation of masonry caused by successive plastic
deformations resulting from cyclic behaviour [15,
16], apart from consideration of the cracking/plastic
process in masonry due to a seismic action and its
progression in time.
The study is set out as follows:
(1)

(2)

(3)

Description of the problem, establishing the


geometric definition of the structure, specifying
height, base section, thickness, etc.
Discretization of the continuum through the
finite element method in order to reduce the
degrees of freedom to a discrete number,
assigning macromodel properties. Boundary
conditions and initial values are then fixed to
set up the model for the seismic action.
Generation of the artificial accelerograms to be
used in the calculations from the response
spectrum proposed in the standards.

215

(4)

(5)
(6)

Setting up yield criteria: with the model and the


seismic action defined, a yield/crack criterion is
needed to register the inelastic behaviour of
masonry during the loading process. Three
different criteria are studied, resulting in three
analyses: linear elastic analysis, elastic-plastic
analysis (Drucker-Prager) and a linear elastic
analysis until cracking or crushing occurs in
masonry (Willam-Warncke).
Calculation and results of the analyses.
Conclusions derived from the analyses.

3 Properties of the chimney


The structure studied in the present work is an
industrial chimney made of masonry. One particular
chimney was chosen whose dimensions and circular
section are representative of the great number of these
chimneys in Mediterranean coastal areas and Europe.
It is 30 m high without reinforcement, as can be seen in
Fig. 2a. The dimensions shown were established by
rules of thumb taken from different handbooks such as
[17] or [18] and compared with the actual dimensions
measured from other 30-m chimneys. Variability in
dimensions is not a parameter studied here.
The structure can be clearly divided into three
parts, [19]:
(1)

(2)

(3)

Base: Its task is to distribute stresses on the


foundation. It is not always necessary and
sometimes does not exist. Shapes can vary
widely: square, hexahedron, octagonal, etc.
Shaft: This is the chimney strictly speaking. Its
tasks were to lead the smoke to a great height to
avoid environmental contamination and to create the necessary draught to facilitate
combustion. It is formed by a tube with varied
cross-sections: circular, square, hexagon, and
octagon. It is variable in height with prismatic
or helicoidal shapes.
Crown: The upper part of the chimney with an
aesthetic purpose.

As previously stated, the discretization of the


continuum is made through 3D solid elements with
plastic and cracking capabilities, depending on the
failure criteria used. Eight node bricks with three
degrees of freedom per node are used.

216
Fig. 2 (a) Longitudinal
section of the industrial
brickwork chimney studied
in the present work.
(b) Discretization using 3D
solid elements. Isometric
view. Tested meshes

Materials and Structures (2009) 42:213226

(a)

Figure 2b shows three tested meshes of the finite


element discretization used for the chimney. The
central mesh was chosen for the calculations, since
accurate enough results were obtained from it compared with the finer mesh, having a total of 20,320
nodes with 60,800 active d.o.f. and 15,280 elements.

4 Seismic action
Since time-dependent transient responses were
looked for, accelerograms were used as input for

(b)

the seismic action. In this way a comparison could be


made between the models of the time (instant of the
seismic motion) in which the initial crack appeared
and how the cracking pattern spread. Each accelerogram was synthetically generated [20] compatible
with the seismic spectrum proposed in the current
European code [7] and in accordance with the Spanish
norms for the Mediterranean coast (Valencia), [8].
A total of five accelerograms were used to compare the
different criteria.
The initial synthetic accelerograms generated from
the standards were scaled to produce cracks

Materials and Structures (2009) 42:213226

(or plastic deformation) and the collapse of the


structure in order to study the failure process. The
peak associated ground acceleration was 0.06 g.
Figure 3a shows the response spectrum for the
city of Valencia from the Spanish norms, while
Fig. 3b shows one of the artificial accelerograms
generated.

5 Calculation values
As stated above, a macro-structural approach was
adopted in order to model the whole structure, due
to the great number of elements forming the
chimney. More detailed models would become
extremely demanding from a computational point
of view, and the heterogeneities and uncertainties
usually present in masonry structures make it
advisable to use models with a small number of
mechanical parameters.
The industrial chimneys were built of masonry due
to its natural properties, which were suited to the use
for which the chimneys were conceived: it was easy
to handle and had good thermal and mechanical
properties.
The number of unavoidable uncertainties is sometimes high when professionals carry out structural
assessments, and discrepancies between calculated
and experimental values may occur. In the present
paper, rough but representative values for the
masonry mechanical parameters were used to perform the calculations.
For the masonry, the values used in the calculations were:

217

Uniaxial compressive strength: fc 6,500,000 N/m2


Uniaxial tensile strength: ft 200,000 N/m2
Elasticity modulus: E 6e8 N/m2

Poisson ratio: m 0:2


q 1,600 kg=m3
Due to the lack of laboratory data, these values
were chosen from references such as [17, 21] or [22].
These provide valuable information on the strengths
of masonry used for chimneys and are typical values
for masonry structures from the end of the 19th until
the middle of the 20th century. The low tensile
strength was chosen in order to consider the lime
mortar used in the first chimneys built in the 19th
century, being a typical mortar used prior to 1950, or
chimneys constructed with cement mortar whose
mechanical properties are deteriorated due to climatic
conditions. This is a key parameter in the seismic
behaviour of chimneys and must be estimated
carefully in each case.
6 Cracking and plastic criteria
The plastic and cracking criteria used in the calculations presented here are the well-established
Drucker-Prager and Willam-Warncke criteria applied
in geo-materials, adapted to the case under study
(masonry).
It has been said that none of the models used
provides or takes into account either the stiffness
degradation in masonry caused by successive plastic
deformations or cracks resulting from low cycle
fatigue, or material fatigue. However, the purpose of
this paper is to compare both frequently used criteria

Fig. 3 (a) Response spectrum. (b) Artificial accelerogram generated; time (s) versus acceleration (a/g)

218

Materials and Structures (2009) 42:213226

in masonry and to understand the cracking progression in the material, since cracking is the main
phenomenon that governs the response and loads
applied to the chimney do not lead to crushing or
plastic deformation due to excessive compression, as
will be shown in the results presented. Nevertheless, a
reduced elasticity modulus is considered to take into
account the degradation process [23].
Both the Drucker-Prager and Willam-Warncke
criteria are frequently used in masonry structures to
determine the end of the elastic behaviour and the
beginning of the non-linear range. The Drucker-Prager
model has been employed in a brittle material such as
masonry in different situations to simulate initial
cracks at the onset of plastic deformation [13]. Addessi
et al. [24] combines both damage and a DruckerPrager plastic criterion to reproduce the behaviour of
masonry to cyclic loading, while Cerioni et al. [25]
applies Drucker-Prager criterion to the case of Parma

qrm ; h

2qc q2c  q2t cos h qc 2qt  qc

Drucker-Prager criterion [29]


This plastic criterion, included in many nonlinear software packages, states the beginning of
plastic strains if:
p
b  I1 J2  ry 0
2

where b, parameter related to the internal friction


angle; I1, first stress invariant; ry, parameter
related to cohesion and J2, deviatoric stress
invariant.
Both parameters are fitted in order to obtain the ft
and fc strengths fixed in (1).
Willam-Warncke criterion [30]
This cracking criterion states initial crack if an
equation such as (3) is satisfied:
p
sm
f rm ; sm ; h 3
10
3
qrm ; sm
where

p
4q2c  q2t cos2 h 5q2t  4qt qc

4q2c  q2t cos2 h qc  2qt 2

Cathedral Bell-Tower, subjected to the El Centro


earthquake comparing different finite elements. The
Willam-Warncke criterion showed good results with
masonry in [26], obtaining a 2D failure surface quite
similar to that presented in [27]. Litewka and Szojda
[28] uses a Willam-Warnke limit surface at material
failure for brittle material, considering brick masonry
in the numerical model proposed, with good agreement
to experimental results. The WW criterion is considered in this paper as the reference criterion because its
use in seismic analysis of masonry chimneys [5] gives
results that agree with actually observed failure modes,
as presented in [6], and it makes suitable allowance for
the cracking phenomenon. This is why it will also be
considered as the reference criterion, theoretically
estimating the appearance of the first crack, although
no experimental results can be provided to support this
piece of information.
Their capacity for reproducing seismic behaviour
in masonry chimneys will be shown in the next
section. These criteria can be stated as:

p2 r
3 ;
h is the angle of similarity given by: cos h 2r21pr
3 J2
J2 second invariant of stress deviator tensor; r1 ; r2 ; r3
are the principal stresses; rm is the mean normal stress:
rm r1 r32 r3 ; qc is the deviatoric length for h = 60,
and qt is the deviatoric length
qfor h = 0; sm is the
mean shear strength: sm 25 J2 :
More information about these criteria can be found
in [31].
Parameters to define the failure surface are those
given in (1).

7 Results
The calculations were carried out in a Pentium-IV
computer, 2.8 GHz processor, 1 GB RAM and about
50 h were needed to complete one seismic motion.
Although several earthquakes were simulated, the
following figures present different instants for only
one of the earthquake motions tested, in order to
compare the criteria appropriately. In the figures,

Materials and Structures (2009) 42:213226

stresses, plastic deformations and cracks are plotted


according to each criterion. Results for the three
criteria are presented below:
7.1 Linear elastic analysis
This elastic model has many weaknesses, but it is
widely used in practical masonry analysis and
conclusions can be obtained. Results for this linear
elastic model with no cracking or plastic criteria are
in agreement with those obtained from the non-linear
analyses until the non-linearities are involved in the
chimney response, which occurs when the equation
for each non-linear criterion is satisfied.
The high level of tensile stresses in the bottom and
middle sections of the chimney suggests the need to
use non-linear models to detect possible cracks, as the
tensile strength is exceeded (see Fig. 4a). The
masonry compressive strength seems not to have a
significant effect on the response of the chimney due
to the low gravity stresses experienced. Furthermore,

219

vertical stresses under self-weight are used to check


that the weight of the model has been properly
introduced, comparing it with simplified calculations
(Fig. 4b). It is confirmed that the chimney remains in
the elastic range as stresses are below the elastic
limit, as expected.
In order to evaluate the quality of the mesh
adopted in the subsequent calculations, three meshes
were tested in this linear elastic model, as can be
observed in Fig. 2b. Results were compared for the
three meshes, to determine the mesh that provided the
best results with the lowest computational effort. No
significant differences could be obtained using more
refined meshes.
The model developed for this elastic study will be
used in further work, currently under development, to
obtain numerical modal frequencies and vibration
modes to compare with those obtained from experimental studies, validating or updating the parameters
of the numerical model.
7.2 Drucker-Prager criterion (D-P)

Fig. 4 (a) Stresses higher than tensile strength at the bottom.


(b) Vertical stresses under self-weight

Figure 5a, b show longitudinal stresses and plastic


strains given by the Drucker-Prager criterion at
t = 5.34 s from the beginning of the seismic
action, this being the time in which plastic
deformation is initiated, while Fig. 6a and b show,
in the same way, at instant t = 5.62 s, longitudinal
stresses and plastic strains when the plastic strains
have progressed and extended to both sides of the
base of the chimney.
Time t = 5.70 s is the instant when plastic strains
reach the shaft, seen in Fig. 7a, where plastifications
at the base and middle sections are shown, while in
Fig. 7b, at t = 5.72 s, plastic strains and stresses are
displayed, the latter remaining constant when the
plastic criterion is satisfied according to the perfect
plasticity criterion used.
Figures 8a, b; 9a, b; 10a, b show the growth of the
plastic strains in the shaft in the right border at
t = 6 s and the final state of the plastic areas when
the seismic action has finished.
Finally, Fig. 11 displays the evolution of plastic
strains with time belonging to the two edge nodes on
both sides of the chimney base as an example of the
different results that can be obtained from the
analysis.

220

Materials and Structures (2009) 42:213226

Fig. 5 t = 5.34 s.
(a) Longitudinal stresses
(vertical axis) throughout
the chimney. Isometric
view, (N/m2); (b) Plastic
strains at chimney base.
Isometric view (zoom).
Maxima at the base of the
chimney

Fig. 6 t = 5.62 s.
(a) Longitudinal stresses
(vertical axis) throughout
the chimney. Frontal view,
(N/m2); (b) Plastic strains
on both sides of chimney
base. Frontal view (zoom)

7.3 Willam-Warncke criterion (W-W)


In order to compare both criteria appropriately, results
are displayed showing the time of the appearance of the

first crack at different points of the chimney and the


final crack pattern for the same seismic action.
Figure 12 shows the instant of the appearance of
the first crack at the base of the chimney. This instant

Materials and Structures (2009) 42:213226

221

Fig. 7 (a) t = 5.70 s. Plastic strains along the right chimney border. Initial plastic strains at a height of 18 m. (b) Longitudinal
stresses (vertical axis) along the right chimney border, t = 5.72 s

sides of the base due to the seismic action. Here, as


before, the agreement with the previous criterion is
clear.
Finally, Fig. 14 presents the instant t = 5.72 s at
which cracks appear in the shaft of the chimney.
Once more, agreement can be found with regard to
the D-P criterion. However, an important item is
shown in Fig. 15, where displacements at the base of
the chimney (fitting with the seismic action imposed
at the base) and amplifications at the crown can be
observed until t = 5.94 s, when the chimney fails.
This failure is not recorded using D-P criterion.
The results displayed for both criteria have been
presented for one synthetically generated seismic
action, but the study was performed for different
seismic actions, for which the conclusions are
outlined in the following section.

8 Conclusions

Fig. 8 t = 6 s. (a) Longitudinal stresses (vertical axis)


throughout the chimney. Isometric view, (N/m2). Maximum
located in the shaft of the chimney. (b) Plastic strains in the
shaft. Isometric view

coincides with the moment at which the beginning of


plastic strains using the D-P criterion is achieved, as
can be checked.
Figure 13 shows stresses for the instant at which
cracks appear in the left edge of the chimney base and
how cracking progresses in depth and extends to both

Several seismic movements were used to study the


seismic behaviour of an industrial masonry chimney in
order to compare widely used masonry failure criteria.
The results of the three numerical approaches were
compared and a discussion presented, mainly intended
for use by practicing engineers, concerning the
suitability and limitations of each numerical approach,
evaluating the seismic response and failure mode in the
type of structure shown. The main characteristics
compared between the criteria, usually required in
seismic analysis, were: appearance of the first crack in
the different parts of the chimney and failure mode.

222

Materials and Structures (2009) 42:213226

Fig. 9 (a) Growth of the plastic strains throughout the chimney, t = 6 s (right border). (b) Enlargement of the plastic strains area at
the base, t = 12 s. Isometric view

Fig. 10 Final state of plastic strains on the left border (a) and the right border (b) of the chimney

As a result of the study, the following conclusions


are drawn and comments are made for each criterion,
with an outline of the general conclusions and a
summary of the whole work:

estimate the appearance of cracks and will indicate


the need for a more refined (non-linear) analysis.

8.1 Linear elastic model

This criterion is capable of representing the appearance of the first cracks, producing similar predictions
of initial cracks to the W-W criterion, as has been
shown. Following the first cracks, however, failure
mode and subsequent cracks can be wrongly predicted by the D-P criterion (as compared to W-W
criterion), since tensile strength still remains after
D-P surface is reached, leading to cases (crack

Elastic linear analysis does not capture the failure


mechanism but indicates qualitatively possible failure
modes and cracking areas, fully resisting the seismic
motion undergone by the chimney. It provides insight
into the response of the structure as a first approximation, and is the fastest and easiest analysis to

8.2 Drucker-Prager criterion

Materials and Structures (2009) 42:213226

patterns and failure modes) not in accordance with


W-W predictions. The residual tensile strength
remaining in the plastic areas can lead to a chimney

Fig. 11 Final state of plastic strains on the left edge (a) and
the right edge (b) of the base

Fig. 12 (a) Longitudinal


stresses (vertical axis)
throughout the chimney.
Isometric view, (N/m2);
(b, c, d) Crack pattern.
Frontal view (b-global,
c-base zoom) and oblique
view (d-base zoom). Cracks
at the base and the lower
part of the chimney

223

fully withstanding some earthquakes when W-W


predicts collapse.
Taking this into account, when as a consequence of a
seismic motion a part of the chimney plastifies (cracks),
e.g. the bottom left part of the base, if the next cracks (or
tensile demanding area) are on the opposite side or in a
different part of the chimney (e.g., shaft), so that tensile
strength is not being demanded from a previous plastic
area, these cracks will be in accordance with the W-W
criterion. However, if the next cracks involve previous
plastic areas through tensile stresses, the subsequent
cracks and failure mode could be erroneous.
In general, this criterion consumes less computer
time than the Willam-Warncke criterion and can give
a good understanding of the seismic behavior, taking
into account the limitations commented.
To sum up, the use of plastic criteria such as the
DP criterion employed here, and used in many
software packages, provides suitable approximations
of the stressstrain fields until the appearance of the

224

Materials and Structures (2009) 42:213226

Fig. 13 Instant of
appearance of cracks on the
left border. (a) Longitudinal
stresses (vertical axis) along
the chimney. Isometric
view, (N/m2); (b, c) Crack
pattern. Frontal view (bbase zoom) and oblique
view (c-base zoom)

first cracks, even in the case of brittle material such as


masonry, but can fail to predict the crack pattern at
failure.

8.4 General conclusions and comments

8.3 Willam-Warncke criterion


This criterion provides more realistic information
than the DP criterion on the progression of the crack
pattern during the earthquake if a time-dependent
analysis is desired. This is because it makes proper
allowance for the cracking phenomenon reducing the
stiffness matrix of the cracked elements and introducing crack planes.
One of its drawbacks is that it requires the largest
computational effort, but results are in accordance
with observations of the failure modes of real
chimneys in earthquakes, showing reasonable results
for both crack pattern and the first crack appearance.

After the study of the different accelerograms, the


chimney studied presents a failure mode by
toppling over the base due to the extent of the
cracks, based on the results from the W-W
criterion. This is not adequately recorded by the
D-P criterion in any of the cases studied or by the
elastic criterion.
Failure proceeds from cracking and not from
crushing of masonry in brickwork chimneys, as a
result of the low tensile strength exhibited by the
masonry and the type of action. The masonry
compressive strength does not have a significant
effect on the response of this type of structure,
due to the low gravity stresses experienced, the
low tensile strength being the key parameter.
The W-W criterion is considered as the reference
criterion in this study, since it makes proper

Materials and Structures (2009) 42:213226

225

Fig. 14 t = 5.72 s. Cracks


in the shaft of the chimney.
Frontal view

Fig. 15 Displacements of the base and crown of the chimney.


The displacement of the base node (2) corresponds to the
seismic motion imposed while the displacement of the crown
node (1) to the amplifications undergone

allowance for cracking phenomenon, as previously


explained, modifying the stiffness matrix when
cracks are produced, and the calculated failure mode
matches well with the failure modes observed in real
chimneys that have experienced earthquakes. Similar results have been reached in [13].
It is really important to take the masonry tensile
strength into account in order to adequately
estimate the failure mode. The results shown
here agree with the observations made in the cited
literature, even though they are based on numerical simulations.

In conclusion, the low tensile strength exhibited by


masonry in this type of construction results in cracking,
which is the main contributing factor to non-linear
behaviour. It is therefore important to include precise
information on this factor in the model. The stress
redistribution and increase of the plastic area influence
the crack pattern and, consequently, the failure mode
obtained, so it is therefore possible to conclude that the
results obtained, mainly with the W-W cracking
criterion, are useful for evaluating the maximum
seismic action that a chimney can withstand and the

failure mode in the case of excessive cracking, while


D-P can provide fair approximations to initial cracks
and crack patterns.
The numerical models shown are reasonably able
to predict the response to the seismic action, according to the degree of accuracy, the results looked for,
and the available computer capacity, giving valuable
information on possible crack pattern distributions in
earthquakes.
These conclusions and comments may provide
useful information on the repair and retrofit applications to this type of structure carried out by practising
engineers or researchers, and some guidelines to
assess the reliability of measures taken to preserve
these or similar constructions from earthquakes.
All the approaches have their limitations, and
uncertainties of masonry properties in this type of
construction should induce professionals to act with
caution.
The conclusions and comments stated could be
extrapolated to similar masonry structures such as
towers, bell towers, minarets or lighthouses.

References
1. Riva G, Zorgno AM (1995) Old brickwork chimneys:
structural features and restoration problems. In: 4th International Conference on Structural Studies, Repairs and
Maintenance of Historical Buildings, STREMAH 95,
Comp Mech Publications, Southampton, Boston, vol 2,
Dynamics, Repairs &Restoration, 1995, pp 317327

226
2. Pistone G, Riva G, Zorgno AM (1995) Structural behaviour of ancient chimneys. In: 5th Internal Conference on
Structural Studies, Repairs and Maintenance of Historical
Buildings, STREMAH 95, Comp Mech Publications,
Southampton, Boston, vol 3, Advances in Architecture
Series, 1995, pp 331341
3. Pallares FJ, Aguero A, Martn M (2006) Seismic behaviour
of industrial masonry chimneys. Int J Solids Struct 43(7
8):20762090
4. Aoki T, Sabia D (2006) Structural characterization of a
brick chimney by experimental tests and numerical model
updating. Masonry Int 19(2):4152
5. Pallares FJ (2005) Contribucion al analisis ssmico de
chimeneas industriales de obra de fabrica mediante el
metodo de los elementos finitos. PhD Thesis (in Spanish).
Universidad Politecnica de Valencia
6. Ghobarah A, Baumber T (1992) Seismic response and
retrofit of industrial brick masonry chimneys. Can J Civil
Eng 19:117128
7. Eurocode 8: Design provisions for earthquake resistance of
structures. Part 1-1: general rules. Seismic actions and
general requirements for structures
8. Norma de Construccion Sismorresistente: Parte General y
Edificacion (NCSE-02) (2002) Ministerio de Fomento,
(Spanish Standard)
9. Zienkiewicz OC, Taylor RL (1991) The finite element
method, 4th edn. McGraw-Hill, London
10. Lourenco PB, Rots JG (1997) Multisurface interface model
for analysis of masonry structures. J Eng Mech
123(7):660668
11. Lotfi HR, Shing PB (1991) An appraisal of smeared crack
models for masonry shear wall analysis. Comput Struct
41(3):413425
12. Middleton J, Pande GN, Liang JX, Kralj B (1991) Some
recent advances in computer methods in structural
masonry. Computer methods in structural masonry. In:
Middleton J, Pande GN (eds) Books and Journals International, Swansea, UK, pp 121
13. Genna F, Di Pasqua M, Veroli M, Ronca P (1998)
Numerical analysis of old masonry buildings: a comparison among constitutive models. Eng Struct 20(12):3753
14. Costa A, Arede A (2006) Strengthening of structures
damaged by the Azores earthquake of 1998. Construct
Build Mater 20:252268
15. Oliveira DV, Lourenco PB (2004) Implementation and
validation of a constitutive model for the cyclic behaviour
of interface elements. Comput Struct 82:14511461

Materials and Structures (2009) 42:213226


16. De Castro DV (2002) Experimental and numerical analysis
of Blocky masonry structures under cyclic loading. Thesis,
Escola de Engenharia, Universidade do Minho
17. Gouilly A (1876) Theorie sur la Stabilite des Hautes
Cheminees en Maconnerie. J Dejey & Cia Imprimeurs
18. Esselborn C (1952) Tratado General de Construccion:
Parte 1, Construccion de Edificios. Gustavo Gili (ed),
Buenos Aires, Argentina
19. Pallares FJ, Martn M (2001) Industrial chimneys in
Valencia City. Structural response model (in Spanish). VI
International Seminar Forum UNESCO, 2001
20. Gasparini DA, Vanmarcke EH (1976) Simulated earthquake motions compatible with prescribed response
spectra. Massachusetts Institute of Technology, Cambridge, Massachusetts
lvarez A (1904) Manual del Ingeniero. Adrian Romo
21. A
(ed), Madrid, Spain
22. Mazzocchi L (1965) Memorial Tecnico. Dossat (ed) (in
Spanish)
23. Naraine K, Shina S (1991) Model for cyclic compressive
behavior of brick masonry. ACI Struct J 88(5):603609
24. Addessi D, Marfia S, Sacco E (2002) A plastic local
nondamage model. Comput Methods Appl Mech Eng
191:1291310
25. Cerioni R, Brighenti R, Donida G (1995) Use of incompatible displacement modes in a finite element model to
analyze the dynamic behavior of unreinforced masonry
panels. Comput Struct 57(1):4757
26. Kappos AJ, Penelis GG, Drakopoulos CG (2002) Evaluation of simplified models for lateral load analysis of
unreinforced masonry buildings. J Struct Eng 128(7):
890897
27. Dhanasekar M, Page AW, Kleeman PW (1985) The failure
of brick masonry under biaxial stresses. Proc ICE (part 2)
79:295313
28. Litewka A, Szojda L (2006) Damage, plasticity and failure
of ceramics and cementitious composites subjected to
multi-axial state of stress. Int J Plast 22:20482065
29. Drucker DC, Prager W (1952) Soil mechanics and plastic
analysis or limit design. Q Appl Math 10:157165
30. Willam KJ, Warncke ED (1975) Constitutive model for the
triaxial behaviour of concrete. Proceedings of the International Association for Bridge and Structural
Engineering, 1975, 19. ISMES. Bergamo, Italy
31. Chen WF, Saleeb AF (1982) Constitutive equations for
engineering materials, vol 1. Elasticity and Modeling. John
Wiley and Sons

Potrebbero piacerti anche