A Simplifies Mechanics-Based Procedure For The Seismic Risk Assessment of Unreinforced Masonry Buildings

I.U.S.S.
Istituto Universitario
Universit degli Studi
di Studi Superiori di Pavia
di Pavia
EUROPEAN SCHOOL OF ADVANCED STUDIES IN

REDUCTION OF SEISMIC RISK
ROSE SCHOOL
A SIMPLIFIED MECHANICS BASED PROCEDURE

FOR THE SEISMIC RISK ASSESSMENT OF
UNREINFORCED MASONRY BUILDINGS
An Individual Study Submitted in Partial

Fulfilment of the Requirements for the PhD Degree in
EARTHQUAKE ENGINEERING
By
LUIS FERNANDO RESTREPO VELEZ
Supervisor: Prof. GUIDO MAGENES
December, 2003
The dissertation entitled A Simplified Mechanics-Based Procedure for the Seismic Risk Assessment
of Unreinforced Masonry Buildings, by Luis Fernando Restrepo Vlez, has been approved as second
dissertation, in partial fulfilment of the requirements for the PhD Degree in Earthquake Engineering.
Guido Magenes
Paolo Emilio Pinto

Abstract
ABSTRACT
A simplified mechanics-based procedure for the seismic risk assessment of unreinforced masonry
buildings is developed, based on some of the ideas presented in the original procedure proposed by
Calvi [1999]. This new procedure is based on the application of a displacement-based approach for
vulnerability evaluation of classes of buildings, but relevant changes have been carried out, in order to
correct the main drawbacks identified. One drawback is related to the out-of-plane behaviour that has
not been originally included. Another drawback is related to the definition of the joint probability
density function JPDF of displacement and period T; the real JPDF must consider the fact that
and T are not two independent variables.
For the in-plane mechanisms the demand is represented by the displacement response spectrum which
is obtained from the regional probabilistic seismic hazard study, and must be defined by the median or
mean value for each spectral period and also by the standard deviation. For the out-of-plane
mechanisms the demand is represented according with the formulation presented in Eurocode 8 to
compute the acceleration for non-structural components.
Appropriate definition of the JPDF would be necessary to be included in the methodology; however,
two main problems arise: the first is the definition of the JPDF, which should include the effect of all
the related sources of uncertainty and that cannot be estimated from the sample with enough accuracy,
but rather have to be possibly chosen by means of a heuristic approaching. The second problem is
related with the integration of the JPDF over the failure domain, in order to compute the probability of
failure. An alternative path has been chosen to overcome these problems, by estimating the CDF of
and T using first order reliability methods (FORM).
Considering all the sources of uncertainty, the strategy used in the procedure can treat the uncertainty
coming from the material properties and structural response, and the uncertainty coming from the
statistics of the population of buildings. The former is directly included in the computation of the
probability of failure for a single building or class of building, whilst the latter is considered in a
second stage.
As a complement of the proposed methodology a survey data form has been designed to gather the
data field required for the application of the methodology in real cases.
i
Acknowledgments
ACKNOWLEDGEMENTS
Being far is not an easy thing, neither for those who stay there, or for us living here. My deepest
gratitude to my family, for all the help and support during this period that I have spent far from them.
Thank you very much for your understanding and unconditional love.
A very special mention to the ROSE School for all the academic and financial opportunities that they
have given me so far, and specially to Professor Guido Magenes for his clear explanations on the
masonry issues, for the excellent guidance, for the opportunity of working as his research assistant and
for his generosity on providing financial support.
A true feeling of gratitude to Prof. Paolo E. Pinto and to Dr. Paolo Franchin for all the invaluable
explanations and help concerning the probability and reliability issues. Hopefully, now I have got the
idea.
The discussion sessions with my colleagues Ms. Helen Crowley and Ms. Rosamaria Iaccino were
always very fruitful; thanks to them that helped me to improve my work.
The most important place in this acknowledgements if for my wife Lina, for her endless patience and
for putting my priorities always first than her priorities. My infinite love is for you.
ii
Acknowledgments
AGRADECIMIENTOS
Estar lejos de la familia no es algo fcil, tanto para los que estn all como para los que estamos ac.
Mis ms profundos agradecimientos para mi familia, por toda la ayuda y el respaldo que me han
brindado durante todo este tiempo que he estado lejano. Muchas gracias por su comprensin y amor
incondicional.
Una mencin muy especial para la ROSE School por todas las oportunidades de carcter acadmico y
financiero que me han dado hasta ahora, y muy especialmente al Profesor Guido Magenes por sus
claras explicaciones relacionadas con los temas de mampostera, por ser un excelente gua, por la
oportunidad de poder trabajar como su asistente de investigacin y por su gran generosidad al
proporcionarme patrocinio.
Un verdadero sentimiento de gratitud para el Profesor Paolo E. Pinto y para el Dr. Paolo Franchin por
toda las invaluables explicaciones relacionadas con los temas de probabilidad y confiabilidad. Espero
haber finalmente captado la idea.
Las sesiones de discusin con mis colegas Helen Crowley y Rosamaria Iaccino fueron siempre muy
provechosas y contribuyeron a mejorar la calidad de mi trabajo.
El sitio ms importante en estos agradecimientos es para mi esposa Lina, por su paciencia sin lmites y
por poner mis prioridades siempre por delante de las suyas. Mi infinito amor es para ti.
iii
Index
A SIMPLIFIED MECHANICS BASED PROCEDURE

FOR THE SEISMIC RISK ASSESSMENT OF
UNREINFORCED MASONRY BUILDINGS
INDEX
page
ABSTRACT . i
ACKNOWLEDGEMENTS . ii
AGRADECIMIENTOS iii
INDEX .. iv
LIST OF TABLES vi
LIST OF FIGURES .. vii
1. INTRODUCTION .. 1
2. VULNERABILITY AND LOSS ESTIMATION PROCEDURES .. 3
2.1 DAMAGE PROBABILITY MATRIX ..... 4
2.2 GNDT I AND II LEVEL APPROACHES .... 6
2.3 VULNUS ... 8
2.4 HAZUS . 10
2.5 FaMIVE .... 14
iv
Index
page
2.6 SIMPLIFIED DISPLACEMENT BASED PROCEDURE FOR CLASESS OF

BUILDINGS . 15
2.7 GENERAL REMARKS .... 18
3. MeBaSe PROCEDURE .... 19
3.1 PROCEDURE FOR IN-PLANE FAILURE MECHANISM .................................... 19
3.1.1 Demand ................................................................. 19

3.1.2 Limit States ....................................................................... 20
3.1.3 Simplified Structural Model ............................................. 20
3.2 PROCEDURE FOR OUT-OF-PLANE FAILURE MECHANISM . 29
3.2.1 Demand ................................................................. 30

3.2.2 Limit States ....................................................................... 33
3.2.3 Simplified Structural Model ............................................. 34
3.3 PROBABILISTIC CONSIDERATIONS ...... 36
3.3.1 Uncertainties ......................................................... 37

3.3.2 Definition of the JPDF for structural capacity . 40
3.3.3 Definition of the Seismic Demand ... 41
3.3.4 FORM Procedure ...................................................... 42
3.3.5 Probability of failure for In-plane and Out-of-plane mechanisms ................ 48
3.4 DAMAGE PROBABILITY AT URBAN OR REGIONAL SCALE 50
4. APPLICATION EXAMPLES . 52
4.1 PROBABILITY OF FAILURE FOR IN-PLANE RESPONSE OF A 4-STORIES

UNREINFORCED MASONRY BUILDING ... 52
4.2 PROBABILITY OF FAILURE FOR OUT-OF-PLANE RESPONSE OF SINGLE

WALLS .. 59
5. SURVEY DATA FORM 63
6. CONCLUSIONS . 68
6.1 SUMMARY ....................................................... 68
6.2 FINAL REMARKS ........................................... 70
6.3 FUTURE RESEARCH .. 70
7. BIBLIOGRAPHY .. 72
ANNEX 1. FORTRAN Code for the MeBaSe Procedure
ANNEX 2. Survey data form
v
Index
LIST OF TABLES
2.1 Classification of vulnerability assessment methods, according to input, methodology and

output [Dolce et al., 1994].
2.2 European Macroseismic Scale EMS [Gruntal, 1998].
2.3 DPM Irpinia [i.e. Braga et al.; Giovinazzi and Lagomarsino, 2001].
2.4 Scores and relative weights to compute Iv.
2.5 Classification and corresponding values of the vulnerability factors
[Bernardini, Gori and Modena, 1990].
2.6 Vulnerability factors related to qualitative judgements and their
corresponding weights [Bernardini, Gori and Modena, 1990].
2.7 Linguistic relationship between a and I3 [Bernardini, Gori and Modena, 1990].
2.8 Limit states for masonry structures, Calvi [1999].
3.1 1 and 2 values, according to the number of stories.
3.2 Displacement ratios for the tri-linear model.
3.3 LS according to the boundary conditions.
4.1 Computation of the seismic coefficient C.
4.2 Computation of median values for LS and TLS..
4.3 Parameters for computing FTLS and F LS using FORM procedure.
4.4 Median and standard deviation values for vector X.
4.5 Probability of failure for each given Limit State.
4.6 Probability of failure for each given Limit State for the same seismic demand.
4.7 Median and standard deviation values for vector X.
4.8 Probability of failure for different demand levels.
vi
Index
LIST OF FIGURES
2.1 Differentiation of structures (buildings) into vulnerability classes [Gruntal, 1998].

2.2 Acceleration versus damage ratio tri-linear curves for masonry buildings proposed in the
GNDT II level approach [after GNDT, 1993].
2.3 Definition of the safety criterion [Bernardini, 2000].
2.4 Example of building capacity curve and demand spectrum [FEMA, 1999].
2.5 Example of fragility curves for slight, moderate, extensive and complete damage states
[FEMA, 1999], for a specific class of buildings.
2.6 Example of building damage estimation process [FEMA, 1999].
2.7 Out-of-plane mechanisms taken into account in the FaMIVE procedure [DAyala and
Speranza, 2002].
2.8 Displacement demand and capacity for LS4 and masonry building class of building (ground
acceleration 0.3g), Calvi, [1999].
3.1 Acceleration and Demand Response Spectra according to Eurocode 8, for ag = 0.3g, S = 1.0
and k = 1.0.
3.2 Simplified equivalent SDOF model.
3.3 Deformed shapes for different limit states and in-plane failure modes.
3.4 Simplified model for the definition of 2.
3.5 Definition of 2 according to the storey at failure.
3.6 Elastic-Perfectly plastic response for two different structures.
3.7 Force distribution, storey shears, storey strengths and seismic coefficients.
3.8 Boundary conditions considered.
3.9 Seismic response of an unreinforced masonry building [Paulay and Priestley, 1992].
3.10 Flexible floor response to ground excitation [Paulay and Priestley, 1992].
3.11 Example of acceleration and displacement response spectra computed with Eurocode 8.
3.12 Tri-linear simplified model.
3.13 Comparison of the limiting displacement 2 and the displacement response spectrum DRS.
3.14 Sources of uncertainty.
3.15 Components for different uncertainty models.
3.16 Sources of uncertainty on capacity response for in-plane walls.
3.17 Sources of uncertainty on capacity response for out-of-plane walls.
3.18 Seismic hazard curve of Peak Ground Acceleration for Gaziantep, Turkey, adapted from
Restrepo-Vlez and Bommer [2003].
3.19 Approximation of the limit state function g in the normalized standard space.
3.20 Subdivision of the hypothetic region in zones.
4.1 Plan view and faade for the in-plane assessment example.
4.2 Comparison of median values of capacity and demand for each given limit state.
vii
Index
4.3 Cumulative density functions CDF for demand FD.

4.4 Cumulative density functions CDF for TLS, and LS conditional to TLS = 1.0 sec.
4.5 Comparison of median values of capacity and demand for each given limit state, for
performance level 2 PGA 0.15g.
4.6 Variation of the probability of failure Pf with the coefficient of correlation ij, for LS4.
4.7 Variation of the sensitivity factors sen with the coefficient of correlation ij, for LS4.
4.8 Comparison of median values of capacity and demand.
4.9 Cumulative density functions CDF for demand FD.
4.10 Cumulative density functions CDF for TLS, and LS conditional to TLS = 1.5 sec.
4.11 Variation of the sensitivity factors sen with the coefficient of correlation ij, for LS4.
5.1 Main screen for the website used to manipulate the database.
5.2 Partial view of the screen used to enter the write the information to the database.
5.3 Partial view of the screen used to control and to modify the information written in the
database.
5.4 Output view of the screen used to write the input file required for the MeBaSe procedure.
viii
Chapter 1. Introduction
1. INTRODUCTION
A big research effort has been made on the prediction of earthquakes in the last decades, and in fact
the exploration of new techniques aiming to foresee the occurrence of seismic events is in a
continuous progress. However, the possibility of knowing in advance the occurrence of a major
earthquake is still far from being a reality, and the preparedness to face an eventual emergency has to
be made from the point of view of the prevention rather than the prediction.
Several analytical tools have been developed around the world in order to estimate, with different
degrees of accuracy, the vulnerability of buildings and the probable loss of lives and economic
resources, due to the occurrence of an earthquake. Those tools are intended to be used by government
agencies, and even by insurance companies, as a mean for planning of emergency preparedness
procedures and response strategies, and also for the reconstruction phases. Nonetheless, most of the
current available tools require a large amount of resources, in terms of money, time and computational
effort, in order to be properly implemented and effectively used.
Additionally, large portions of buildings in non-developed countries, and an important portion of

buildings in developed countries, especially old constructions, are built with unreinforced masonry,
under several forms. In most modern design codes the use of unreinforced masonry has been banned
for the construction of new buildings in moderate to high seismic regions. However, in low seismic
regions or in those countries were the use of an official building code has not been yet implemented,
the use of unreinforced masonry is still one of the most preferable choices for new low-rise buildings,
considering its relatively low cost and easy construction procedures.
From the arguments exposed in the previous paragraphs, it is possible to say that the development of a
simplified methodology for the seismic vulnerability, or better seismic risk assessment of unreinforced
masonry buildings, would have a considerable relevance, especially taking into account the social
1
Chapter 1. Introduction
benefits that some new developments in this field will bring to poor communities in developing
countries, providing useful tools for governments and other agencies.
The simplified methodology presented in this dissertation includes, apart from the considerations for
the in-plane failure mechanism, a formulation for the out-of-plane behaviour. Out-of-plane collapse
mechanism has received less research compared to the in-plane behaviour, and the aspects related to
the dynamic behaviour and vulnerability assessment still offer a lot of space for new developments.
This dissertation is organized into four chapters, covering from the theoretical development of the
methodology to the discussion of some numerical examples.
Chapter 2 is a general review of some of the methodologies most extensively used in Europe and
America, for the assessment of the vulnerability and loss estimation. This review includes the
Damage Probability Matrix DPM, the GNDT first and second level approaches, the VULNUS , the
FaMIVE and the HAZUS procedures.
Chapter 3 is dedicated to the development of a new simplified methodology for the seismic risk
assessment of unreinforced masonry buildings. The methodology includes the procedures for the
evaluation of in-plane and out-of-plane failure mechanisms, and also the complete analysis and
development of the probabilistic framework, necessary in order to considerer the main sources of
uncertainties existing in both the methodology and the physical phenomena. The procedure included
in this dissertation is presented as a first step of a major development that is being carried out as a part
of a PhD research project.
In Chapter 4 some numerical examples are presented with the purpose of illustrating in a detailed way
the steps of the procedure, including the probabilistic considerations.
Finally, the survey form used to gather the required information for the proposed procedure is
presented in Chapter 5.
2
Chapter 2. Vulnerability and Loss Estimation Procedures
2. VULNERABILITY AND LOSS ESTIMATION

PROCEDURES
The purpose of this chapter, rather than doing a full description of some of the most used vulnerability
and loss estimation procedures, is to explain the basic ideas of those procedures, showing some of
their advantages and limitations, aiming to highlight the interest and potential of developing rational
mechanics based methodologies.
Several methods for vulnerability assessment have been developed in recent years, considering
different approaches for in the input data and for the output. Corsanego and Petrini [1990] proposed
the following classification for the methods, according to the type of building:
Direct: these methods are subdivided in two groups. The typological group is based on data
gathered after the occurrence of real earthquakes that are statistically manipulated to obtain
damage probability matrices for a limited number of classes of buildings. The mechanical group
is based on numerical models and the output is usually the level of damage for a given building of
a given class.
Indirect: these techniques use both the data obtained after the occurrence of real earthquakes and
the data obtained through numerical models.
Conventional: these correspond to methodologies based on judgements given by experts. The

output is a vulnerability index for a given class of building but this vulnerability is not directly
correlated with any specific level of damage.
A different classification was proposed by Dolce et al. [1994], based on the input, the methodology
and the output, considering the options shown in Table 2.1.
3
Table 2.1. Classification of vulnerability assessment methods, according to input, methodology and
output [Dolce et al., 1994]
Input Method Output
1. Damage data 1. Statistical methods 1. Absolute vulnerability
2. Geometric and qualitative 2. Mechanical methods 2. Relative vulnerability
features 3. Methods based on expert
3. Mechanic features judgements
4. Seismic demand features
5. Geological and geotechnical data
Different input-method-output combinations are possible, resulting in methods with different levels of
applicability and accuracy. Nonetheless, Dolce et al. [1994] recognized that for masonry buildings
just hybrid expert-statistical methods are available, and that no method based on mechanical features
exists. Absolute quantification of the vulnerability is not available neither.
The selection of the adequate methodology must be based on the following criteria [Dolce, 1997]:
Purpose and regional scale of the damage scenario.
Accuracy of the earthquake scenario used for the damage scenario.
Available resources in terms of time and funds to prepare the building catalogue.
Quantity and accuracy required in the information according to the different types of feasible
methodologies.
Existing information, either from government bodies, databases, libraries, etc.
From the available literature and considering the vast collection of methodologies, the following six
procedures are explained in more detail, taking into account that they have been used or proposed to
be used for the assessment of unreinforced masonry buildings.
2.1 DAMAGE PROBABILITY MATRIX
The damage probability matrix method or DPM is based on the idea that a set of buildings having a
common structural typology would have the same behaviour under the action of an earthquake, and as
a consequence, the level of damage would be the same for the set of buildings. The damage is
characterized by some level of uncertainty described by a damage probability matrix [i.e. Whitman,
1973; Di Pasquale et al., 2001]. Each element of the matrix is expressed according to Eq.(2.1):
4
DPM (DV , I , T ) = P(DV | I , T ) (2.1)
where DV corresponds to a given level of damage, T is an specific structural typology and I is the
earthquake intensity, normally described by some macroseismic scale, for example the European
Macroseismic Scale EMS , given in Table 2.2 [Gruntal, 1998].
Table 2.2. European Macroseismic Scale EMS [Gruntal, 1998]

Scale Description
I Not felt
II Scarcely felt
III Weak
IV Largely observed
V Strong
VI Slightly damaging
VII Damaging
VIII Heavily damaging
IX Destructive
X Very destructive
XI Devastating
XII Completely devastating
This methodology is considered as a direct method because it allows estimating the vulnerability in
one single step, considering the building as a member within a specific class. The level or class of
vulnerability according to the structural typology of masonry buildings is shown in Figure 2.1,
according to the proposal given in the EMS-98, being A the highest vulnerability and F the lowest.
Figure 2.1. Differentiation of structures (buildings) into vulnerability classes [Gruntal, 1998]
According to Eq.(2.1), the DPM would be a matrix having the probability of reaching some level of
damage for a given earthquake intensity. An example of a DPM is shown in Table 2.3 prepared after
the Irpinia earthquake in Italy [i.e. Braga et al., 1982; Giovinazzi and Lagomarsino, 2001].
5
Table 2.3. DPM Irpinia [i.e. Braga et al, 1982; Giovinazzi and Lagomarsino, 2001]
EMS Intensity V VI VII VIII IX X
Class A 0.020 0.284 0.423 0.726 0.860 0.923
Class B 0.010 0.185 0.284 0.501 0.700 0.850
Class C 0.005 0.065 0.167 0.334 0.500 0.700
One of the main disadvantages of the DPM method is the use of a discrete measure of the earthquake
intensity through the use of a macroseismic intensity scale, rather than using a different definition of
intensity, for example a continuous parameter like acceleration or displacement, with a better
correlation with the level of damage. A key issue here is that the input is described by the effects of
the ground motion on structures, rather than other parameters measuring directly the input ground
motion, which is somehow like using the responses to compute the response. Another disadvantage is
that the DPM does not consider the uncertainty on the demand, which one of the components of
uncertainty that are required to be included in a complete vulnerability assessment, as discussed in
Section 3.3.1. Finally, the DPM method does not allow the estimation of the vulnerability for a single
building, but just the evaluation as a part of a class of buildings; thus, it not possible to individuate all
the features of each specific building.
2.2 GNDT I AND II LEVEL APPROACHES
The Gruppo Nazionale per la Difesa dai Terremoti GNDT is the Italian government research body in
charge of the seismic risk evaluation and definition of the required measures to reduce it. Two,
somehow complementary, approaches have been published by this group, with the aim of being
applied in the assessment of the seismic risk in the Italian territory.
The GNDT I level approach is nothing more than a DPM method, having three classes of vulnerability,
from A to C, each of these having a DPM. For this approach the demand is considered through the
use of the EMS-98 intensity scale and the damage is described by means of a qualitative description,
according to the level of damaged reached by the building. Further description of this methodology
can be found elsewhere [e.g. GNDT, 1993].
The GNDT II level approach is based on a survey form designed to gather information regarding the
typology and constructive features for each single building, that are combined afterwards to get a
vulnerability index IV. Eleven parameters are combined with different scores and relative weights,
depending on four classes of vulnerability, as shown in Table 2.4. IV is an absolute measure from 0 to
382.5 but eventually can be normalized from 0 to 100, being 0 the best vulnerability condition and 100
the worst.
6
Table 2.4. Scores and relative weights to compute Iv

Parameter Class Weight
A B C D
Type and layout of resistant system 0 5 20 45 1.00
Quality of resistant system 0 5 25 45 0.25
Conventional resistant 0 5 25 45 1.50
Location of building and foundations 0 5 25 45 0.75
Horizontal elements (floor, diaphragm) 0 5 15 45 variable
Configuration in plan 0 5 25 45 0.50
Configuration in height 0 5 25 45 variable
Max. distance between walls 0 5 25 45 0.25
Roof 0 15 25 45 variable
Non-structural elements 0 0 25 45 0.25
State of conservation 0 5 25 45 1.00
For each vulnerability index there is a corresponding curve correlating the damage ratio and the
demand represented by the PGA, by means of a tri-linear curve resembling somehow the so called
fragility curves, which are better explained in Section 2.4. The damage is computed in terms of
economical loss, correlated as a function of the peak ground acceleration PGA [Giovinazzi and
Lagomarsino, 2001]. Figure 2.2 shows the acceleration versus damage ratio tri-linear curves for
masonry buildings proposed in the GNDT II level approach.
1.0
Iv = 100
0.8
Damage ratio
0.6
0.4
Iv = 0
0.2
0.0
0.0 0.1 0.2 0.3 0.4 0.5
PGA [g]
Figure 2.2. Acceleration versus damage ratio tri-linear curves for masonry buildings proposed in the
GNDT II level approach [after GNDT, 1993]
Some important drawbacks of the I and II level approaches is that their intervals of confidence are
very large, both for the arbitrary way used to defined the points and weight for the vulnerability index,
and for lack of correlation between PGA and the level of damage of a given building [Giovinazzi and
Lagomarsino, 2001].
7
2.3 VULNUS
The VULNUS procedure was developed in the second half of the 80s at the University of Padova,
with the purpose of evaluating the seismic vulnerability of a single building or group of buildings
[Bernardini, Gori and Modena, 1990]. The methodology is based on the evaluation of the geometrical
and mechanical characteristics of each building, which is combined with the evaluation of some other
important factors controlling the response of the structure, that are handled through qualitative
judgments. The whole procedure is developed under the fuzzy set theory that is used for the definition
of the safety criterion. This method could be considered inside the mechanical group because it makes
use of the so called collapse multipliers.
The geometrical and mechanical characteristics are described with two indices or collapse multipliers,
according to Eqs (2.2) and (2.3):
min (V x , V y )
I1 = (2.2)
W
i
(
I 2 = min I 2' + I 2'' )
i
(2.3)
I1 is the collapse multiplier for in-plane behaviour considering shear failure at ground floor, being W
the total weight of the building and Vx and Vy the strength at mid-storey height of the ground floor,
according to Eq.(2.4):
1
2
{Vx ,V y } = {Fx , Fy } f t 1 + ( W ) (2.4)

1.5 f t Fx + Fy
{ }
where Fx , Fy are the total areas of the walls in the x and y direction respectively, ft is the tensile
strength of masonry [Ministero dei Lavori Pubblici, 1981] and is a factor to include the effects of
plan regularity. In this expression it has been assumed that the walls are rigidly jointed to the slabs
and are subjected to uniform vertical compression.
I2 is the collapse multiplier for the out-of-plane behaviour, considering each single wall i and several
failure modes, namely: overturning, flexural tension, arch crushing, shoulders overturning and tension.
A detailed description of the computation procedure for this collapse multiplier can be found in
Bernardini, Gori and Modena [1990].
8
Once the in-plane I1 and out-of-plane I2 indices have been computed, a safety criterion is chosen to
estimate the vulnerability of a given building. The definition of the safety criterion is shown in
Figure 2.3, where c1 = 0.5, c2 = 1.0, c3 = 0.1 and c4 = 1 are the values typically assumed for buildings
in Italy. The parameter u is defined with Eq.(2.5), being A the maximum base shear divided by the
total weight of the building W.
c + c c + [(I 1 / A c1 )(I 2 / A c1 )]
1
2
u= 3 1 2 (2.5)
2c3 + ac 4
The parameter a depends on the qualitative judgments that are performed taken into account large
databases gathered in Italy by means of several surveys performed in the past. An important feature of
this parameter is that it can be updated after the continuous improving of the database [Bernardini,
2000]. The qualitative judgments are expressed as a combination of seven vulnerability factors Si and
their corresponding weights Wi, and is evaluated with Eq.(2.6).
Figure 2.3. Definition of the safety criterion [Bernardini, 2000]
Wi S i
I3 = (2.6)
i 45 3.15
Tables 2.5 and 2.6 shows the proposed values for the size and the weight of each vulnerability factor,
respectively.
9
Table 2.5. Classification and corresponding values of the vulnerability factors

[Bernardini, Gori and Modena, 1990]
Class Size S
1 Good or corresponding to code 0
2 Almost good 15
3 Almost poor 30
4 Poor or unsafe 45
Table 2.6. Vulnerability factors related to qualitative judgements and their

corresponding weights [Bernardini, Gori and Modena, 1990]
Vulnerability factors Weight W
1 Walls system quality 0.15
2 Soil and foundations interaction 0.75
3 Floors interaction 0.50
4 Elevation regularity 0.50
5 Roof interaction 0.50
6 Interaction of non-structural elements 0.25
7 General maintenance conditions 0.50
Total 3.15
Once I3 has been obtained, a linguistic relationship between a and I3 is established, according to
Table 2.7.
Table 2.7. Linguistic relationship between a and I3 [Bernardini, Gori and Modena, 1990]
j=1 If I3 is very large then a is very large
j=2 If I3 is large then a is large
j=3 If I3 is medium then a is medium
j=4 If I3 is small then a is small
j=5 If I3 is very small then a is very small
After the three parameters I1, I2 and I3 have been computed and the linguistic relationship has been
established, fuzzy set theory is applied in order to compute the vulnerability value. Details of the
corresponding fuzzy set theory are not reported here but can be found in Bernardini, Gori and Modena
[1990].
The main advantages of this method are that it allows to compute an absolute value for the
vulnerability with respect to the intensity of the given ground motion, and also allows to classify the
surveyed buildings in an orderly way with respect to the vulnerability measure.
2.4 HAZUS
HAZUS is an acronym for HAZard in the U.S. and corresponds to a methodology developed by the
Federal Emergency Management Agency [Whitman et al., 1997; FEMA, 1999]. The methodology is
based on three fundamental concepts: capacity curve, design point and fragility curve.
10
Capacity curve is the relationship between the lateral load resistance of a given structure and its
characteristic lateral displacement [Kircher et al., 1997], and is typically obtained by means of a static
pushover analysis. The capacity curve is then converted to spectral acceleration and roof displacement
in order to be compared with the demand spectrum. Figure 2.4 shows an example of the capacity
curve and the demand spectrum for a given building. It can be seen that the capacity curve is
controlled by the yield capacity, or restoring force, and the ultimate capacity, being possible to
represent with this single curve the corresponding strength at a certain displacement limit state, which
is turns can be directly correlated with some damage limit state. The damage limit states considered
by HAZUS are: slight, moderate, extensive and complete; their description can be found in Kircher et
al . [1997].
For a given building class, defined by the structural system, the building type and the occupancy class,
a specific capacity curve is defined, for which the design point (Sa, Sd) is obtained. The design point
corresponds to the intersection of capacity and demand curves.
Figure 2.4. Example of building capacity curve and demand spectrum [FEMA, 1999]
The fragility curve concept represents the cumulative density function CDF for the probability of
reaching or exceeding a specific damage limit state for a given peak response to a given ground
motion demand. In HAZUS the fragility curves are assumed to be represented by lognormal functions,
hence they can be fully described by the median and the standard deviation, according to Eq.(2.7):

1 Sd
P[ds S d ] = ln _ (2.7)
ds
S d ,ds
11
_
where: S d ,ds is the median spectral displacement for which the building reaches the damage limit state
ds, ds is the standard deviation of the natural logarithm of spectral displacement for the damage limit
state and is the standard normal cumulative density function.
The variability of a given damage state ds is obtained with Eq.(2.8), where D is the variability of the
ground-motion demand, C is the variability of the capacity response and T,ds is the variability of the
damage state threshold.
ds = (CONV [ C , D ])2 + ( T ,ds )2 (2.8)
In Eq.(2.8) CONV represents the convolution process applied to the variability of the ground-motion
demand and the capacity response that is necessary to carry out, taken into account that demand
spectrum is dependent on building capacity. The convolution process is explained in Kircher et al .
[1997]. Nonetheless, when the variability of the ground-motion demand has already been
incorporated into the seismic hazard assessment, which is the normal practice, the variability of a
given damage state must be computed with Eq.(2.9) in order to avoid a double counting of D.
ds = ( C )2 + ( T ,ds )2 (2.9)
_
In HAZUS, typical median values S d , ds of spectral displacement and standard deviation values ds of
the natural logarithm of spectral displacement are given for 36 different classes of buildings, based on
experimental tests, experience and judgment. Those parameters are strictly valid for buildings within
the United States. Figure 2.5 shows an example of fragility curves for slight, moderate, extensive and
complete damage states obtained as explained above.
Once the design point and the fragility curve for a specific class of building have been computed, the
probability of reaching or exceeding a specific damage limit state is obtained. The process of building
damage estimation is schematically shown in Figure 2.6. It is worth noting that in HAZUS fragility
curves for non-structural drift sensitive components and non-structural acceleration sensitive
components are also provided.
12
Figure 2.5. Example of fragility curves for slight, moderate, extensive

and complete damage states [FEMA, 1999], for a specific class of buildings
Figure 2.6. Example of building damage estimation process [FEMA, 1999]
An important advantage of HAZUS is its ability to estimate the damage under a given earthquake
scenario, considering damages not just in buildings but also in lifelines, transportation systems, utility
systems, essential and high potential loss facilities. In fact, with HAZUS is possible to compute the
damage due to earthquake hazard, inundation, fire and hazardous materials.
13
The main disadvantage stems from its complexity. HAZUS is so complete and powerful that is takes a
lot of resources to be implemented in a real application for a small to medium community. In
HAZUSs users manual [FEMA, 1999] it is stated that the methodology allows for different levels of
funding which means different levels of inventory collection. However, when low levels of funding
are available, the information that is missing has to be assumed as similar to the one provided in the
HAZUS database, which has been validated to be used in the United States.
2.5 FaMIVE
The Failure Mechanism Identification and Vulnerability Evaluation method, FaMIVE [DAyala and
Speranza, 2002], has been developed in recent years aiming to the assessment of the seismic
vulnerability of historic buildings in town centres.
The procedure makes use of an electronic survey form that helps to gather the information on the field
and allows also to store it in a database.
Limit state and collapse mechanism analyses are in the core of the procedure, taking into account in
the process the friction coefficient of masonry, a detailed modeling of the restriction provided by the
lateral supporting walls and also the structural retrofit carried out in the buildings.
FaMIVE uses a predefined set of rules in the definition of the most feasible collapse mechanism, for
which a load factor or collapse multiplier is associated by means of a static equivalent procedure. This
load factor is further manipulated to lead to the definition of some vulnerability value.
DAyala and Speranza claim that this procedure can be applied to medium size samples of buildings
without sacrificing the benefit of a detailed knowledge of the mechanical and geometrical features of
each individual building.
The application of the procedure starts with the identification of all the structural typologies existing in
the zone being studied. Afterwards, a detailed survey of the relevant geometrical parameters (height,
length, thickness, number of stories, strengthening devices) for every faade wall on each building is
carried out. The surveyor must associate each building with one of the structural typologies
previously defined for the zone; then, according with existing information and the type of building, the
equivalent shear capacity ESC is computed for each faade wall, varying the number of stories
participating in the mechanism. The eight collapse mechanisms considered in the procedure are
presented in Figure 2.7. The selected collapse mechanisms are then ordered according to their ESC
14
and finally a judgment is made, considering the danger posed by each mechanism. The final result is
the vulnerability of the specific building.
The advantages of this procedure are the use of mechanical criteria for the definition of the
vulnerability and the inclusion of the most common out-of-plane failure mechanisms. Other
advantages are the use of the friction coefficient in the definition of the failure mechanisms and also
the possibility of applying the procedure to specific buildings.
The main drawbacks are not to consider the displacement rather than the acceleration demand and also
to ignore the inherent uncertainty existing in the seismic demand; being these two aspects quite
relevant, as will be discussed in Chapter 3.
Figure 2.7. Out-of-plane mechanisms taken into account in the FaMIVE procedure [DAyala and
Speranza, 2002]
2.6 SIMPLIFIED DISPLACEMENT BASED PROCEDURE FOR CLASSES OF

BUILDINGS
A simplified procedure for the evaluation of the vulnerability of masonry buildings, subject to in-plane
failure mechanisms, was developed by Calvi [1999], based on the application of the displacement-
based approach for vulnerability evaluation of classes of buildings. That procedure considers the
energy dissipation and displacement capacity of the existing buildings, and through a very simplified
probabilistic approach, computes the probability of occurrence of a specific damage state limit for a
given earthquake motion.
15
In general, the steps suggested in the procedure are:
Definition of the demand. For a given site or area being studied, the demand is defined in terms of
an elastic displacement response spectra DRS, based on the selection of an specific return period,
or better probability of exceedance. The selected response spectra must include the effects of soil
condition and geographical location. Response spectra taken from seismic design codes could be
a suitable choice to define the demand.
Definition of performance levels or limit states. The limit states are defined in terms of inter-
storey drift. Table 2.8 shows the limit states for masonry structures used by Calvi [1999] and
based on a work by Magenes and Calvi [1997]. The first two levels of limit state corresponding to
Undamaged Buildings and Slightly Damaged-Usable Buildings have been mixed into one single
limit state, taking into account that masonry buildings do not exhibit a clear definition of the
elastic limit, because cracking develops at very early loading stages. No coefficient of variation
for drift is shown for LS1-LS2 and LS3; however, from experimental results it can be stated that
the scatter for ultimate limit state is much more reduced than the scatter for initial and
intermediate limit states, [Magenes and Calvi, 1997].
Table 2.8. Limit states for masonry structures, Calvi [1999]

Limit State Median Drift LS (%) Coefficient of (%) LS
Variation for Drift
(%)
LS1 LS2 0.1 --- 2 1
LS3 0.3 --- 10 1 + 3/n
LS4 0.5 10.9 15 1 + 6/n
Definition of simplified structural models. The simplified models are defined considering a linear
displacement profile for LS1-LS2 and a soft-storey displacement profile with failure at the ground
floor for the LS3 and LS4. The corresponding displacement for any given limit state is computed
with Eq. (2.10), for which hs is the typical storey height and n is the number of stories.
LS = 0.67hs n y + ( LS y )hs (2.10)
The period of vibration is computed with Eq. (2.11), using the same ductility factors LS shown in
Table 2.8.
TLS = 0.04(nhs )
3
4 LS (2.11)
16
Definition of a joint probability density function JPDF in the displacement period T space, for
which a uniform distribution is assumed, limited by the computed minimum and maximum
expected displacement capacities, and by the minimum and maximum expected period of
vibration.
Computation of the probability that the demand exceeds the capacity. The probability is computed
by integrating the JPDF of -T in the areas above the DRS.
Figure 2.8 shows the displacement demand and the capacity for masonry building class and a
ground acceleration of 0.3g.
Figure 2.8. Displacement demand and capacity for LS4 and masonry building class
of building (ground acceleration 0.3g), Calvi, [1999]
The main advantage of this methodology is the possibility to evaluate the seismic vulnerability of a
given class of buildings, considering different levels of quality in the data and in the refinement in
modelling and analysis [Calvi, 1999]. Here it is worth stressing the importance of this methodology in
proposing a mechanical formulation for the solution of the problem.
Two main drawbacks can be identified in this methodology: First, the out-of-plane behaviour has not
been included arguing that out-of-plane collapse mechanism is only feasible when ties or transversal
walls are absent. Out-of-plane behaviour has received little attention in the literature, and has been
considered as a marginal, or at least, a local structural problem. However, the experience in recent
earthquakes and numerous works available in the literature [e.g. Giuffr, 1999; DAyala et al., 1997]
have shown that the outward out-of-plane collapse of faade walls is the most probable failure mode.
They also have shown that the out-of-plane collapse occurs before the activation of some in-plane
17
mechanism. Besides, it would be possible that some out-of-plane collapse occurs, even in buildings
with ties or good connection of the out-of-plane walls, especially considering that these walls are
sensitive to displacement rather than to acceleration, according to what has been shown by Doherty et
al [2002].
The second drawback is related to the definition of the JPDF of and T. To treat the JPDF
f ,T (, T ) as a uniform distribution for the T space, considering both variables as independent,
was useful for the description of the original methodology, but the real JPDF must take into account
the fact that and T are not two independent variables.
2.7 GENERAL REMARKS
The purpose of the revision reported in this chapter was to identify possible features that ideally would
be considered also in the new procedure, and problems that would be either solved or avoided in order
to have a sounder methodology.
The features that have been identified to be ideally included in the new procedure are:
The use of displacement or drift as an indicator of the demand level, which has a better
correlation with the level of damage.
To consider the uncertainty on the demand, recognized to be one of the components of

uncertainty that are required to be included in a complete seismic risk assessment.
The use of a mechanics based approach for the definition of the structural response.
Conversely, the problems identified to be overcome are:
The use of a discrete measure of the earthquake intensity through the use of a macroseismic
intensity scale.
The uncertainties in the methodology originated in the arbitrary definition of factors and
weights.
Complexity in the application of the procedure in real cases.
18
Chapter 3. MeBaSe Procedure
3. MeBaSe PROCEDURE
In this Chapter a new proposal for the seismic risk assessment of unreinforced masonry buildings is
presented. The methodology is based on the general ideas proposed by Calvi [1999], but relevant
changes have been carried out, in order to correct the main drawbacks identified in Section 2.6.
3.1 PROCEDURE FOR IN-PLANE FAILURE MECHANISM
3.1.1 Demand
The definition of the demand representing the ground motion was originally proposed to be
represented by the elastic displacement response spectrum DRS ( = 5%) for the area of interest, given
Tr, source, path and site-effects. Being the demand represented by a displacement response spectrum,
then displacement demand is a function of period, as shown in Eq.(3.1). The importance of
underlining such an obvious concept will be clear in the following paragraphs.
demand = f (T ) (3.1)
The demand can be obtained from regional probabilistic seismic hazard studies or from available
seismic codes. In order to apply the probabilistic procedure presented in Section 3.3, the demand must
be defined by the median or mean value for each spectral period and also by the standard deviation of
the corresponding demand parameter. Figure 3.1 shows an example of the acceleration response
spectra ARS and the displacement response spectra DRS for ag = 0.3g, S = 1.0 and k = 1.0, defined
according to Eurocode 8 [CEN, 2001] with 5% of critical damping ratio, and illustrating the random
nature of the demand.
19
0.8 0.16
0.7 0.14
0.6 ARS 0.12

DRS
0.5 0.10
Sa ('g)
0.4 0.08
0.3 0.06
0.2 0.04
0.1 0.02
0.0 0.00
0.0 1.0 2.0 3.0 4.0
Figure 3.1. Acceleration and Demand Response Spectra according to Eurocode 8,

for ag = 0.3g, S = 1.0 and k = 1.0
Given the fact that additionally to acceleration or displacement, some other ground-motion parameters
could influence the response of masonry structures [Penazzo, 2001], the demand could also include
duration/energy-related parameters. Nonetheless this issue is not explored in the current development.
3.1.2 Limit States
Appropriate median values for the drift limit, corresponding to each given limit state, must be used.
Brick masonry, stone masonry, tuff masonry and other types of masonry have different drift limits that
have to be obtained from testing or from the available literature. In case they need to be assumed, a
higher level of uncertainty must be set accordingly. Alternatively, the values presented in Table 2.8,
proposed by Calvi [1999], could be used.
3.1.3 Simplified Structural Model
3.1.3.1. Definition of the Displacement Limit State LS. A simplified equivalent SDOF system,
based on the well known substitute structure method [Shibata and Sozen, 1976], suitable to model
the in-plane behaviour of masonry buildings must be used here. Figure 3.2 shows the general features
of the SDOF, in which a multi degree of freedom system MDOF with masses mfi located at each storey
height hi and subjected to a system of lateral forces Fi is represented by a SDOF with effective mass
meff and effective stiffness Keff, given by the ratio between the yielding force of the system Fy and the
displacement at the effective height heff for a given displacement demand or limit state LS.
20
i
mfi
Fi F
meff
Fy
j eff
hT
hi
Keff
heff
y LS
Figure 3.2. Simplified equivalent SDOF model
Being presented the simplified SDOF model now in Figure 3.3 four possible displacement profiles for
different limit states and in-plane failure modes are shown. The profile (a) corresponds to the limit
states LS1 and LS2 for which none of the members of the structure has reached the yield displacement
y (i.e. an appreciable extension of cracking). The yield displacement at the effective height he,
assuming a triangular displacement shape, can be computed with Eq.(3.2). The triangular shape is
justified for low-rise buildings dominated by a shear deformation failure mode, as is the case for most
old-unreinforced masonry buildings.
y = 1 hT y (3.2)
The profiles (b), (c) and (d) correspond to the limit states LS3 and LS4, in which either a soft storey
mechanism or a weak beam mechanism has been developed as the most probable failure mode.
Profile (b) is the most typical situation for low rise unreinforced masonry buildings, but profile (c)
could be also possible, either at top or intermediate stories, depending on the relative strengths of the
stories; or also profile (d), depending on the relative strengths of the spandrel beams and the piers.
(a) (b) (c) (d)

Figure 3.3. Deformed shapes for different limit states and in-plane failure modes
21
The maximum displacement for a given limit state LS, can be computed as the summation of the yield
displacement y Eq.(3.2), and the plastic displacement p Eq.(3.3), being hsp the effective height of the
piers going into the inelastic range, when openings are present, or the storey height when there are no
openings.
p = 2 ( LS y )hsp (3.3)
LS = y + p (3.4)
LS = 1 hT y + 2 ( LS y )hsp (3.5)
When the building has a regular distribution of masses and a uniform storey height 1 is equal to
0.667. However, when the buildings is better represented by a distributed mass corresponding to the
mass of the masonry walls plus lumped masses corresponding to the mass at each floor, the 1 has to
be computed in the way described below.
In Eqs. (3.3) and (3.5) a key issue is the definition of 1 and 2 that have been defined by carrying out
the following analysis:
A simplified model like the one shown in Figure 3.4 was used, for which lumped masses mf were
considered at each floor, and the masonry is assumed to have a distributed mass mm per unit
length. The definition of some other relevant parameter is shown in the figure.
x
mfi yi
pi
mmdx hT -h1
dx j
hsp
hT
hi
x
h1
Figure 3.4. Simplified model for the definition of 2
22
The effective displacement was computed according to Eq.(3.6), assuming that for any given
storey i, the yield displacement y i is obtained with Eq.(3.7) and the plastic displacement p i
with Eq.(3.8).
( )
n
mi yi + p i
e = i =1
(3.6)
m ( )
n
i yi + pi
i =1
y i = hi y (3.7)
p i = ( LS y )hsp (3.8)
Eq.(3.6) can be written in an explicit way, including the terms corresponding to the distributed and
the lumped masses, as appears in Eq.(3.9), being Mm computed with Eq.(3.10) and Nm computed
with Eq.(3.11).
n n n
y2 hi2 m f + 2 y p hsp hi m f + p2 hsp2 m f + M m
i i i
i =1 i= j i= j
e = n n
(3.9)
y hi m f + p hsp m f + N m
i i
i =1 i= j
h1 hT
M m = mm (x y ) dx + mm (x y + p hsp ) dx
2 2
(3.10)
0 h1
h1 hT
N m = mm x y dx + mm (x y + p hsp )dx (3.11)

0 h1
From Eq.(3.9) and assuming a ductility factor = 1, which means p = 0, it is possible to compute
e for the yielding state. Then equalling e to LS in Eq.(3.5) it can be computed 1.
Now, with Eq.(3.9) and for different ductility factors, an effective displacement e is computed,
which is then equated to LS in Eq.(3.5), with the 1 previously defined. By rearranging Eq.(3.5),
2 is obtained with Eq.(3.12):
23
2 = ( LS 1 hT y ) / ( LS y )hsp (3.12)
As an example, Figure 3.5 shows the variation of 2 for different ductility factors , and considering
failure at different heights, for a five storey building with a mass ratio mf / mm equal to 0.3. It can be
noticed that 2 is almost insensitive to the variation of the ductility level.
From the analysis of a set of buildings, ranging from 1 to 6 stories and all the buildings with the same
mass ratio mf / mm equal to 0.3, it has been computed the following values of 2 according to the
number of stories n, which are presented in Table 3.1. These values are considered valid when the
weak storey is formed in the two lower thirds of the building.
Finally, with 1 and 2 from Table 3.1, the displacement for any given limit state LS can be computed
with Eq.(3.5).
1.4
1.2
1.0
0.8
=1
2
0.6 =2
=3
0.4
=4
0.2 =5
0.0
1 2 3 4 5 6
Story at failure
Figure 3.5. Definition of 2 according to the storey at failure
Table 3.1. 1 and 2 values, according to the number of stories

Number of Stories 1 2
n
1 0.790 0.967
2 0.718 0.950
3 0.698 0.918
4 0.689 0.916
5 0.684 0.900
6 0.681 0.881
24
3.1.3.2. Definition of the Ductility for a Given Limit State LS. From Eqs.(3.2) and (3.5), the
ductility for a given limit state LS, is obtained with Eqs.(3.13) and (3.14) for the case when there are
openings, or with Eq.(3.15) for the case when there are not openings and the height hsp is the same for
all the stories.
LS y + p p
LS = = = 1+ (3.13)
y y y
( LS y )hsp
LS = 1 + (3.14)
hT y
( LS y )
LS = 1 + (3.15)
n y
3.1.3.3. Definition of the Capacity for Given Limit State LS. Unlike Eq.(3.1), in this case the
displacement capacity for a given limit state computed with Eq.(3.5) does not depend on the period of
the structure, but rather on its material and geometrical properties. For two buildings with similar
materials and geometrical properties, this relationship will depend just on the total height hT as shown
in Eq.(3.16). Thus, it is necessary to find a relationship between structural period and total
height hT = g (T ) , in order to be able to compare demand and capacity.
capacity = f (hT ) (3.16)
capacity = f (g (T )) (3.17)
Let us consider for example two structures with the same given height and with one and two walls,
corresponding to curves 1 and 2 in Figure 3.6, respectively. These structures will reach the yielding
displacement y when the applied lateral force reaches the strength of each wall Fy, assuming a simple
bilinear elasto plastic idealisation. Beyond this point, no additional force is necessary to take the
structures to some displacement limit state LS. The effective period of these structures will be a
function of the secant stiffness to the displacement capacity for the given limit state.
25
Fy,2
KLS,2
Fy,1
KLS,1
y LS
Figure 3.6. Elastic-Perfectly plastic response for two different structures
From Figure 3.6, Eqs.(3.18) and (3.19) can be stated, relating the stiffness and displacement at yield,
with the equivalent stiffness and displacement for a given limit state.
Fy = K y y = K LS LS (3.18)
K y y Ky
K LS = = (3.19)
LS LS
Two different approaches can be followed beyond this stage. The first one would be to consider the
elastic period of the structure Ty as a function of total height hT, according to Eq.(3.20), which is
equivalent to those type of formulations used in seismic codes. In this case a fundamental assumption
is made: two buildings with the same height have the same or approximately the same elastic period,
regardless their yielding strength.
T y = hT (3.20)
From Eq.(3.19), the relationship shown in Eq.(3.21) can be established between effective period for a
given limit state and elastic period, including the effect of ductility, and combining the standard-code
form shown in Eq.(3.20), to relate elastic period with total height of the structure, with Eq.(3.21), a
general expression for the effective period in terms of total height and ductility is obtained, according
to Eq.(3.22).
TLS = T y LS (3.21)
26
TLS = hT LS (3.22)
Now, by using Eqs.(3.2), (3.13) and (3.22), a simple expression relating displacement and effective
period for a given limit state is found, recalling the assumption already explained:
y
LS = 2 1
TLS2 (3.23)
h
2
T
Eq.(3.23) must be carefully understood, because some misleading interpretation could be drawn at first
sight. Through the entire derivation of this equation it was already shown that the dependency of LS
on TLS is just a mathematical manipulation made in order to be able to compare the capacity given by
this equation, with the demand given by Eq.(3.2). In fact, this relationship must not be read as if the
effective period of a structure is changed, its capacity will increase; the true meaning of this equation
is that for a given displacement capacity an effective period is obtained.
In the first approach described above, just the height and some ductility measure of the building was
considered, and no reference to the strength was made, which leads to inconsistencies in the
procedure, as shown in Figure 3.6.
For the second approach let us consider two structures, the first with one single wall and strength Fy,1
and the second with two identical walls and strength Fy,2 that will be the double than the first one, but
the yielding and limit state displacement capacities will be about the same. For the two structures the
secant stiffness and the effective period will not be the same. Thus, it is necessary to find an
expression to relate the effective period to both ductility and strength.
A second and more rigorous alternative, which is the one followed in this dissertation, considers that
two buildings of the same height can have different strength and hence different effective stiffness.
Thus, it is required that the maximum displacement at given limit state LS, to be expressed as a
function of the strength Fy and the effective period TLS, through Eqs.(3.24) and (3.25).
M LS
TLS = 2 (3.24)
Fy
Fy
LS = TLS2 (3.25)
4 M
2
27
Having found an expression for the relationship defined in Eq. (3.17), an additional step is yet
necessary to define the strength of the structure Fy, in order to do a proper use of the modified
methodology. The strength Fy can be computed in terms of the seismic coefficient C with Eq.(3.26).
Fy = CW = CMg (3.26)
Combining Eqs.(3.25) and (3.26), a relationship between displacement LS and effective period TLS for
a given limit state, Eq.(3.27) can be defined, being C computed by Eq. (3.28). Figure 3.7 shows the
meaning of C for every storey m. For a given force distribution (a) the storey shears can be computed
(left b), and then considering the strength distribution (right b), the ratio between the strength and the
shear is obtained. This ratio is assumed to be the implicit seismic coefficient for every storey, and the
minimum Cm value will be the one controlling the overall response of the structure.
Cg 2
LS = TLS (3.27)
4 2
C = min{C m } (3.28)
Cm can be computed with Eq. (3.29) [Benedetti & Petrini, 1984], where: Am is the normal area of all
the resistant elements in direction of minimal strength, km the reference shear strength for masonry at
the level m, n the number of floors, and m the ratio between Am and the maximum area Bm of the
longitudinal or transversal walls. This expression is similar to Eq.(2.3) used in the VULNUS
methodology.
1
n
2
1 Wi
Cm = A 1 + i=m
(3.29)
1.5 km Am (1 + m )
m km
n
hiWi
Wt n
i=m
h jW j
j =1
28
Fi Vm Fym Cm
(a) (b) (c)
Figure 3.7. Force distribution, storey shears, storey strengths and seismic coefficients
It has to be recognized that Eq.(3.29) is rather simplistic because it considers the in-plane shear
strength as the only mechanism to withstand the seismic action. Besides, in this equation it has not
been included any influence of three-dimensional effects, like torsion and distribution of strength in
plan, for the computation of the seismic coefficient.
A slight modification of Eq.(3.28) is proposed here under the form of Eq.(3.30), in which c is a
correction coefficient included to cope with the possible three-dimensional effects and other sources of
strength not included in the original formulation. Research is currently being carried out on the
development of some analytical expression to allow for the computation of c based on some basic
information.
1
C= min{C m } (3.30)
c
In order to apply all the steps proposed in the methodology by Calvi [1997], two additional steps are
still missing: the definition of a joint probability density function JPDF in the -T space, and finally,
the computation of the probability that the demand exceeds the capacity. Those two steps are treated
in detail in Section 3.3.
3.2 PROCEDURE FOR OUT-OF-PLANE FAILURE MECHANISM
A similar procedure to that proposed for the in-plane failure mechanism for unreinforced masonry
walls has to be developed for the out-of-plane failure mechanism, in order to have a complete model to
29
be used on the seismic assessment of unreinforced masonry buildings, considering both modes of
failure.
The general procedure is intended to follow the same steps outlined in the original procedure by Calvi
[1999], but new approaches are necessary to define both the demand and the response of the structural
model to considerer the features of the out-of-plane behaviour.
The procedure developed so far and explained below is restricted to parapets and simple supported
walls, either with loadbearing or non-loadbearing walls, according to the boundary conditions shown
in Figure 3.8.
Timber bearer boundary condition

W W
Slab boundary condition
t Slip joint boundary condition
(a) (b) (c) (d)

Parapet wall Simply-supported Simply-supported
Non-loadbearing Loadbearing wall
wall parapet wall
Figure 3.8. Boundary conditions considered
3.2.1 Demand
In the same way as the demand defined for the in-plane failure mechanism, the objective for the out-
of-plane failure mechanism is also to define the demand as a function of period, as shown in Eq.(3.1).
Figure 3.9 shows the path of seismic energy through an unreinforced masonry building [Paulay and
Priestley, 1992]. The input acceleration and displacement at each level for the out-of-plane walls is
due to the combined effect of the end in-plane-walls response and the floor diaphragm response.
Paulay and Priestley proposed a procedure to estimate the input acceleration for the out-of-plane walls,
considering the effects already described, as shown in Figure 3.10.
30
Figure 3.9. Seismic response of an unreinforced masonry building [Paulay and Priestley, 1992]
Figure 3.10. Flexible floor response to ground excitation [Paulay and Priestley, 1992]
A similar procedure is adopted in Eurocode 8 [CEN, 2001] to compute the acceleration for non-
structural elements, according to Eq.(3.31):
h
3 1 + ne
hT
S ane = 2
(3.31)
Tne
1 + 1
T1
where: is equal to the peak ground acceleration of the given ground motion, hne is the height at
which the non structural element is placed, hT is the total height of the building, Tne is the period of
vibration of the non-structural element and T1 is the period of the first mode of vibration of the
building.
31
The numerator in Eq.(3.31) considers the amplification of the acceleration due to the level of the non-
structural element with respect to the ground level; the higher the element is placed, the larger the
amplification is, being maximum for an element placed at the roof level. The denominator in the same
equation represents the dynamic interaction described by the ratio between the period of vibration of
the non-structural element and the period of the first mode of vibration of the building. When both
periods of vibration are the same there is no effect on the acceleration of the non-structural element,
when the period of the non-structural element goes to zero the de amplification effect of the period
ratio is null, and the period tends to infinite the denominator tends to infinite and the acceleration goes
to zero. Also a de amplification takes place when Tne is larger than T1.
Little experimental evidence is available to compute the effect of location in height and period ratio on
the displacement for the non-structural element, and apparently no definitive conclusions have been
drawn so far. In the New Zealand code for assessment [NZSEE, 2002] it has been suggested that the
displacement for non-structural elements can be computed with Eq.(3.32), but being this a cautious
assumption.
Tne2
S d ne = S ane (3.32)
4 2
As an example Figure 3.11 shows the acceleration and displacement response spectra for a non-
structural element placed at 12 meters from the ground level, in a 15 m high building with T1 =
0.40 sec, for PGA = 0.30g.
Demand Response Spectra

1.2 0.10
1.0 Sa
0.08
Sd
0.8
0.06
Sa ('g)
Sd (m)
0.6
0.04
0.4
0.02
0.2
0.0 0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
T ne (sec)
Figure 3.11. Example of acceleration and displacement response spectra computed with Eurocode 8
32
It can be assumed that the previous equations developed for non-structural elements can also be used
for out-of-plane walls.
3.2.2 Limit States
To define proper limit states for the assessment of out-of-plane unreinforced masonry walls it is useful
to refer to the work done by Doherty et al. [2002] and Griffith et al. [2003], in which the semi-rigid
nonlinear behaviour of a bending wall is described by means of a tri-linear simplified model, as shown
in Figure 3.12.
Fo corresponds to the force at incipient rocking and is also called the Rigid threshold resistant, u is
the ultimate displacement or the displacement at the point of static instability. The straight line
defined by these two parameters means that the wall is cracked before the motion starts, and that the
wall is behaving as a set of rigid bodies.
The real behaviour of the semi-rigid wall, which is a consequence of the stiffness of the wall before
cracking and also of the finite rather than infinitesimal dimensions of the pivot points, is represented
by the continuous curve. Doherty et al [2002] and Griffith et al. [2003] have shown that the nonlinear
nature of the masonry walls, subjected to out-of-plane motion, can be described by the tri-linear
model, defined by the parameters 1 and 2, being these parameters estimated from the material
properties and the state of degradation of the mortar at the pivot points, as a proportion of the ultimate
displacement. Table 3.2 shows the ratios proposed by Doherty et al [2002] for three different
conditions of the cracked joint.
F
Fo
K2
1 2 u
Figure 3.12. Tri-linear simplified model
33
Table 3.2. Displacement ratios for the tri-linear model

State of Degradation at Cracked Joint 1/u 2/u
(%) (%)
New 6 28
Moderate 13 40
Severe 20 50
Griffith et al. [2003] have shown that the displacement 2 can be used as a valid parameter, when the
main objective is to determine whether a wall will collapse or not, and in fact they stated:
The inaccuracy in the evaluation of 2 has no relevant effect in the application of the simplified
procedure.
The maximum strength defined by the force plateau limited by 2 is a rough estimate of the real
maximum strength.
The maximum wall strength estimated is relatively insensitive to the mechanical material
properties like the elastic modulus of elasticity and the compressive strength.
It is important to underline that for the current definitions of limit state it is possible just to identify
whether a wall would collapse or not, thus stable or collapsed and the only possible limit state at
this stage.
3.2.3 Simplified structural models
The ultimate displacement or the displacement at the point of instability is a function of the thickness
of the wall, the boundary conditions and , which is the ratio between the overburden weight (axial
force applied at the top the wall) and the self-weight of the upper-half of the wall above mid-height.
u = f (t , B.C., ) (3.33)
Fo is the force required to initiate the rocking motion, and is a function of the overburden ratio , the
effective mass and the geometry of the wall, and can be expressed according to Eq.(3.34):
4(1 + )M e gt
Fo = (3.34)
h
34
From Figure 3.12 and once the parameters Fo, 2 and u have been computed or estimated, it is
possible to compute the force corresponding to the plateau, by using equation (3.35), and the stiffness
to the point of limiting displacement is determined with equations (3.36) and (3.37).

F = Fo 1 2 (3.35)
u
F
K2 = (3.36)
2
Fo 2
K2 = 1 (3.37)
2 u
Now, considering the relationship between stiffness and period, Eq.(3.38), and Eqs. (3.34) and (3.37),
it is possible to get an expression to link the period T2 and the limiting displacement 2, as is presented
in Eq.(3.39).
4 2 M e
K2 = (3.38)
T22
1
2

22h
T2 = (3.39)
1 2 (1 + )gt

u
For a given limiting displacement than can be conservatively assumed as LS = u, being a factor
between 0.8 and 1.0, and knowing the ratio 2 = 2/u, it is now possible to compute the effective
period, with the use of Eq. (3.40), and this value has to be compared with the demand, defined in
Section 3.2.1, to evaluate whether for the computed period TLS, the limiting displacement LS would be
greater or lower than the demand defined by the displacement response spectrum DRS, as shown in
Figure 3.13.
1
2 2 u h 2
TLS = (3.40)
(1 2 )(1 + )gt
35
Now, the two additional steps: definition of a joint probability density function JPDF in the -T space,
and the computation of the probability that the demand exceeds the capacity, can be applied, according
to the details shown in the next section.
Sd
Safe
Fails
LS
T2 TLS T
Figure 3.13. Comparison of the limiting displacement 2 and the displacement response spectrum DRS
It was already said that the current development for out-of-plane mechanism includes just the
configurations shown in Figure 3.8. Extension of this ideas to more complex failure mechanisms is
currently under way, assuming that all the mechanism being studied can be represented by a tri
linear model similar to the one shown in Figure 3.12.
3.3 PROBABILISTIC CONSIDERATIONS
In Sections 3.1 and 3.2 the simplified methodologies for the vulnerability assessment of in-plane and
out-of-plane unreinforced masonry walls were presented. These new methodologies are based on two
basic concepts: a mechanical-based approach for the formulation of the response of the structure and
the use of a joint probability density function JPDF for the computation of the probabilities of
exceedance of a given limit state.
In this Section all the steps required for the probabilistic considerations of the methodologies are
developed, from the identification of the sources of uncertainty, to the definition of the JPDF and the
integration procedures to compute the probability of exceedance.
36
3.3.1 Uncertainties
The problem of estimation of the seismic risk of a structure is a very cumbersome matter, especially
when a full description of the uncertainty of all the parameters involved is required. In any seismic
risk assessment procedure the uncertainties come from both the demand and the response sides.
The uncertainty is recognized as originated in three different sources, namely: ground-motion demand,
damage state threshold and capacity response, according to the definition used in the HAZUS
procedure [Whitman et al., 1997; FEMA, 1999] as shown in Figure 3.14.
Ground motion demand
Sources of uncertainty Damage state threshold
Capacity response
Figure 3.14. Sources of uncertainty
3.3.1.1. Uncertainties on ground-motion demand. Despite the fact that there is a generalized
consensus about the sources of the uncertainty, there are different approaches and methodologies used
to consider the uncertainty in a seismic risk and loss estimation analysis, as is explained below.
In a general sense, the uncertainties in ground-motion demand are classified in two different types
[Kennedy et al., 1980]: randomness R and uncertainty U components. Abrahamson [2000] proposed
a similar model for the variability, as formed by the aleatory and epistemic components. The
uncertainty known as aleatory is related to the randomness of the earthquake generating process and
there is no way to reduce it. The second type is the epistemic uncertainty, which is originated in the
lack of knowledge of the phenomenon and it is possible to reduce it with additional research and a
better understanding of the different parameters that affect the phenomenon.
Abrahamson [2000] and Bommer [2002] have proposed that the two types of uncertainty can be
treated in different ways to compute the ground-motion demand in seismic hazard assessment: the
aleatory uncertainty is customarily represented by the standard deviation of the PDF for the prediction
equation, and the epistemic uncertainty is represented with the formulation of a logic tree.
Some researchers [e.g. Restrepo-Vlez and Bommer, 2003] have indicated that apart from the scatter
in the probability density function of the prediction equations, the uncertainty in seismic hazard
assessment has some other sources, for example the definition of the geometric features of the seismic
sources, the definition of activity features of the seismic sources such as maximum magnitude,
37
recurrence model and parameters for the recurrence model, and also the selection of the prediction
equation; these sources of uncertainty belong to the group of epistemic uncertainty [Abrahamson,
2000].
Whitman et al. [1997] considered that it is not practical to separate these two components of
variability, and instead they proposed to compute composite best-estimate variability. In HAZUS,
the uncertainty related with the ground-motions demand is represented by the spatial and the temporal
variability components. The spatial variability is implicitly included in the variability of damage
probability matrices or fragility curves, while the uncertainty due to the temporal variability could be
evaluated and/or chosen by the user through the selection of the magnitude M or the recurrence rate
of the earthquake selected for the earthquake scenario [FEMA, 1999].
Figure 3.15 shows the different components for the ground-motion uncertainty models related above.
It is worth underlining that the temporal component proposed in HAZUS, rather than a component of
uncertainty is a deterministic variable, because once a recurrence rate or magnitude M is fixed, the
uncertainty about the temporal component, defined according to HAZUS, does not exist any longer.
Uncertainty / Variability
Kennedy et al. , [1980] Randomness Uncertainty
Abrahamson, [2000] Aleatory Epistemic

of PDF of prediction eq. Logic tree
Spatial Temporal Spatial Temporal
Whitman et al. , [1997] Spatial Temporal

Variability in fragility curves Selection of
Figure 3.15. Components for different uncertainty models
In any case, regardless the definition of uncertainty in ground-motion used, all the components of
variability related to source, path, site and temporal features have to be considered, either in the
seismic hazard assessment and/or directly in the seismic risk assessment, in order to include the full
effect of this uncertainty.
Increasing the level of complexity of this component of uncertainty, the demand could be expressed
also as combination of several peak ground-motion parameters, given the fact that additionally to
acceleration, velocity or displacement, some other peak ground-motion parameters could have
influence in the response of masonry structures [Penazzo, 2001]. Bazzurro and Cornell [2002]
proposed a new methodology to compute the ground-motion demand for several peak ground-motion
parameters at the same time, with a vector-valued probabilistic seismic hazard assessment or VPSHA.
38
This type of analysis is required when several peak-ground motion parameters are going to be
included in the hazard assessment, because it is not correct to assume that peak ground acceleration,
peak ground displacement, duration, spectral ordinates, etc, are all simultaneously certain number of
standard deviation above the median value [Restrepo-Vlez and Bommer, 2003].
3.3.1.2. Uncertainties on damage state threshold. Tables 2.8 and 3.2 show the parameters needed for
defining the damage state thresholds for in-plane and out-of-plane failure mechanisms, respectively.
Those can be considered as the median values for the threshold, and the uncertainty has to be
represented by the standard deviation or the coefficient of variation of every one of the damage state
threshold.
3.3.1.3. Uncertainties on capacity response. The uncertainties on capacity response come from the
propagation of the uncertainty on each one of the parameters affecting the response. Figures 3.16 and
3.17 show the sources of uncertainty on capacity response for in-plane and out-of-planes, respectively.
Those sources have been listed, according to the methodologies proposed in Sections 3.1 and 3.2.
Height
Geometrical Area of resistant walls
Mass distribution
Irregularities
Mixed Geometrical/
Sources of uncertainty Material Properties Type of slabs and roof
on capacity response
for in-plane walls Masonry specific weight
Material Properties Quality of masonry
Age (state of conservation, maintenance, design code)
Others Use (loads)

Pre-existing damages
Figure 3.16. Sources of uncertainty on capacity response for in-plane walls
Height of building and walls

Geometrical Boundary conditions
Thickness of the walls
Mixed Geometrical/ Type of slabs and roof
Sources of uncertainty Material Properties Presence of chains or other restrainers
on capacity response
for out-of-plane walls Masonry specific weight
Material Properties Texture of masonry (double leaf, single leaf,
staggering ratio, shape factor, etc)
Quality of masonry
Others Pre-existing damages
Figure 3.17. Sources of uncertainty on capacity response for out-of-plane walls
39
3.3.2 Definition of the JPDF for Structural Capacity
It has to be recognized that Eqs. (3.27) and (3.39) are not proper descriptions of the dependency of
on T, and this dependency must rather be expressed with Eq. (3.41), by including the uncertainty in the
relationship.
= g (T , ) (3.41)
According to the original proposal by Calvi, an appropriate definition of the JPDF f ,T (, T ) would
be necessary to be included in the methodology at this stage. However, two main problems might
arise: the first is the definition of the JPDF, which should include the effect of all the sources of
uncertainty related in Section 3.3.1 and that cannot be estimated from the sample with enough
accuracy, but rather would be needed to be possibly chosen by means of a heuristic approaching [Pinto
and Giannini, 2001]. The second problem is related with the integration of the JPDF over the failure
domain, in order to compute the probability of failure. In most of the cases the JPDF has a very
complicated mathematical form, and in addition the integration must be performed in a
multidimensional space, being the integral difficult to compute. To avoid complications from the
mathematical point of view, Monte Carlo methods could be applied in order to compute the
probability that DRS exceed the capacity. Some special Monte Carlo techniques, like Importance
Sampling or Latin Hypercube Sampling [McKay, Beckman and Conover, 1979] would be useful to
reduce the computing time.
An alternative path, and somehow an easier one, can be followed to overcome the mathematical
problems related above. Let us consider the classical reliability formulation represented in Eq. (3.42),
for which many references can be found [e.g. Pinto and Franchin, 2003], in which FD() represents the
CDF of the demand being less or equal to some value , and fC() is the PDF of the structural
capacity SC:

Pf = [1 FD ( )] f SC ( )d (3.42)
0
In the particular formulation representing the capacity for a given limit state LS in terms of the
displacement capacity for that given limit state LS and the corresponding effective period TLS,
Eq.(3.42) takes the form of Eq. (3.43), where Pf is the probability of failure, FD is the CDF of the
40
displacement demand LS = x conditional on period TLS = y and f LS TLS is the JPDF for the
displacement limit state LS and effective period at the same limit state TLS.
Pf = [1 FD ( LS = x TLS = y )] f LS TLS (x, y )dxdy (3.43)

y x
The JPDF f LS TLS can be expressed as the product of the PDF of LS conditional to some TLS times the
PDF of TLS, according to Eq. (3.44), leading to the computation of the probability of failure Pf with
Eq.(3.45):
f LS TLS ( x, y ) = f LS TLS (x y ) f T ( y )
LS
(3.44)
Pf = [1 FD ( LS = x TLS = y )] f LS TLS (x y ) f T ( y )dxdy

LS
(3.45)
y x
Extensively known reliability methods like first order (FORM) or second order (SORM) reliability
methods can be used to compute, in an approximate way, the CDF of a complicated function of
random variables [i.e. Pinto and Giannini, 2001], from which the corresponding PDF can be
afterwards numerically evaluated. This is the approach that is proposed here to compute the required
f LS TLS (x y ) and f TLS ( y ) included in Eq.(3.45).
In order to compute the CDF of a random variable using FORM analysis, for example TLS, a G limit
state function must be created, as shown in Eq.(3.46):
G = TLS (3.46)
By assigning parametric values to in a suitable range, for example from 0.01 sec to 4.0 sec, and
running FORM analysis for each discrete value of , the result will be the probability of TLS being less
or equal than , i.e. the cumulative density function CDF of TLS. As a final step, the CDF obtained is
differentiated numerically to obtain f TLS ( y ) .
In a similar fashion, f LS TLS (x y ) can be obtained, taking due care of the fact that this is a conditional
distribution. This problem can be solved by using Eq.(3.27) and fixing the period to some specific
value Ti, considering as random variables just those required to compute the seismic coefficient C, and
41
then following the procedure suggested above. The resulting PDF would be the distribution of LS
conditional to Ti, f LS Ti (x Ti ) . The same procedure must be followed for every single value of T.
3.3.3 Definition of the Seismic Demand
In the classical reliability formulation, the probability that the demand exceeds some given value ,
[1 FD ( )], is represented as the product of the hazard function H(I) of the intensity parameter of
interest I times some factor to take into account the non-linear response of the structure, including
also the uncertainty on the structural response. The hazard function is obtained through a probabilistic
seismic hazard analysis of the site of interest, and in general has the shape shown as an example in
Figure 3.18.
1.0.E-02
Annual frequencies of exceedance
1.0.E-03
1.0.E-04
1.0.E-05
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
PGA [g]
Figure 3.18. Seismic hazard curve of Peak Ground Acceleration for Gaziantep, Turkey, adapted from
Restrepo-Vlez and Bommer [2003]
For example, in a typical application of the reliability theory known as the SAC/FEMA method, which
is intended for the seismic design and assessment of moment resisting frames [Cornell et al., 2002],
the probability that the seismic demand D exceeds a given value d, [1 FD ( d )] , is given by Eq.
(3.47):
k
1 k2
d b
2
D
P (D > d ) = k o e 2 b
2
(3.47)
a
where the first factor is the probability that the seismic hazard exceeds the demand D required to
produce the response d, and the factor with the exponential function accounts for the uncertainty on
42
the structural response. It is worth noting that the probability obtained with Eq.(3.42) corresponds to
the total annual probability of failure.
Now that the concept of demand in the classical seismic reliability has been clarified, it is appropriate
to explain the definition of the seismic demand proposed to be used in this dissertation.
One of the main objectives of the methodology being developed is its application in the evaluation of
damage distribution under a given earthquake scenario. In this case it is assumed that the level of
seismic hazard is chosen from the start, when the earthquake scenario is defined, for example: a
seismic event for an annual frequency of exceedance N equal to 0.0021 (equivalent to a return period
of 475 years), corresponds to the 10% probability of being exceeded q in a lifetime L of 50 years,
assuming the extensively used Poissonian model in Eq. (3.48):
q = 1 e LN (3.48)
When such a selection is made, it is implicit that all the aleatory uncertainties on the seismic hazard
analysis given by the scatter on the prediction equation have been already considered, through the
integration process across the scatter, being left just to consider the epistemic uncertainties inherent to
the definition of the earthquake scenario or the computation of the seismic hazard itself.
In this dissertation it is proposed that for a scenario event the CDF of demand will be computed from
the median values of the displacement response spectrum given by the probabilistic seismic hazard
analysis, and the standard deviation representing the epistemic uncertainty in the definition of the
spectrum. For the computation of this CDF it will be used the widely accepted assumption of
lognormality.
The aleatory component of the uncertainty on the demand it is assumed to be included when the
probabilistic seismic hazard assessment is carried out, thus the uncertainty represented by the standard
deviation of the prediction equation is considered when the integration across the scatter is performed.
To include the aleatory component also as a part of the uncertainty in the definition of the spectrum
would be a double counting of this uncertainty, being an overestimation of the uncertainty and a
mistake from the conceptual point of view.
43
3.3.4 FORM Procedure
Having discussed the goal of using FORM, let us describe in a brief way the general procedure using
as an example Eq.(3.24). In short, the purpose of FORM is to approximate, in the normalized standard
space, the limit state g function by means of a linear function tangent to the g function at the design
point which is the most likely failure point, as shown in Figure 3.19. is the distance from the origin
to the design point and is a unitary vector on the direction of . The components of are called
sensitivity factors and they represent the influence of each random variable ui in the norm of , which
can be assigned with a probabilistic meaning. Details regarding FORM analysis can be found
elsewhere [e.g. Pinto and Giannini, 2001].
Eq.(3.24) can be written in a more explicit way in Eq.(3.49), hence the G limit state function will be
expressed with Eq.(3.50):
1 hT y + 2 ( LS y )hsp
TLS = 2 1
+ TLS (3.49)
1 K 2
K 1 km 1 + 2 g
c km
G = TLS ( km , y , LS , hsp , hT ) (3.50)
U2 Safe domain g(U) = 0

FORM
approximation Failure domain
Design point
u2*
SORM
2 approximation

1 u1* U1
Figure 3.19. Approximation of the limit state function g in the normalized standard space
44
For the purpose of this dissertation, TLS has been assumed equal to zero, and c ,km, y, LS, hsp and hT
and considered as the only random variables on which TLS depends. In upcoming research that is
being currently carried out, a more complete and proper dependence of TLS on a higher number of
random variables is been used, and appropriate values for c and TLS are proposed.
Before applying the procedure, the following statements have to be done:
Due to the nonlinear nature of the limit state function G, as can be seen from Eqs.(3.46) and
(3.49), it is necessary to use the Hasofer-Lind index, which is invariant with respect to the form of
the limit state function.
Considering that the random variables X: { c , hT, hsp, y, LS, km,} are not Gaussian, a
transformation is needed. The variables in X are assumed to be non Gaussian because is not
possible to have negative values for none of them. Besides, in order to use the Hasofer-Lind
index, all the random variables must be transformed into standard Gaussian random variables.
Hereafter it is assumed that all the variables on vector X are lognormally distributed. Nonetheless,
the FORM procedure can be applied to any type of marginal distribution of the individual random
variables, provided that a proper transformation is carried out.
Taken into account that JPDF of X is unknown, the Rosemblatt transformation is not applicable
and the Nataf transformation has to be used [Liu and Der Kiureghian, 1986].
The steps required to found the so called design point or most likely failure point are the
following:
1. To create the correlation matrix P with the coefficient of correlation between variables xij.
Correlation between random variables can be estimated from the statistical data gathered during in
the survey.
[ ]
P = ij (3.51)
2. To uncorrelate the random variables by computing the corrected correlation matrix P by using
the Nataf transformation.
45
3. To factorize matrix Cy = P by using the Cholesky decomposition. Cy = [ij ij' ] = P because
are the covariance matrices of standard normal variables, hence i = j = 1.0:
P = AAT (3.52)
4. To compute the inverse A-1.
5. To select starting point x1 (could be the mean values of X, i).
6. To transform random variables X to correlated standard normal variables Y, with Eqs.(3.53) to

(3.55) for the case of lognormally distributed variables, where correspond to the mean and is
the standard deviation for each random variable on vector X:

2

y = ln x + 1 (3.53)
x

y = ln x 0.5 y2 (3.54)
ln x y
y= (3.55)
y
A transformation from the X space to the Y space of the standard normal variables is always possible,
regardless the type of distribution assumed for any of the random variables. The key point is just to
equate the probability of failure, in order to find the corresponding standard normal value y
corresponding to x. For example, assuming that x is distributed according to the Beta distribution,
with parameters p, q, a and b, the corresponding transformation can be performed with Eq.(3.56),
being 1 the inverse standard normal cumulative density function:
P[x x ] = FBeta ( p ,q ,a ,b ) ( x ) y = 1 (FBeta ( p ,q ,a ,b ) ( x )) (3.56)
7. To transform correlated standard normal variables Y to independent standard normal variables U,

with Eq.(3.57):
U = A-1 Y (3.57)
46
8. To compute the Jacobian J of the transformation and the inverse J-1 .
(u1 ,..., u n )
J= (3.58)
(x1 ,..., x n )
9. To compute the gradient of the limit state function G: G(x).
10. To compute the gradient of the limit state function in the standard normal space g with Eq.(3.59):
g(u) = G(x) J-1 (3.59)
11. To compute u(2) with Eq.(3.60):
g (u (1) )u (1) g (u (1) )

(2)
u = g (u (1) ) (3.60)
g (u )
(1) 2
12. To compute x(2) with Eq.(3.61):
x(2) = x(1) + J-1 (u(2) - u(1)) (3.61)
13. To compare x(1) and x(2) and iterate until convergence is achieved.
14. Once the procedure has converged, to compute the vector of sensitivity factors sen with
Eq.(3.62):
g (u )
sen = (3.62)
g (u )
2
15. To compute the Hasofer-Lind reliability index HL, with Eq.(3.61):
HL = sen u (3.63)
16. To compute the CDF for TLS with Eq.(3.64). It has already been said that FORM is a procedure to
get the approximate CDF of a random variable, based on the CDF of the random variables of
which the first is a function. The accuracy of the approximation will depend on the shape of the g
47
limit state and specially on the flatness or the curvature of the g function in the neighbourhood
around the design point.
FTLS ( HL ) (3.64)
Once this procedure has been coded, it can readily be used to compute the approximate CDF of
basically any random variable related to some others random variables through a complex non-linear
function.
The FORM procedure described above has been coded into a FORTRAN program which is presented
in Annex 1. As an alternative another computer program is freely available and can be adapted to be
used in this assessment procedure [Haukaas, 2001].
3.3.5 Probability of failure for In-plane and Out-of-plane mechanisms
With all the tools required to compute the probability of failure Pf, already available, the procedure is
nearly complete. The details regarding each specific failure mechanism are explained next.
3.3.5.1. Probability of failure for In-plane mechanism. According to Eq.(3.45), apart from the CDF
of the demand FD, it is required to compute the PDFs f TLS ( y ) and f LS TLS (x y ) .
From Eqs.(3.24), (3.26), (3.30) and (3.46), the limit state function for TLS is expressed according to
Eq.(3.65):
1 hT y + 2 ( LS y )hsp
G = 2 1
(3.65)
1 K 2
K 1 km 1 + 2 g
c km
where K1 and K2 are constant for a given building, and are computed with Eqs. (3.66) and (3.67):
1
K1 = Am (3.66)
n
hiWi
Wt n
i=m
h W
j =1
j j
48
W i
K2 = i=m
(3.67)
1.5 Am (1 + m )
Now, by applying the FORM procedure described in Section 3.3.3 the CDF of TLS is computed. The
corresponding PDF f TLS is obtained with Eq.(3.68):
dFTLS
f TLS = (3.68)
dTLS
For LS an equivalent computation is carried out, considering the fact that in this case we are interested
in computing a conditional PDF. Firstly, from Eqs. (3.25), (3.29), (3.30), (3.66) and (3.67) the
displacement LS is expressed as a function of period TLS and the other random variables. Secondly,
the period is set to some deterministic value TLSi , and afterwards the FORM procedure is applied to
the limit state function G shown in Eq.(3.70). Finally, the conditional PDF f LS TLS i (x y ) is
computed with Eq.(3.71):
1
T2 1 K 2
LS = LS2 K1 km 1 + 2 g (3.69)
4 c km
1
TLS2 i 1 K 2
G= K 1 km 1 + 2 g (3.70)
4 c
2
km
dF LS TLS i
f = (3.71)
LS T LS i
d LS
3.3.5.2. Probability of failure for Out-of-plane mechanism. The limit state function G for TLS,
corresponding to the out-of-plane failure mechanism described in Section 3.2, is computed with
Eq.(3.72), being 2, LS, h, , and t the random variables on which G depends, and LS computed
according to Table 3.3 depending on the boundary conditions. The PDF f TLS is also obtained with
Eq.(3.68).
49
1
2 2 LS h 2
G = (3.72)
(1 2 )(1 + )gt
In a similar fashion to the in-plane mechanism, the CDF of LS would be conditional to period TLSi ,
because both LS and TLS depends on some common random variables, namely and , and as a
consequence they are correlated. From Eq.(3.40) the limit state function for LS will be Eq.(3.73):
(1 2 )(1 + )gt
G = TLS2 i (3.73)
2 2h
Additionally, the CDF F LS TLS i (x y ) depends on the boundary conditions, as shown in Table 3.3. The
corresponding PDF f ( x y ) is also computed with Eq.(3.71).

LS T LS i
Table 3.3. LS according to the boundary conditions

Boundary conditions LS = u
Simply-supported wall with vertical load
(1 + 0.75 ) t
at walls top centerline (1 + )
Simply-supported wall with vertical load
t
at walls top edge
Parapet wall with no vertical load t
3.4 DAMAGE PROBABILITY AT URBAN OR REGIONAL SCALE
The main goal of the proposed methodology is the loss estimation at urban or regional scale; thus, a
general framework is still required in order to apply the procedures developed for in-plane and out-of-
plane failure mechanisms, going from the simple computation of the probability of failure of a specific
class of buildings, to a more general loss estimation procedure, including several classes of buildings
for the building stock within a geographical region. This final step in the MeBaSe methodology is
presented in the following.
Figure 3.20 shows a hypothetic region that has been subdivided in n1 zones, each of them
corresponding to a subregion, a census tract, a mail-code zone, a neighbourhood, a block within the
city, or any other subdivision considered suitable for the purpose of the loss assessment process. The
building stock comprised within the particular zonei (i = 1 to n1) is subdivided in building classes
50
CBj(mt,#s,st) (j = 1 to n2), according to the material (masonry type, age, quality) mt, number of stories
#s and possibly, soil type st.
Once the region of interested has been subdivided in zones and the corresponding building classes
have been identified, the probability of failure Pf ij , which is the probability of reaching or exceeding
the specified limit state of interes, for each CBj(mt,#s,st) in each zonei is computed, as explained in
sections 3.3.4.1 for in-plane failure mechanism or 3.3.4.2 for out-of-plane failure mechanism.
3
1
2 5
4
i
7
6
Zonei: {CB1, CB2, CB3, , CBj, , CBn2}

8
n1
Figure 3.20. Subdivision of the hypothetic region in zones
The probability of failure in a given zone for a specific class of buildings CBj(mt,#s,st) would be
computed with Eq.(3.74), where Pf is the probability of failure for a given class of buildings,
CB ( mt , # s , st )
Pf ij is the probability of failure for a single building of the class CBj(mt,#s,st) within the zonei,
computed with Eq.(3.45), and Pmt , # s , st is the joint probability that a building within the population has
a given material mt, certain number of stories #s and an specific soil type st, which can be statistically
inferred from the sample gathered during the data survey:
Pf = Pf ij Pmt , # s , st (3.74)
CB j ( mt , # s , st )
The total probability of failure in the whole region for the same class of buildings CBj(mt,#s,st) would
be computed with Eq.(3.75):
n1
Pf = Pf (3.75)
CB j ( mt , # s , st )T CB j ( mt , # s , st )i
i =1
51
Chapter 4. Application Examples
4. APPLICATION EXAMPLES
4.1 PROBABILITY OF FAILURE FOR IN-PLANE RESPONSE OF A 4-STOREY

UNREINFORCED MASONRY BUILDING
The first application example corresponds to a hypothetical 4-story building, built with unreinforced
brick masonry. The plan view of the typical storey and the faade are shown in Figure 4.1. For the
material properties it has been assumed a specific weight pm equal to 2.0 MN/m3, and a reference shear
strength km equal to 0.20 MPa.
Figure 4.1. Plan view and faade for the in-plane assessment example
The computation of the seismic coefficient C, according to Eqs.(3.28) and (3.29) is shown in Table
4.1, being C = 0.27. Here it has been assumed the correction factor c = 1.0. The weights for the slabs
and applied load have been assumed as typical for reinforced concrete slabs and residential use.
52
Table 4.1. Computation of the seismic coefficient C

Story hs j hj Vj pm W walls W slab W load W j-tot % W walls
[m] [m] [m3] [MN/m3] [MN] [MN] [MN] [MN] [%]
Ground 3.15 3.15 38.21 0.02 0.76 0.45 0.31 1.53 50
1 3.15 6.30 38.21 0.02 0.76 0.45 0.31 1.53 50
2 3.15 9.45 38.21 0.02 0.76 0.45 0.31 1.53 50
3 3.15 12.60 38.21 0.02 0.76 0.90 0.31 1.60 48
h t = 12.6
h sp = 2.0
Story mj Wj km Am Bm m hjWj h i W i W i Cm
2 2 2
[MN-s /m] [MN] [MPa] [m ] [m ] [MN-m] [MN-m] [MN]
Ground 0.156 1.53 0.20 5.24 6.89 1.31 4.82 49.08 5.43 0.27
1 0.156 1.53 0.20 5.24 6.89 1.31 9.64 44.26 3.90 0.27
2 0.156 1.53 0.20 5.24 6.89 1.31 14.46 34.62 2.37 0.31
3 0.163 1.60 0.20 5.24 6.89 1.31 20.16 20.16 0.80 0.46
W t = 6.19 h j W j = 49.08 C = 0.27
If a deterministic assessment was to be performed, just the median values of the drift for a given limit
state would be required. By using median drift values and Eqs. (3.2), (3.3), (3.5), (3.24) and (3.26) it
would be possible to compute the corresponding displacement and effective period for each given
limit state, as shown in Table 4.2. In Figure 4.2 the comparison of median values of capacity and
demand for three different limit states and performance levels (levels of seismic input) is shown.
According to these results it could be concluded that for an expected performance level 1 the capacity
of the building is higher than the demand, hence for this case the building is safe. For the performance
level 2 the building is marginally unsafe but for the performance level 3 the building is clearly unsafe.
The key question would be: how safe or unsafe is the building, considering the uncertainty in demand,
capacity and the definition of the limit state thresholds? Now, let us apply the proposed MeBaSe
methodology to find an answer this question.
Table 4.2. Computation of median values for LS and TLS

LS LS y p LS T LS
[%] [m] [m] [m] [sec]
1-2 0.10 0.009 0.000 0.009 0.36
3 0.30 0.009 0.004 0.012 0.43
4 0.50 0.009 0.007 0.016 0.49
First, the definition of the displacement demand is required. In this example it has been used the same
DRS shown in Figure 3.1, for ag = 0.05g, 0.15g and 0.3g, for performance levels 1, 2 and 3
respectively; S = 1.0 and k = 1.0, defined according to Eurocode 8 [CEN, 2001]. For each limit state,
damping ratios of 5%, 10% and 15% have been used, respectively. The variability of the ground-
53
motion demand D has been assumed to be equal to 0.5, which means that the standard deviation is
0.65 times the mean value.
0.04
Perf. Level 1
Perf. Level 2
Sd (m) 0.03 Perf. Level 3
0.02
LS 4
LS 3
0.01 LS 1-LS 2
0.00
0.0 0.2 0.4 0.6 0.8 1.0
T (sec)
Figure 4.2. Comparison of median values of capacity and demand for each given limit state
Figure 4.3 shows the cumulative density functions CDF for the demand FD(LS) given that TLS = y.
The surface shown in this figure is not the JPDF of TLS-LS; referred in Sections 2.5 and 3.3.2, but
rather it is the CDF for the demand required in Eq.(3.45).
The next step is the definition of the CDF for TLS and LS conditional to TLSi, by using the FORM
procedure. In Tables 4.3 and 4.4 the required parameters, and median and standard deviation values
for the X vector are presented. The standard deviations have been assumed to be representative and
have been used just with the purpose of carrying out this example. Proper values must be selected
from the literature and/or computed from the survey data. Figure 4.4 shows the cumulative density
function CDF for TLS and, as an example, the CDF for LS conditional to TLS = 1.0 sec, that correspond
to the limit state LS4, computed according to the FORM procedure presented in Section 3.3.3. In this
figure different values for the coefficient of correlation ij' have been used to consider the correlation
between all variables in the X vector, in order to show the influence on the estimation of the CDF.
54
FD(LS=x|TLS=y)
T LS [sec]
LS [m]
Figure 4.3. Cumulative density functions CDF for demand FD
Table 4.3. Parameters for computing FTLS and F LS using FORM procedure
Parameters
0.00
1 0.689
2 0.916
K1 0.87
K2 0.39
Table 4.4. Median and standard deviation values for vector X

Vector X x x
km [MPa] 0.20 0.05
hT [m] 12.60 2.00
y 0.001 0.001
LS 0.005 0.001
hsp [m] 2.00 0.50
c 1.00 0.20
55
1.0 1.0
0.9 0.9
0.8 0.8
0.7 0.7
= 0.0
0.6 = 0.0 0.6
LS)
CDF (TLS )
= 0.1
= 0.1
0.5 0.5 = 0.5
CDF (
= 0.5
0.4 0.4 = 0.8
= 0.8
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
T LS [sec] LS [m]
Figure 4.4. Cumulative density functions CDF for TLS, and LS conditional to TLS = 1.0 sec
Finally, by computing the integral in Eq.(3.39), the probability of failure for each given limit state is
obtained, as shown in Table 4.5. For these results a coefficient of correlation ij' = 0.5 has been used.
The selection of the same coefficient of correlation between all the possible pairs of random variables
is rather arbitrary. In fact, intuitively it could be said that a high correlation is expected for example
between the ht and km, whilst hsp could be poorly correlated with all the rest of variables. In a formal
application, an appropriate estimation of the coefficients of correlation can be done.
For each limit state a different level on the demand has been used, corresponding the performance
levels 1, 2 and 3 shown in Figure 4.2.
Table 4.5. Probability of failure for each given Limit State

Limit State Probability of Failure
LS1-LS2 0.002
LS3 0.278
LS4 0.819
When the same seismic demand is used to compute the probability of failure for each limit state, the
probability decreases when going to more demanding limit states, as expected. Table 4.6 shows the
probability of failure for each limit state under performance level 2.
Table 4.6. Probability of failure for each given Limit State for the same seismic demand
Limit State Probability of Failure
LS1-LS2 0.392
LS3 0.278
LS4 0.211
56
In this specific example the differences in the probabilities of failure are not as large as would be
expected, due to the similitude in the displacement response spectra in the range of effective period
corresponding to this building, scaled to the respective damping values, as can be seen in Figure 4.5.
0.04
Perf. Level 1
Perf. Level 2
0.03 Perf. Level 3
Sd (m)
0.02
LS 4
LS 3
0.01 LS 1-LS 2
0.00
0.0 0.2 0.4 0.6 0.8 1.0
T (sec)
Figure 4.5. Comparison of median values of capacity and demand for each given limit state, for
performace level 2 PGA 0.15g
The shear resistant areas on each direction, Am and Bm, have been assumed as deterministic variables
due to the fact that this a specific building, whose dimensions are known. In case that instead of this
building, a set of 4-storey brick masonry buildings were to be analyzed, with no specific information
on single buildings, Am and Bm could be treated both as random variables for the computation of FTLS .
In Figure 4.4 are also presented the CDFs corresponding to coefficients of correlation equal to 0.0, 0.1
and 0.8. It can be seen that the selection of the coefficient of correlation has a quite limited effect in
the computation of FTLS but has some notable influence in F LS . In the same way, the computed
probability of failure varies, as shown in Figure 4.6 for the limit state LS4, being conservative to use a
low or even null coefficient of correlation.
The influence of each random variable in the computation of FTLS and F LS is presented in Figure 4.7,
showing the variation of the sensitivity factors sen with respect to the variation of the coefficient of
correlation ij. The sensitivity factors vary for each design point, this is for each point on the CDF.
The results showed in Figure 4.7 correspond to the sensitivity for the median values of TLS and LS
conditional to TLS = 1.0 sec
57
0.26
0.25
0.24
0.23
0.22
Pf
0.21
0.20
0.19
0.18
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
' ij
Figure 4.6. Variation of the probability of failure Pf

with the coefficient of correlation ij, for LS4
0.6 1.2
ht 1.0
0.4
y
0.8
0.2 LS
h sp 0.6
0.0 km km
0.4
c c
-0.2 0.2
sen
sen
-0.4 0.0
-0.2
-0.6
-0.4
-0.8
-0.6
-1.0 -0.8
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
' ij ' ij
Figure 4.7. Variation of the sensitivity factors sen with the coefficient of correlation ij, for LS4
In this figure it can be seen that for low coefficients of correlation all the random variables used in the
computation of CDF for TLS have influence in the uncertainty, whilst for high coefficients of
correlation the biggest influence in the uncertainty comes from the referential shear strength km, the
total height hT and the yield drift limit y and the correction factor c. In the case of the conditional
CDF for LS, both random variables have a relatively equal influence for low coefficients of
correlation, and for high coefficients of correlation the biggest influence in the uncertainty comes from
the correction factor c, having km an almost negligible influence.
58
4.2 PROBABILITY OF FAILURE FOR OUT-OF-PLANE RESPONSE OF SINGLE

WALLS
The second application example is related with the assessment of the out-of-plane behaviour of the
faade wall at the fourth storey of the building used for the first example. The building has reinforced
concrete slabs and the faade wall at each storey is bearing on the floor slab; hence it can be assumed
that each wall at intermediate levels will behave as simply-supported load-bearing walls, according to
Figure 3.7(c).
It has been selected a median value for the overburden weight ratio equal to 0.50, a factor equal to
0.8, and 2 equal to 0.40. With this median values the corresponding median values for displacement
limit state LS and effective period TLS were computed and then compared with the demand. Figure
4.8 shows the displacement response spectra corresponding to Eurocode 8, for PGA equal to 0.3g, 0.5g
and 1.0g, along with the mean capacity computed for the out-of-plane wall used in this example. It
can be seen that for PGA equal to 0.3g and 0.5g the mean capacity of the wall exceeds the demand,
and the wall would not collapse. When PGA is increased to 1.0g, the capacity of the wall is below the
spectral demand, which means that the wall would be unstable under such demand.
0.40
0.35 Sa 0.3g
0.30 Sa 0.5g
Sa 1.0g
0.25
Sd (m)
0.20 Capacity of the out-of-plane wall
0.15
0.10
0.05
0.00
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
T ne (sec)
Figure 4.8. Comparison of median values of capacity and demand
As the first step in the application of the MeBaSe procedure it has been used the DRS for PGA 1.0g
shown in Figure 4.8. The variability of the ground-motion demand D was again assumed to be equal
to 0.5. Figure 4.9 shows the corresponding cumulative density functions CDF for the demand FD(LS)
given that TLS = y.
59
Then, to obtain the CDF for TLS, and LS conditional to TLSi by using the FORM procedure, the median
and standard deviation values for the X vector presented in Table 4.7 were used. In this example a
coefficient of correlation ij' = 0.50 was also used for the correlation between all variables in the X
vector. Figure 4.10 shows the cumulative density functions CDF for TLS and the CDF for LS
conditional to TLS = 1.5 sec, calculated with the FORM already described.
FD(LS = x|TLS = y)
T LS [sec]
LS [m ]
Figure 4.9. Cumulative density functions CDF for demand FD
Table 4.7. Median and standard deviation values for vector X

Vector X x x
2 0.40 0.10
h [m] 3.00 0.50
0.80 0.20
0.50 0.20
t [m] 0.30 0.05
60
1.0 1.0
0.9 0.9
0.8 0.8
0.7 0.7
= 0.0 = 0.0
0.6 0.6
LS)
CDF (TLS )
= 0.1 = 0.1
0.5 0.5 = 0.5
CDF (
= 0.5
0.4 = 0.8 0.4 = 0.8
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
T LS [sec] LS [m]
Figure 4.10. Cumulative density functions CDF for TLS, and LS conditional to TLS = 1.5 sec
The probability of failure for each given limit state, obtained with the integral shown in Eq.(3.39), are
presented in Table 4.8.
Table 4.8. Probability of failure for different demand levels

PGA Probability of Failure
0.3g 0.047
0.5g 0.183
1.0g 0.540
In the same way as for the in-plane mechanism, in Figure 4.10 are also presented the CDFs
corresponding to coefficients of correlation equal to 0.0, 0.1 and 0.8. In this example the selection of
the coefficient of correlation has an almost negligible effect in the computation of FTLS and a moderate
effect for F LS . Moreover, the influence of each random variable in the computation of FTLS and
F LS has also been analyzed and is presented in Figure 4.11, showing the variation of the sensitivity
factors for the median values of TLS and LS.
For low coefficients of correlation used in the computation of CDF for TLS the biggest influence comes
from the displacement ratio 2 and height of the wall h, and the lowest influence comes from the
thickness of the wall t, being in fact feasible to consider it as a parameter rather than a random
variable, taking into account its null influence. For high coefficients of correlation the biggest
influence in the uncertainty comes again from the displacement ratio 2 whilst the rest of the random
variables have low to null influence. In the case of the conditional CDF for LS, the biggest influence
comes again from the displacement ratio 2, and the other random variable have a medium to low
influence, for all the coefficients of correlation.
61
0.6 1.0
0.4 0.8
0.6
0.2
0.4
0.0
0.2
2 2
-0.2
sen
0.0
sen
h h
-0.4 -0.2

-0.4
-0.6
t -0.6 t
-0.8
-0.8
-1.0 -1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
' ij ' ij
Figure 4.11. Variation of the sensitivity factors sen with the coefficient of correlation ij, for LS4
62
Chapter 5. Survey Data Form
5. SURVEY DATA FORM
The MeBaSe procedure presented in Chapter 3 is complemented with a survey data form, developed
with the purpose of being use on field surveys. The new survey form resembles the appearance of the
GNDT first and second level forms [GNDT, 1993], but includes new parameters required for the
application of the new procedure.
A version of the survey data form with the corresponding Users Manual written in Italian is presented
in Annex 2. The survey form is divided in eight sections, each one being described as follows:
Data related to the survey form
Basic information like date, number of survey form and surveyor is included in this section and is
intended to be used for identification and reference purposes.
Location of the building
In this section the information required for the complete identification of the building within a data
base is collected.
63
Type of usage
Residential, commercial or public usage is considered in this section, including the frequency of use
and the type of activity in case that the main use is commercial. This information is used in the
estimation of the storey masses. This type of data is fundamental in the application of the procedure.
Site morphology
Site features like the position of the building within the block, slopes and the historic origin condition
of the building is considered in this section. This information is presently not used in the procedure
but it has been included for eventual application of other procedures used for calibration or
comparison purposes.
Age and maintenance condition
The age of construction, age and type of eventual retrofit interventions, maintenance condition and
state of degradation of the mortar joints are surveyed in this part. The last two features are
fundamental in the application of the procedure, especially for the selection of the displacement ratio
2.
Structural typology
The structural typology of the vertical elements, the slabs and roof is considered here, using the same
classification proposed by the GNDT [1993]. This information is fundamental in the application of
the procedure, particularly for the definition of the reference shear strength for masonry and the
identification of the possible out-of-plane collapse modes.
General features of the building
The information gathered in this section is related to the plan and height regularity, the direction of the
joists in the floor systems, demolitions, attachments, restrictions, etc. The use of this information has
not been included yet in the procedure, but could have relevance in the determination of the
64
probability of failure for certain types of buildings, thus it is included for eventual updates or
refinements of the procedure.
Other information compiled in this section refers to the total area, the storey height, the shear resistant
areas of walls, and the direction, thickness, length and other features of the walls in the perimeter of
the building. These parameters are fundamental for the evaluation of the in-plane and out-of-plane
probabilities of failure.
Overturning mechanisms associated to existing damages
The last section on the survey form is devoted to the compilation of the possible overturning
mechanisms associated to existing damages, for which the more general out-of-plane collapse modes
have been included in the form.
Finally, as part of the survey form, a website has been developed in order to write the information into
a database. Figure 5.1. shows the main screen of the website.
Figure 5.1. Main screen for the website used to manipulate the database
The website has three different modules, namely: Module 1 which is used to write the information to
the database, Module 2 that helps to read the information stored in the database and allows to modify
65
the information when some correction is required, and Module 3 that creates a text file with the
relevant information required for the application of the MeBaSe procedure. Figures 5.2 to 5.4. shows
some partial views of the three modules.
Figure 5.2. Partial view of the screen used to enter the write the information to the database
Figure 5.3. Partial view of the screen used to control and to modify the information written in the
database
66
Figure 5.4. Output view of the screen used to write the input file required for the MeBaSe procedure
67
Chapter 6. Conclusions
6. CONCLUSIONS
6.1 SUMMARY
A simplified mechanics-based procedure for the seismic risk assessment of unreinforced masonry
buildings (MeBaSe) has been developed, based on the concept of displacement-based approach for
vulnerability evaluation of classes of buildings, that was originally proposed by Calvi [1999]. The
procedure developed can be classified as a direct method because allows the estimation of the
probability of failure of a given building or class of building by using numerical models (mechanical
group), but also makes use of data collected from surveys on populations of buildings from which
relevant statistical parameters used in the analysis are obtained (typological group).
As a first step, a revision of the state-of-the-art procedures for vulnerability assessment of buildings, in
particular unreinforced masonry buildings, was carried out, from which advantages and disadvantages
of the current procedures were identified. Considering this, it can be stated that the MeBaSe procedure
complies with the advantages and solves, at least partially, the disadvantages identified in existing
procedures, because it is a procedure based on displacement capacity and demand, includes the
uncertainty in the demand response spectra, and allows the computation of the probability of reaching
or exceeding certain limit state for a building or class of buildings, which is the direct measure of the
seismic risk being estimated. Besides, the procedure uses a continuous definition of the earthquake
intensity and can be readily applied in real cases. Additionally, the procedure proposed here includes
the explicit computation of the probability of failure for out-of-plane behaviour, and includes also the
treatment of the uncertainty on the and T space, avoiding a complicated definition of the JPDF
f ,T (, T ) .
68
Considering the in-plane mechanisms, the demand represented by the displacement response spectrum
is to be obtained from the regional probabilistic seismic hazard study, and must be defined by the
median or mean value for each spectral period and also by the standard deviation. Concerning the
limit states, appropriate median values corresponding to each given limit state must be used. Brick
masonry, stone masonry, tuff masonry and other types of masonry have different drift limits that have
to be obtained from testing or from the available literature. In case they need to be assumed, a higher
level of uncertainty must be set accordingly. Statistical evaluation of available experimental data is
currently under way.
For the out-of-plane mechanism, the procedure developed in this dissertation is based on the response
of parapets and simple supported walls, either with loadbearing or non-loadbearing walls. Equivalent
procedures for other wall configuration and failure modes are currently under development. In the
same way as the demand defined for the in-plane failure mechanism, the demand is expressed as a
function of period, which is taken from Eurocode 8 to compute the acceleration for out-of-plane walls.
To define proper limit states for the assessment of out-of-plane unreinforced masonry walls it has been
used a semi-rigid nonlinear behaviour of a bending wall described by means of a tri-linear simplified
model. This model is defined by the parameters 1 and 2, estimated from the material properties and
the state of degradation of the mortar.
Regarding the probabilistic aspects of the new procedure, it is accepted that the uncertainty is
originated in three different sources, namely: ground-motion demand, damage state threshold and
capacity response, and that all these are necessary to be considered for a complete description of the
uncertainty in a vulnerability assessment analysis. In order to overcome the numerical problems that
might arise in the definition of the JPDF for TLS and LS an alternative path has been chosen, by using
FORM reliability method to compute in an approximate way the cumulative density function CDF of
LS and TLS. Taking into account all the possible sources of uncertainty, the strategy used here can
treat the uncertainty coming from the material properties and structural response, and the uncertainty
coming from the statistics of the population of buildings. The former is directly included in the
computation of probability of failure of a single building or class of buildings, whilst the latter is
consider in a second stage.
Finally, as a complement of the proposed methodology, a survey data form has been designed to
gather the data field required for the application of the methodology at urban scale in real cases. This
survey form allows gathering the information required for the application of the MeBaSe procedure
and also additional information useful in the application of other procedures that could be eventually
used for calibration or comparison purposes.
69
6.2 FINAL REMARKS
In the definition of the 1 and 2 parameters for the computation of the displacement for any given
limit state, presented in Section 3.1.3.1, a simplified model considering the distributed mass of the
masonry walls and the lumped masses at each storey has been proposed. For this definition a mass
ratio mf / mm equal to 0.3 was used, showing that the parameter 2 is almost insensitive to the variation
of the ductility level. The values proposed for 1 and 2 are considered valid when the weak storey is
formed in the two lower thirds of the building. Nonetheless, a complete and formal computation of
the displacement for the given limit state could also be performed by means of Eq.(3.9).
From the results obtained in the numerical examples presented in Chapter 4 it was stated that the
selection of the coefficient of correlation has null or very limited effect in the computation of the
cumulative density function for the effective period TLS and the displacement limit state LS, and that
the use of a null or low coefficient of correlation between the random variables is conservative, for the
computation of the probability of failure. The relative influence of each random variable used in the
estimation of the cumulative density function for TLS and LS was also performed, showing a variation
of the influence on the total uncertainty of each variable, according to the level of correlation
considered between the random variables.
6.3 FUTURE RESEARCH
The mechanical models used in the definition for the in-plane and out-of-plane response are simple
and do not include factors to consider, for example, irregularities in plan and height for the case of in-
plane failure, or bidirectional bending for out-of-plane failure. More refined mechanical models could
be developed in order to consider more complex behaviour for both failure mechanisms. The MeBaSe
procedure is flexible enough to allow the inclusion of new mechanical models with more parameters;
this task would not complicate in excess the application of the procedure.
The demand model used for the out-of-plane behaviour, taken from Eurocode 8, is still has to be
validated as an adequate representation of the displacement demand for out-of-plane walls, specially
considering the presence of flexible slabs and the possible out-of-phase movement between the floor
and ceiling of a given storey.
Regarding also the out-of-plane failure mechanism, the displacement ratios 1/u and 2/u proposed
by Doherty et al [2002] for three different conditions of the cracked joint, must be confirmed or
70
modified according to the type and quality of the masonry customarily used in the region where the
methodology is to be applied. Besides, the definition of other limit states for out-of-plane failure
modes, apart from stable or collapsed would allow a better correlation with the level of damage.
Finally, considering that unreinforced masonry is usually a material showing highly non-linear
degradation properties, the effect of duration as a component of the demand also needs to be
addressed.
71
Chapter 7. Bibliography
7. BIBLIOGRAPHY
1. Abrahamson, N.A. [2000]. State of the Practice of Seismic Hazard Evaluation, Proceed.
Earthquake Engineering, GeoEng2000 Conference, Melbourne, Australia, November 2000.
2. Bazzurro, P. and Cornell, C.A. [2002]. "Vector-Valued Probabilistic Seismic Hazard Analysis
(VPSHA)", Proceed. 7th U.S. National Conference on Earthquake Engineering, Boston, MA,
July.
3. Braga, F., Dolce, M. and Liberatore, D. [1982]. Southern Italy November 23, 1980 Earthquake:
Sect. 5 A Satistical Study on Damaged Buildings and a Ensuing Review of the MSK-76 Scale.
Proceed. 7th European Conference on Earthquake Engineering, Athens, September.
4. Benedetti, D. and Petrini, V. [1984]. Sulla Vulnerabilit Sismica di Edifici in Muratura: Proposta
di un Metodo di Valutazione. Lindustria delle Costruzioni No.149, marzo. (In Italian).
5. Bernardini, A. [2000]. La Vulnerabilit degli Edifici: Valutazione a Scala Nazionale della

Vulnerabilit Sismica degli Edifici Ordinari. (A cura di A. Bernardini) CNR-GNDT, Roma. (In
Italian).
6. Bernardini, A., Gori, R. and Modena, C. [1990]. An Application of Coupled Analytical Models
and Experiential Knowledge for Seismic Vulnerability Analyses of Masonry Buildings. In A.
Koridze (ed) Engineering Aspects of Earthquake Phenomena, Vol. 3: 161-180. Oxon: Omega
Scientific.
7. Bommer, J.J. [2002]. Deterministic vs. Probabilistic Seismic Hazard Assessment: An

Exaggerated and Obstructive Dichotomy. Journal on Earthquake Engineering, Vol. 6, Special
Issue No. 1, 43-73.
8. Calvi, G.M. [1999]. A Displacement-Based Approach for Vulnerability Evaluation of Classes of

Buildings. Journal on Earthquake Engineering, Vol 3, N.3, 411-438.
9. CEN. [2001]. Eurocode 8: Design of Structures for Earthquake Resistance. Part 1: General
Rules, Seismic Actions and Rules for Buildings. Draft 4, Final Project Team Draft (Stage 34),
December 2001.
10. Cornell, A. C., Jalayer, F., Hamburger, R., and Fouth, D. [2002]. The Probabilistic Basis for the
2000 SAC/FEMA Steel Moment Frame Guidelines. Journal of Engineering Mechanics ASCE,
128(4):526-533.
72
11. Corsanego, A. and Petrini, V. [1990]. Seismic Vulnerability of Buildings - Work in Progress.
Proceed- SEISMED 3. Trieste, Italy.
12. DAyala, D., Spence, R., Oliveira C. and Pomonis, A. [1997]. Earthquake Loss Estimation for
Europes Historic Town Centres. Earthquake Spectra, 13(4), 773-794.
13. DAyala, D. and Speranza, E. [2002]. An Integrated Procedure for the Assessment of Seismic
Vulnerability of Historic Buildings. Proceed. 12th European Conference on Earthquake
Engineering, London. Paper No.561.
14. Di Pasquale, G., Goretti, A., Dolce, M. and Martinelli, A. [2001]. Confronto fra Differenti
Modelli di Vulnerabilit degli Edifici. LIngegneria Sismica in Italia, Atti X Convegno
Nazionale ANIDIS, Potenza-Matera. (In Italian).
15. Doherty, K.T., Griffith, M.C., Lam, N. and Wilson, J. [2002]. Displacement-based Seismic
Analysis for Out-of-plane Bending of Unreinforced Masonry Walls . Earthquake Engineering
and Structural Dynamics, 31: 833-850.
16. Dolce, M. [1997]. La Valutazione della Vulnerabilit per le Analisi di Rischio e gli Scenari di
Danno. LIngegneria Sismica in Italia, Atti 8 Convegno Nazionale ANIDIS, Taormina. (In
Italian).
17. Dolce, M., Kappos, A., Zuccaro, G., and Coburn, A.W. [1994]. State of the Art Report of W.G.
3 Seismic Risk and Vulnerability. Proceed. 10th European Conference on Earthquake
Engineering, Wien.
18. FEMA [1999]. HAZUS 1999. Earthquake Loss Estimation Methodology. Technical Manual.
Federal Emergency Management Agency, Washington D.C.
19. Giovinazzi, S., Lagomarsino, S. [2001]. Una Metodologia per lanalisi di Vulnerabilit Sismica
del Costruito. LIngegneria Sismica in Italia, Atti X Convegno Nazionale ANIDIS, Potenza-
Matera. (In Italian).
20. Giuffr, A. [1999]. Sicurezza e Conservazione del Centri Storici, Il Caso Ortigia. (A cura di
A. Giuffr) CNR-GNDT, Ed. Laterza, 2nd ed, Bari. (In Italian).
21. GNDT. [1993]. Rischio Sismico di edifici Pubblici Parte I Aspetti Metodologici. Consiglio
Nazionale delle Richerche - Gruppo Nazionale per la Difesa dai Terremoti, Roma. (In Italian).
22. Griffith, M.C., Magenes, G., Melis, G. and Picchi, L. [2003] Evaluation of Out-of-Plane Stability
of Unreinforced Masonry Walls Subjected to Seismic Excitation. Journal on Earthquake
Engineering, Vol. 7, Special Issue No. 1, 141-169.
23. Gruntal, G. [1998]. European Macroseismic Scale 1998. Chaiers dur Centre Europen du
Godynamique et de Sismologie, Volume 15, Luxembourg.
24. Haukass, T. [2001]. FERUM, Finite Element Reliability Using Matlab.

http://www.ce.berkeley.edu/~haukaas/FERUM/ferum.html.
25. Kennedy, R.P., Cornell, C.A., Campbell, R.L., Kaplan, S., and Perla, H.F. [1980]. Probabilistic
Seismic Safety on an Existing Nuclear Power Plant. Nuclear Engineering and Design, Vol
59(2): 315-338
26. Kircher, C., Nassar, A., Kustu, O. and Holmes, W. [1997]. Development of Building Damage
Functions for Earthquake Loss Estimation. Earthquake Spectra, 13(4): 663-682.
73
27. Liu, P and Der Kiureghian, A. [1986]. Multivariate Distribution Models with Prescribed
Marginals and Covariances. Probabilistic Engineering Mechanics, Vol 1, N2.
28. Magenes, G. and Calvi, G.M. [1997]. In-Plane Seismic Response of Brick Masonry Walls.
Earthquake Engineering and Structural Dynamics, Vol 26, 1091-1112.
29. McKay, Beckman and Conover. [1979] Technometrics, 21(2):239-245.
30. Ministero dei Lavori Pubblici. [1981]. Istruzioni per la Compilazione della Scheda di
Rilevamento, Esposizione e Vulnerabilit Sismica degli Edifici. GNDT, Regioni Emilia-
Romagna e Toscana, Firenze. (In Italian).
31. NZSEE. [2002]. Assessment and Improvement of the Structural Performance of Buildings in
Earthquake. Recommendations of an NZSEE Study Group on Earthquake Risk Buildings
Prepared for the Building Industry Authority.
32. Paulay, T., and Priestley, M.J.N. [1992]. Seismic Design of Reinforced Concrete and Masonry
Structures, John Wiley & Sons, New York.
33. Penazzo, P. [2001]. Influence of strong-motion characteristics on inelastic demand in masonry

structures. MSc Dissertation. London: Imperial College.
34. Pinto, P.E. and Giannini, R. [2001]. Methods for Seismic Reliability Analysis. Notes for a
course at the European School of Advanced Studies in Reduction of Seismic Risk-ROSE School.
35. Pinto, P.E. and Franchin, P. [2003]. Probabilistic Seismic Risk Assessment of Structures.
Notes for a Short Course at the European School of Advanced Studies in Reduction of Seismic
Risk-ROSE School.
36. Restrepo-Vlez, L.F., and Bommer, J.J. [2003] An Exploration of the Nature of the Scatter in
Ground-Motion Prediction Equations and the Implications for Seismic Hazard Assessment..
Journal on Earthquake Engineering, Vol. 7, Special Issue No. 1, 1-29.
37. Shibata, A. and Sozen, M.A. [1976]. Substitute-Structure Method for Seismic Design in R/C.
Journal of the Structural Division ASCE, 102:1-18.
38. Sudret, B. and Der Kiureghian, A. [2000]. Stochastic Finite Element Methods and Reliability, A
State of the Art Report. Department of Civil and Environmental Engineering , University
of California at Berkeley, Report UCB/SEMM-2000/08.
39. Whitman, R.V. [1973]. Damage Probability Matrices for Prototype Buildings. Massachusetts
Institute of Technology, Department of Civil Engineering Research, Report R73-57, Cambridge,
Massachusetts.
40. Whitman, R.V., Anagnos, T., Kircher, C., Lagorio, H.C., Lawson, R.S., and Schenider, P. [1997].
Development of a National Earthquake Loss Methodology. Earthquake Spectra, 13(4): 643-
661.
74
Annex 1
FORTRAN Code for the MeBaSe Procedure
c**********************************************************************]
c
c MeBaSe
c MECHANICAL BASED PROCEDURE FOR THE SEISMIC RISK ESTIMATION OF
c UNREINFORCED MASONRY BUILDINGS
c
c MAIN PROGRAM: main.for
c Coded: Luis F. Restrepo V.
c January 2002
c Ver 1.0
c
c INPUT VARIABLES :
c
c OUTPUT VARIABLES :
c
c**********************************************************************]
program main
implicit none
integer io, n1, n2, n3, n4, n5
c----------------------------------------------------------------------]
c Reads input file: input.dat
c----------------------------------------------------------------------]
write(*,*) ' 1. Reading input file'
call readfirstline(io,n1,n2,n3,n4,n5)
call mebase(io,n1,n2,n3,n4,n5)
end
A1.1
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: mebase.for
c November 2002
c Ver 1.0
c
c INPUT VARIABLES :
c
c io : Flag 1: In-plane 2:Out-of-plane
c n1 : Number of constants
c n2 : Number of random variables for T LS
c n3 : Number of random variables for Delta LS
c n4 : Number of points defining DRS
c n5 : Number of discretization points
c
c
c pf : Probability of failure
c
c**********************************************************************]
subroutine mebase(io,n1,n2,n3,n4,n5)
integer io, n1, n2, n3, n4, n5, i, j, eqcn, naux

double precision xd(n1+1), m(n2), sd(n2), x(n2), p(n2,n2), d1, d2
double precision prob, tls(n5), cdftls(n5), deltals(n5)
double precision cdfdeltals(n5,n5), pdftls(n5), pdfdeltals(n5,n5)
double precision pf, drs(n4,2), drsint(n5), sddrs, sdy, my, y
double precision cdfdemand(n5,n5), xaux(n2)
c----------------------------------------------------------------------]
c----------------------------------------------------------------------]
call readdata(n1,n2,d1,d2,xd,m,sd,x,p,drs,sddrs)
c----------------------------------------------------------------------]
c Computes the CDF of Tls
c----------------------------------------------------------------------]
open(6,file = 'output.dat')
write(*,*) ' 2. Computing CDF of Tls'
write(6,1000)
xd(1) = 0.0d0
xd(n1+1) = 0.0d0
eqcn = 1
naux = n2
prob = 0.0d0
do i = 1,n2
xaux(i) = x(i)
end do
do i = 1,n5
xd(1) = xd(1)+d1
call form(io,eqcn,n1,n2,naux,xd,m,sd,xaux,p,prob)
tls(i) = xd(1)
cdftls(i) = prob
write(6,1100) tls(i), cdftls(i)
end do
c----------------------------------------------------------------------]
c Computes the PDF of Tls
c----------------------------------------------------------------------]
write(*,*) ' 3. Computing PDF of Tls'

call pdf(n5,tls,cdftls,pdftls)
c----------------------------------------------------------------------]
c Computes the CDF of Deltals conditional on Tls(i)
c----------------------------------------------------------------------]
open(6,file = 'output.dat')
write(*,*) ' 4. Computing CDF of Deltals'
write(6,1200)
do i = 1,n5
write(*,*) 'tls(',i,')'
write(6,*) 'tls(',i,')'
xd(1) = 0.0d0
xd(n1+1) = tls(i)
eqcn = 2
naux = n3
A1.2
Annex 1
prob = 0.0d0
do j = 1,n2
xaux(j) = x(j)
end do
do j = 1,n5
xd(1) = xd(1)+d2
call form(io,eqcn,n1,n2,naux,xd,m,sd,xaux,p,prob)
deltals(j) = xd(1)
cdfdeltals(j,i) = prob
c ******************** These lines are to avoid ********************
c ********* numerical problems when CDF is already > 0.999 *********
c *************** when displacement is extremely small *************
if(prob.ge.0.999) then
xd(1) = xd(1)-d2
do k = j,n5
xd(1) = xd(1)+d2
deltals(k) = xd(1)
cdfdeltals(k,i) = prob
write(6,1100) deltals(k), cdfdeltals(k,i)
end do
goto 5
else
end if
c ******************************************************************
write(6,1100) deltals(j), cdfdeltals(j,i)
end do
5 continue
end do
c do i = 1,n5
c do j = 1,n5
c write(6,1100) deltals(j), cdfdeltals(i,j)
c end do
c end do
c----------------------------------------------------------------------]
c Computes the PDF of Deltals conditional on Tls(i)
c----------------------------------------------------------------------]
write(*,*) ' 5. Computing PDF of Deltals'
do i = 1,n5
write(6,*) 'tls(',i,')'
call pdf(n5,deltals,cdfdeltals(:,i),pdfdeltals(:,i))
do j = 1,n5
write(6,1100) deltals(j), pdfdeltals(j,i)
end do
end do
c----------------------------------------------------------------------]
c Interpolates the DRS to the same Tls of the PDF
c----------------------------------------------------------------------]
write(*,*) ' 6. Interpolating the Demand Response Spectrum'
do i = 1,n5
j = 0
10 j = j+1
if(tls(i).eq.drs(j,1)) then
drsint(i) = drs(j,2)
else
if(tls(i).lt.drs(j+1,1)) then
drsint(i) = (drs(j+1,2)-drs(j,2))*(tls(i)-drs(j,1))/
1 (drs(j+1,1)-drs(j,1))+drs(j,2)
else
goto 10
end if
end if
end do
c----------------------------------------------------------------------]
c Computes the CDF of the demand for a given Tls
c----------------------------------------------------------------------]
write(*,*) ' 7. Computing CDF of the demand for a given Tls'

call probdemand(n5,sddrs,drsint,deltals,cdfdemand)
c----------------------------------------------------------------------]
c Computes the probability of failure Pf
c----------------------------------------------------------------------]
write(*,*) ' 8. Computing the probability of failure Pf'
pf = 0.0d0
do i = 1,n5
A1.3
Annex 1
do j = 1,n5
pf = pf+(1-cdfdemand(i,j))*pdfdeltals(j,i)*pdftls(i)
end do
end do
write(*,1500) pf
write(6,1500) pf
c----------------------------------------------------------------------]
c Formats
c----------------------------------------------------------------------]
1000 format(/, 10x, 'Tls(i) CDF of Tls(i)', /)

1100 format(5x, F10.4,3x, F12.10)
1200 format(/, 6x, 'Deltals(i) CDF of Deltals(i)', /)
1300 format(/, 10x, 'Tls(i) PDF of Tls(i)', /)
1400 format(/, 6x, 'Deltals(i) PDF of Deltals(i)', /)
1500 format(2x, 'PROBABILITY OF FAILURE Pf = ', F5.3)
return
end
A1.4
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: readfirstline.for
c November 2002
c Ver 1.0
c
c INPUT VARIABLES :
c
c
c n3 : Number of random variables for Delta LS
c n4 : Number of points defining DRS
c n5 : Number of discretization points
c
c**********************************************************************]
subroutine readfirstline(io,n1,n2,n3,n4,n5)
implicit none
integer io, n1, n2, n3, n4, n5
c----------------------------------------------------------------------]
c----------------------------------------------------------------------]
open(5,file = 'input.dat')
read(5,*) io, n1, n2, n3, n4, n5

return
end
A1.5
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: readdata.for
c November 2002
c Ver 1.0
c
c INPUT VARIABLES :
c
c
c
c d1 : Step increment for T LS
c d2 : Step increment for Delta LS
c xd : Vector of constants
c m : Vector of mean values of X
c sd : Vector of standard deviation values of X
c x : Vector of initial values of X in FORM analysis
c p : Matrix of correlation coefficients
c drs(i,1) : Periods of the Demand Response Spectra
c drs(i,2) : Ordinates of the Demand Response Spectra
c sddrs : Standard deviation of thd Demand Response Spectra
c
c**********************************************************************]
subroutine readdata(n1,n2,d1,d2,xd,m,sd,x,p,drs,sddrs)
implicit none
integer i, j, n1, n2
double precision d1, d2
double precision xd(n1), m(n2), sd(n2), x(n2), p(n2,n2), drs(20,2)
double precision sddrs
c----------------------------------------------------------------------]
c----------------------------------------------------------------------]
read (5,*) d1, d2
do i = 1,n1
read(5,*) xd(i)
end do
do i = 1,n2
read(5,*) m(i), sd(i), x(i)
end do
read (5,*) ((p(i,j), j = 1,n2), i = 1,n2)

read (5,*) ((drs(i,j), j = 1,2), i = 1,20)
read(5,*) sddrs
return
end
A1.6
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: form.for
c Coded: Luis F. Restrepo V. - Convergence criteria taken from FERUM
c November 2002
c Ver 1.0
c
c This subroutine performs the FORM (First Order Reliability Method)
c analysis using the Nataf transformation
c
c INPUT VARIABLES :
c
c eqcn : Equation type (1. Tls, 2. Deltals)
c n2 : Number of random variables
c naux : Number of 'efective' random variables
c x : Vector of initial values of X
c p : Correlation matrix P
c
c
c prob : Probability of failure
c
c**********************************************************************]
subroutine form(io,eqcn,n1,n2,naux,xd,m,sd,x,p,prob)
implicit none
integer i, j, k, n1, n2, naux, io, eqcn, flag, iter, maxiter

double precision m(n2), sd(n2), xd(n1+1), x(n2), p(n2,n2), sdy, my
double precision maux, sdaux, paux, alpha, pp, low, ilow, y, u, ux
double precision xaux, dja, ja, ija, gradgcap, gradg, delta, xx
double precision gcap0, gcap, gcapf, g, tol1, tol2, err
double precision normg, od, normd, d(n2), beta, prob
allocatable :: maux(:), sdaux(:), paux(:,:), alpha(:), pp(:,:)
allocatable :: low(:,:), ilow(:,:), y(:), u(:), ux(:), xaux(:)
allocatable :: dja(:,:), ja(:,:), ija(:,:), gradgcap(:), gradg(:)
allocatable :: xx(:)
c----------------------------------------------------------------------]
c Form vectors y matrices for maux, sdaux, xaux and paux
c----------------------------------------------------------------------]
allocate(maux(naux),sdaux(naux),paux(naux,naux),alpha(naux))
allocate(pp(naux,naux),low(naux,naux),ilow(naux,naux),y(naux))
allocate(u(naux),ux(naux),xaux(naux),dja(naux,naux),ja(naux,naux))
allocate(ija(naux,naux),gradgcap(naux),gradg(naux),xx(naux))
do i = 1,naux
maux(i) = m(i+n2-naux)
sdaux(i) = sd(i+n2-naux)
xaux(i) = x(i+n2-naux)
do j = 1,naux
paux(i,j) = p(i+n2-naux,j+n2-naux)
end do
end do
c----------------------------------------------------------------------]
c Computes correlation matrix P' by using the Nataf transformation
c----------------------------------------------------------------------]
call nataf(naux,maux,sdaux,paux,pp)
c----------------------------------------------------------------------]
c Performs the Cholesky decomposition of P' = A * Atrans
c----------------------------------------------------------------------]
call cholesky(naux,pp,low)
c----------------------------------------------------------------------]
c Performs the inversion of A
c----------------------------------------------------------------------]
call inversion(naux,low,ilow)
c----------------------------------------------------------------------]
c Transforms r.v.'s X to correlated standard normal r.v.'s Y
c----------------------------------------------------------------------]
A1.7
Annex 1
do i = 1,naux
sdy = sqrt(log((sdaux(i)/maux(i))**2+1))
my = log(maux(i))-0.5d0*sdy**2
y(i) = (log(xaux(i))-my)/sdy
end do
c----------------------------------------------------------------------]
c Transforms r.v.'s Y to uncorrelated standard normal r.v.'s U
c---------------------------------------------------------------------]
do i = 1,naux
u(i) = 0.0d0
end do
do i = 1,naux
do j = 1,naux
u(i) = u(i)+ilow(i,j)*y(j)
c ********* numerical problems when the transformed r.v.' **********
c *********************** are extremely large **********************
if(ux(i).gt.1000000) then
prob = 0.999999
goto 5
else
end if
c ******************************************************************
end do
end do
c----------------------------------------------------------------------]
c Starts iteration to find the 'design point'
c----------------------------------------------------------------------]
flag = 0
tol1 = 0.001
tol2 = 0.001
iter = 1
maxiter = 100
call feval(io,eqcn,n1,naux,xd,xaux,gcap0)
do while(flag.eq.0)
c----------------------------------------------------------------------]
c Transforms r.v.'s U to the original space of X
c----------------------------------------------------------------------]
do i = 1,naux
ux(i) = 0.0d0
do j = 1,naux
ux(i) = ux(i)+low(i,j)*u(j)
end do
end do
do i = 1,naux
my = log(maux(i))-0.5d0*sdy**2
xaux(i) = exp(ux(i)*sdy+my)
end do
c----------------------------------------------------------------------]
c Computes Jacobian of the transformation J and the inverse of J
c----------------------------------------------------------------------]
do i = 1,naux
do j = 1,naux
dja(i,j) = 0.0d0
ja(i,j) = 0.0d0
ija(i,j) = 0.0d0
end do
end do
do i = 1,naux
do j = 1,naux
dja(i,i) = 1/sdy/xaux(i)
end do
end do
do i = 1,naux
do j = 1,naux
do k = 1,naux
ja(i,j) = ja(i,j)+ilow(i,k)*dja(k,j)
end do
end do
end do
A1.8
Annex 1
call inversion(naux,ja,ija)
c----------------------------------------------------------------------]
c Computes gradient of the G limite-state function
c----------------------------------------------------------------------]
call feval(io,eqcn,n1,naux,xd,xaux,gcap)
do i = 1,naux
xx(i) = 0.0d0
gradgcap(i) = 0.0d0
end do
do i = 1,naux
delta = sdaux(i)/500.0d0
do j = 1,naux
xx(j) = xaux(j)
end do
xx(i) = xaux(i) + delta
call feval(io,eqcn,n1,naux,xd,xx,gcapf)
gradgcap(i) = (gcapf-gcap)/delta
end do
c----------------------------------------------------------------------]
c Computes gradient of the transformed g limite-state function
c----------------------------------------------------------------------]
do i = 1,naux
gradg(i) = 0.0d0
end do
do i = 1,naux
do j= 1,naux
gradg(i) = gradg(i)+gradgcap(j)*ija(j,i)
end do
end do
c----------------------------------------------------------------------]
c Checks convergence
c----------------------------------------------------------------------]
normg = 0.0d0
od = 0.0d0
normd = 0.0d0
do i = 1,naux
normg = normg+gradg(i)**2
alpha(i) = 0.0d0
xx(i) = 0.0d0
end do
normg = sqrt(normg)
do i = 1,naux
alpha(i) = -gradg(i)/normg
end do
do i = 1,naux
od = od+u(i)*alpha(i)
end do
do i = 1,naux
xx(i) = u(i)-od*alpha(i)
end do
do i = 1,naux
normd = normd+xx(i)**2
end do
normd = sqrt(normd)
od=abs(gcap/gcap0)
if((abs(gcap/gcap0)<tol1).and.(normd<tol2)) then
flag = 1
else
call searchcriteria(naux,gcap,gradg,u,d)
do i = 1,naux
u(i) = u(i)+0.30*d(i)
end do
iter = iter+1
end if
if(iter.eq.maxiter) then
write(*,*) 'Maximum number of iterations exceeded'
stop
else
A1.9
Annex 1
end if
c ************ numerical problems when the probability *************
c ************************ is extremely low ************************
if(iter.eq.(maxiter-1)) then
prob = 0.0d0
goto 5
end if
c ******************************************************************
c *********** It seems that these lines are not required ***********
if(iter.eq.maxiter) then
write(*,*) 'Maximum number of iterations exceeded'
stop
else
end if
c ******************************************************************
end do
c----------------------------------------------------------------------]
c Computes Hasofer-Lind reliability index and probability of failure
c----------------------------------------------------------------------]
beta = 0.0d0
do i = 1,naux
beta = beta+alpha(i)*u(i)
end do
call normalcdf(-beta,prob)
5 return
end
A1.10
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: nataf.for
c November 2002
c Ver 1.0
c
c This subroutine computes the correlation matrix P' by using the
c Nataf transformation (Liu and Der Kiureghian, 1986). Is valid just
c for variables with lognormal distribution
c
c INPUT VARIABLES :
c
c n : Number of random variables
c p : Correlation matrix P
c
c
c pp : Correlation matrix P'
c
c**********************************************************************]
subroutine nataf(n,m,sd,p,pp)
integer i, j, n
double precision m(n), sd(n), p(n,n), pp(n,n), f, cvi, cvj
do i = 1,n
cvi = sd(i)/m(i)
do j = 1,n
cvj = sd(j)/m(j)
f = (log(1+p(i,j)*cvi*cvj))/(p(i,j)*sqrt((log(1+cvi**2))*
1 (log(1+cvj**2))))
pp(i,j) = p(i,j)*f
end do
end do
return
end
A1.11
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: cholesky.for
c Coded: Luis F. Restrepo V. - Taken from NUMERICAL RECIPES IN FORTRAN
c November 2002
c Ver 1.0
c
c This subroutine performs the decomposition of A = L * Ltransp
c
c INPUT VARIABLES :
c
c n : Dimension of matrix 'A'
c x : Positive-definite symmetric matrix 'A' to decompose
c
c
c lo : Lower triangular matrix L
c
c**********************************************************************]
subroutine cholesky(n,x,lo)
integer i, j, k, n
double precision x(n,n), a(n,n), lo(n,n), d(n), sum
do i = 1,n
do j = 1,n
a(i,j) = x(i,j)
end do
end do
do i = 1,n
do j = i,n
sum = a(i,j)
do k = i-1,1,-1
sum = sum-a(i,k)*a(j,k)
end do
if(i.eq.j)then
if(sum.le.0.)pause 'cholesky failed'
d(i) = sqrt(sum)
else
a(j,i) = sum/d(i)
endif
end do
end do
do i = 1,n
do j = 1,n
lo(i,j) = a(i,j)
end do
end do
do i = 1,n
do j = 1,n
if(j.gt.i) then
lo(i,j) = 0.0d0
else
if(j.eq.i) then
lo(i,j) = d(i)
else
continue
end if
end if
end do
end do
return
end
A1.12
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: inversion.for
c Coded: Luis F. Restrepo V. - Taken from www.nauticom.net/www/jdtaft
c November 2002
c Ver 1.0
c
c This subroutine performs the inversion of A
c
c INPUT VARIABLES :
c
c n : Dimension of matrix 'A'
c a : Positive-definite symmetric matrix 'A' to invert
c
c
c ia : Inverse matrix of 'A'
c
c**********************************************************************]
subroutine inversion(n,a,ia)
integer i, j, k, n, n2
double precision a(n,n), ia(n,n), d(n,2*n), alpha, beta
c----------------------------------------------------------------------]
c Creates the reduction matrix
c----------------------------------------------------------------------]
n2 = 2*n
do i = 1,n
do j = 1,n
d(i,j) = a(i,j)
d(i,n+j) = 0.0d0
end do
d(i,n+i) = 1.0d0
end do
c----------------------------------------------------------------------]
c Performs the reduction
c----------------------------------------------------------------------]
do i = 1,n
alpha = d(i,i)
if(alpha.eq.0.0d0) go to 10
do j = 1,n2
d(i,j) = d(i,j)/alpha
end do
do k = 1,n
if((k-i).eq.0.0d0) go to 5
beta = d(k,i)
do j = 1,n2
d(k,j) = d(k,j) - beta*d(i,j)
end do
5 end do
end do
c----------------------------------------------------------------------]
c Copy results into output matrix
c----------------------------------------------------------------------]
do i = 1,n
do j = 1,n
ia(i,j) = d(i,j+n)
end do
end do
return
10 print *,'*** ERROR: Trying to invert a singular matrix ***'
end
A1.13
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: feval.for
c November 2002
c Ver 1.0
c
c This subroutine evaluates the G limit state function
c
c INPUT VARIABLES :
c
c eqcn : Equation type (1. Tls, 2. Deltals)
c n2 : Number of random variables
c x : Vector of random variables
c
c
c g : function value
c
c**********************************************************************]
subroutine feval(io,eqcn,n1,n2,xd,x,g)
integer n1, n2, io, eqcn

double precision xd(n1+1), x(n2), g, num, den, grav, pi
grav = 9.81d0
pi = 3.14159265358979
if(io.eq.0) then
if(eqcn.eq.1) then
num = xd(2)*x(1)*x(2)+xd(3)*(x(3)-x(2))*x(4)
den = grav/x(6)*xd(4)*x(5)*(1+xd(5)/x(5))**0.5
g = 2.0d0*pi*sqrt(num/den)-xd(1)
else
if(eqcn.eq.2) then
num = grav*xd(4)*x(1)/x(2)*(1+xd(5)/x(1))**0.5
g = num/4/(pi**2)*(xd(6)**2)-xd(1)
else
write(*,*) 'Wrong Equation number!'
stop
end if
end if
else
if(eqcn.eq.1) then
num = pi**2*x(1)*x(3)*x(5)*x(2)
den = x(3)*(1-x(1))*(1+x(4))*grav*x(5)
g = sqrt(num/den)-xd(1)
else
if(eqcn.eq.2) then
num = x(3)*(1-x(1))*(1+x(4))*grav*x(5)
den = pi**2*x(1)*x(2)
g = (xd(2)**2)*num/den-xd(1)
else
write(*,*) 'Wrong Equation number!'
stop
end if
end if
end if
return
end
A1.14
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: searchcriteria.for
c Coded: Luis F. Restrepo V. - Adapted from FERUM
c November 2002
c Ver 1.0
c
c This subroutine computes the search direction
c
c INPUT VARIABLES :
c
c gcap : G limite-state function
c gradgcap : Gradient of the G limite-state function
c u : Random variables in the U space
c
c
c d : Search direction
c
c**********************************************************************]
subroutine searchcriteria(n2,gcap,gradg,u,d)
implicit none
integer i, n2
double precision gcap, gradg(n2), u(n2), d(n2), alpha(n2),normg
double precision daux
normg = 0.0d0
daux = 0.0d0
do i = 1,n2
normg = normg+gradg(i)**2
alpha(i) = 0.0d0
d(i) = 0.0d0
end do
normg = sqrt(normg)
daux = gcap/normg
do i = 1,n2
alpha(i) = -gradg(i)/normg
end do
do i = 1,n2
daux = daux+alpha(i)*u(i)
end do
do i = 1,n2
d(i) = daux*alpha(i)-u(i)
end do
return
end
A1.15
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: normalcdf.for
c Coded: Luis F. Restrepo V. - some parts taken from NUMERICAL RECIPES IN FORTRAN
c November 2002
c Ver 1.0
c
c This subroutine computes the normal cdf of Beta
c
c INPUT VARIABLES :
c
c beta : Normal standard variable
c
c
c prob : Probability of being less or equal than beta
c
c**********************************************************************]
subroutine normalcdf(beta,prob)
integer i, it, n
double precision beta, prob, betahl, pi, delta, x, sum
pi = 3.14159265358979
n = 20
prob = 0.0d0
if(beta.ge.0.0d0)then
betahl = beta
it = 2**(n-2)
delta = (betahl)/it
x = 0.5*delta
sum = 0.0d0
do i = 1,it
sum = sum + (1/sqrt(2.0d0*pi)*exp(-0.5d0*(x)**2))
x = x + delta
end do
prob = 0.5+sum*delta
else
betahl = -1*beta
it = 2**(n-2)
delta = (betahl)/it
x = 0.5*delta
sum = 0.0d0
do i = 1,it
x = x + delta
end do
prob = 1-(0.5+sum*delta)
end if
if(prob.lt.0) then
prob = 0.0d0
else
end if
return
end
A1.16
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: pdf.for
c December 2002
c Ver 1.0
c
c This subroutine computes the PDF of a given CDF
c
c INPUT VARIABLES :
c
c n : size of vector x
c x : vector of variable x
c cdfx : CDF of variable x
c
c
c pdfx : PDF of variable x
c
c**********************************************************************]
subroutine pdf(n,x,cdfx,pdfx)
integer i, n
double precision x(n), cdfx(n), pdfx(n), sum
sum = 0.0d0
pdfx(1) = 0.0d0
do i = 2,n
pdfx(i) = cdfx(i)-cdfx(i-1)
end do
return
end
A1.17
Annex 1
c**********************************************************************]
c
c MeBaSe
c
c SUBROUTINE: probdemand.for
c Coded: Luis F. Restrepo V. - based on the normalcdf.for routine
c November 2002
c Ver 1.0
c
c This subroutine computes the normal cdf of the demand for a given period
c
c INPUT VARIABLES :
c
c sddrs : Standard deviation of the demand
c drsint : Ordinates of the demand response spectra
c deltals : Vector of periods
c
c
c cdfdemand : Matrix with the CDF of demand given Tls
c
c**********************************************************************]
subroutine probdemand(sddrs,drsint,deltals,cdfdemand)
integer i, j, k, it, n, m
double precision sddrs, sddrs1, drsint(100), deltals(100)
double precision cdfdemand(100,100), sdy, my, funca, funcb
double precision beta1, beta2, prob, betahl, pi, delta, x, sum
pi = 3.14159265358979
n = 20
m = 6
do i = 1,100
prob = 0.0d0
sddrs1=drsint(i)*(1-sddrs)
sdy = log((sddrs1/drsint(i))**2+1)
my = drsint(i)*exp(-0.5*sdy)
do j = 1, 100
beta1 = (log(deltals(j)/my))/sqrt(sdy)
if(beta1.ge.0.0d0)then
if(j.eq.1) then
betahl = beta1
it = 2**(n-2)
delta = (betahl)/it
x = 0.5*delta
sum = 0.0d0
do k = 1,it
x = x + delta
end do
prob = 0.5+sum*delta
else
beta2 = (log(deltals(j-1)/my))/sdy
betahl = beta1 - beta2
it = 2**(m-2)
delta = betahl/it
x = beta2 + 0.5*delta
sum = 0.0d0
do k = 1,it
sum = sum+1/sqrt(2.0d0*pi)*exp(-0.5d0*(x)**2)
x = x + delta
end do
prob = prob+sum*delta
end if
else
if(j.eq.1) then
betahl = -1*beta1
it = 2**(n-2)
delta = betahl/it
x = 0.5*delta
sum = 0.0d0
do k = 1,it
x = x + delta
end do
prob = 1-(0.5+sum*delta)
else
beta2 = (log(deltals(j-1)/my))/sdy
betahl = beta1 - beta2
it = 2**(m-2)
delta = betahl/it
x = beta2 + 0.5*delta
sum = 0.0d0
A1.18
Annex 1
do k = 1,it
sum = sum+1/sqrt(2.0d0*pi)*exp(-0.5d0*(x)**2)
x = x + delta
end do
prob = prob+sum*delta
end if
end if
if(prob.lt.0) then
prob = 0.0d0
else
end if
cdfdemand(i,j) = prob
end do
end do
return
end
A1.19
Annex 1
Input file for In-plane example

0 5 6 2 20 100
0.012 0.0015
0.00
0.689
0.916
0.87
0.39
12.60 2.00 12.60
0.001 0.001 0.001
0.005 0.001 0.005
2.00 0.50 2.00
0.20 0.05 0.20
1.00 0.20 1.00
1.000 0.001 0.001 0.001 0.001 0.001
0.001 1.000 0.001 0.001 0.001 0.001
0.001 0.001 1.000 0.001 0.001 0.001
0.001 0.001 0.001 1.000 0.001 0.001
0.001 0.001 0.001 0.001 1.000 0.001
0.001 0.001 0.001 0.001 0.001 1.000
0.00 0.0000000
0.01 0.0000040
0.02 0.0000172
0.03 0.0000413
0.04 0.0000780
0.05 0.0001290
0.06 0.0001960
0.08 0.0003851
0.10 0.0006589
0.15 0.0014825
0.20 0.0026356
0.25 0.0041182
0.30 0.0059302
0.35 0.0080716
0.40 0.0105425
0.45 0.0167731
0.50 0.0186368
0.60 0.0223641
2.00 0.0745471
4.00 0.0745471
1.65
A1.20
Annex 1
Input file for Out-of-plane example
1 1 5 5 20 100
0.04 0.0250
0.00
0.40 0.10 0.40
3.00 0.50 3.00
0.80 0.20 0.80
0.50 0.20 0.50
0.30 0.05 0.30
1.000 0.800 0.800 0.800 0.800
0.800 1.000 0.800 0.800 0.800
0.800 0.800 1.000 0.800 0.800
0.800 0.800 0.800 1.000 0.800
0.800 0.800 0.800 0.800 1.000
0.001 0.00000
0.010 0.00003
0.020 0.00010
0.030 0.00024
0.040 0.00044
0.050 0.00070
0.060 0.00103
0.080 0.00192
0.100 0.00314
0.150 0.00791
0.300 0.04135
0.500 0.11990
0.700 0.16859
0.900 0.17456
1.000 0.17096
1.400 0.15115
2.000 0.13120
2.400 0.12326
3.000 0.11547
4.001 0.10796
1.65
A1.21
Annex 2
ISTRUZIONI PER LA COMPILAZIONE DELLA SCHEDA DI

RILIEVO DEGLI EDIFICI IN MURATURA
INDICE
pag.
INDICE ..... 2
1. NOTE ESPLICATIVE GENERALI ....... 3
2. SEZIONE 1: DATI RELATIVI ALLA SCHEDA .............................. 4
3. SEZIONE 2: LOCALIZZAZIONE EDIFICIO ... 4
4. SEZIONE 3: CONDIZIONI DUSO .. 5
5. SEZIONE 4: MORFOLOGIA DEL SITO .. 5
6. SEZIONE 5: ETA E STATO DI MANUTENZIONE .... 6
7. SEZIONE 6: TIPOLOGIA STRUTTURALE ..... 6
8. SEZIONE 7: CARATTERISTICHE GENERALI DELLEDIFICIO .... 8
9. SEZIONE 8: MECCANISMI DI RIBALTAMENTO ASSOCIATI A DANNO

ESISTENTI .........................................................................................................................
10
10. SEZIONE 9: SEZIONE DI CONTROLLO ............................................
12
A2.1
Annex 2
1. NOTE ESPLICATIVE GENERALI
1.1 ISTRUZIONI GENERALI
Per la compilazione della scheda dovr essere utilizzata sempre la matita in modo da poter facilmente
correggere eventuali errori.
Gli zeri vanno scritti barrati.
I dati numerici vanno riportati allineati con il margine di destra.
Il punto, nelle caselle in cui presente, separa la parte intera de quella decimale; ove non presente, i
numeri sono interi.
Le caselle definiscono il numero massimo di caratteri che si devono riportare. Qualora il valore
numerico superasse lo spazio massimo previsto, si inseriscono delle lettere sulla sinistra per le cifre
eccedenti (Z = 0, A = 1, B = 2, ...) escludendo per approssimazione le ultime cifre sulla destra. Ad
esempio, nelle caselle 488 a 491, si riporta:
B 3 4 1 Se la superficie media coperta di 23.414 mq
A C 2 8 Se la superficie media coperta di 132.751 mq
A C 2 7 Se la superficie media coperta di 132.749 mq
Le tabelle presente nella scheda verranno compilate rilevando ledificio a partire dai piani inferiori.
Nellimpossibilit di reperire un dato lasciare in bianco la casella o le caselle corrispondenti.
I manuali di istruzioni per la compilazione delle schede di I e II livelli, publicati per il CNR GNDT,
si intendono come complemento di queste manuale.
1.2 QUALIT DELLINFORMAZIONE
E Qualit elevata: Informazioni prevalentemente dirette (misure effettuate in sito, letture di

elaborati grafici affidabili, visione diretta degli elementi di informazione) con un grado di attendibilit
vicino alla certezza.
M Qualit media: Informazioni prevalentemente dedotte (letture indirette quali quelle desunte da
fotografie, misure desunte da elaborati non esecutivi, saggi non distruttivi di scarsa attendibilit,
letture dirette su situazioni analoghe, informazioni orali di persone di fiducia del rilevatore) con un
grado di attendibilit intermedio fra il precedente (E) ed il seguente (B).
B Qualit bassa: Informazioni prevalentemente presunte (misure dedotte da ragionevoli ipotesi

conoscitive quali quelle sulle usuali modalit e sulle pi frequenti scelte progettuali, informazioni orali
diverse dalle precedenti con un grado di attendibilit di poco superiore ad una scelta puramente
casuale della classe.
A Informazione assente: Con un grado di attendibilit intorno ai limiti di una scelta

casuale. In questi casi la valutazione del rilevatore ha valore puramente indicativo.
A2.2
Annex 2
2. SEZIONE 1: DATI RELATIVI ALLA SCHEDA
Data (caselle 1 a 6): Giorno, mese ed anno della data del rilievo.
Scheda # (caselle 7 a 10): Numero progressivo che identifica la scheda tra tutte quelle compilate da
una stessa squadra: tale numero viene confermato sulla scheda nel momento in cui questa viene
consegnata, compilata, dal caposquadra al coordinatore. chiaro che ogni squadra deve proporre
una sua numerazione progressiva per le corrispondenti schede compilate.
Rilevatore (caselle 11 a 26): Nome e cognome del capo della squadra di rilievo. Si inseriscono tutte
le lettere che entrano nelle caselle senza lasciare caselle bianche fra parole distinte.
3. SEZIONE 2: LOCALIZZAZIONE EDIFICIO
Comune (caselle 27 a 41): Nome del Comune per esteso. Si inseriscono tutte le lettere che entrano
nelle caselle senza lasciare caselle bianche fra parole distinte.
Foglio (caselle 42 a 44): Si riporta il numero del foglio della cartografia di rilevazione in cui si trova
ledificio.
Mappale (caselle 45 a 47): Si riportano, ove possibile, i dati relativi alledificio.
Particella (caselle 48 a 51): Si riportano, ove possibile, i dati relativi alledificio.
Provincia (caselle 52 a 53): Riportare il numero di codice ISTAT della Provincia.
Via (caselle 54 a 65): Riportare il nome della via, viale, ecc. tenendo presenti le seguenti convenzioni:
Si inseriscono tutte le lettere che entrano nelle caselle.

Non si lasciano spazi bianchi tra parole distinte.
Si omettono tutte le preposizioni e congiunzioni.
I numeri vanno sempre riportati in fondo in cifre arabe.
Per i nomi di persona si riporta liniziale del nome e del cognomem, ove questo presente; in caso
contrario, il nome per esteso.
I nome dei Santi, Santo, San, ecc. si abbreviano con Sseguita immediatamente dal nome.
Esempi:
Via della Battaglia di San Martino: BATTAGLIASMA

Piazza Giacomo Leopardi: GLEOPARDI
Corso Vittorio Emanuele II: VITTORIOEMA2
Piazza XXV Aprile: APRILE25
# Civico (caselle 66 a 68): Riportare il numero pi recente del portone principale (o presunto tale);
ove presenti (e ove possibile) riportare anche le lettere. Il portone di cui indicato il numero civico
deve essere sempre individuato nella cartografia di rilevazione con una freccia.
Altro # civico (caselle 69 a 71): Ove questo presente oppure ove un secondo portone o accesso
principale esiste, riportare il numero ed anche indicarlo nella cartografia di rilevazione con una freccia.
A2.3
Annex 2
Denominazione storica del bene (caselle 72 a 84): Riportare il nome oppure la denominazione
storica del palazzo; tenendo conto delle stesse convenzioni usati per la Via (caselle 54 a 65).
# Aggregato (caselle 85 a 87): Si indica in queste caselle il numero di aggregato strutturale

provvisorio assegnato dal coordinatore.
# Edificio (caselle 88 a 90): Numero dordine provvisorio delledificio. Queste caselle vengono
riempite sono nel caso in cui laggregato strutturale sia costituito da due o pi edifici. In questo caso il
numero dordine dellaggregato strutturale da mettere in scheda rimane costante, e gli edifici dello
stesso isolato vengono numerati progressivamente a partire da 1.
4. SEZIONE 3: CONDIZIONI DUSO
Uso piano terra (casella 91): Si riportano le condizioni duso del piano terra.
Uso piani superiori (casella 92): Si riportano le condizioni duso dei piani superiori. Se i piani
superiori hanno diversi condizioni duso si riporta quella prevalente per i piani superiori.
# Appartamenti (caselle 93 a 94): Si riporta il numero di appartamenti nelledificio per cui le

condizioni duso sono residenziale.
Attivit commerciale (caselle 95 a 109): Si riporta la attivit commerciale prevalente daccordo con
le seguenti categorie:
Produzione
Commercializzazione
Deposito, stoccaggio
Uffici
Misto, servizi pubblici
Locali amministrazione
Palestre
Condizioni duso (casella 110): Si intende riferirsi soltanto alle considerazioni temporali. Per
abbandonato si intende quando la mancata utilizzazione funzionale ha comportato anche uno stato di
degrado fisico.
5. SEZIONE 4: MORFOLOGIA DEL SITO
Morfologia del sito (casella 111): Si riporta la morfologia del sito.
Fondazioni in roccia (casella 112): Lopzione roccia dovr essere indicata in presenza di roccia
affiorante, anche se sovrastata da depositi superficiali parzialmente disgregati.
Pendenza percentuale del terreno (caselle 114 a 115): Si vuole rilevare la situazione altimetrica
media del terreno su cui insiste il fabbricato, valutata nella direzione ortogonale alla isoipse.
Posizione nellaggregato (casella 117): Si riporta la posizione deledificio nellaggregato.
A2.4
Annex 2
Inquadramento storico nellaggregato (casella 118): Si riporta linquadramento storico deledificio

nellaggregato.
6. SEZIONE 5: ET E STATO DI MANUTENZIONE
Et di costruzione (casella 119): Si riporta il codice che distingue lepoca della prima costruzione.
Nei casi di difficolt nellindividuare let della costruzione, si pu cercare di interpretare
induttivamente sulla base della tipologia e tecnologia edilizia (nei casi dubbi tra due classi contigue
scegliere la precedente). Evitare in ogni caso di omettere il dato.
Et ultimo intervento (casella 124): Si riporta il codice relativo allepoca dellultimo intervento pi
significativo. Se nelledificio non stato effettuato alcun intervento la colonna dovr essere sbarrata
Tipo ultimo intervento (casella 129): Si riporta il codice presente in tabella, relativo al tipo
dellultimo intervento pi significativo (si riporta 0 in caso di nessun intervento).
Stati di fatto (casella 130): Si tiene conto con questa voce dello stato di conservazione degli edifici,
secondo le seguente definizione:
P Pessimo: Edifici che presentano pareti fuori piombo e/o lesioni gravi anche se non diffuse.
Edifici caratterizati da grave deterioramento dei materiali. Edifici che, pur non presentando
lesioni, sono caratterizati da uno stato di conservazione delle murature tale da determinare una
grave diminuzione di resistenza.
S Scadente: Edifici con lesioni di media entit (ampiezza della lesione: 2-3 mm) o con lesioni
capillari di origine sismica. Edifici che, pur non presentando lesioni, sono caratterizati da uno
stato di conservazione delle murature tale da determinare una significativa diminuzione di
resistenza
D Discreto: Edifici che presentano lesioni capillari non diffuse, ad eccezione del caso in cui
queste siano state prodotte da terremoti.
B Buono: Murature in buone condizioni senza lesioni visibili.
Condizione dei giunti (casella 132): Si riporta il grado di deterioramento dei giunti tenendo conto la
seguente classificazione:
N Nuovo: Se la malta dei giunti in perfetto stato di conservazione, senza fessure e senza
nessun grado di deterioramento.
M Moderato: Se la malta esibisce fessure di qualsiasi lunghezza ma con una profondit minore
al 90% dello spessore della parete. In questa categoria possibile avere qualche grado di
deterioramento.
S Severo: Se la malta esibisce fessure di una profondit superiore al 90% dello spessore della
parete.
7. SEZIONE 6: TIPOLOGIA STRUTTURALE
Tipologia strutturale (caselle 134 a 145): Individua la tipologia dominante secondo le codifiche
indicate nella scheda. Riportare per ciascun elemento strutturale (verticali, orinzzontali e coperture) la
tipologia corrispondente ed il numero di piani per i quali si ripete. La scheda prevede la descrizione di
un massimo di quattro tipologie differenti. Qualora queste siano pi numerose occorre accorpare con
A2.5
Annex 2
gli stessi criteri usati per i dati metrici. La tipologia relativa alle coperture deve essere presente solo
nellultima riga riguardante lultimo piano, per cui in caso di coesistenza al penultimo piano di
parziale copertura e struttura orizzontale si indicher solo la tipologia di questultima, ponendo invece
sullultima riga la tipologia della copertura pi elevata (ovviamente solo se riguarda un piano preso in
considerazione e non una superfetazione irrilevante).
Di seguito si riportano, con qualche maggior dettaglio, le definizioni gi contenute nella scheda.
Strutture verticali
A : Muratura a sacco o a due paramenti formata da pietre di pezzature molto varie, male intessuta
e priva di collegamento tra i due fogli.
B : Muratura a sacco formata da pietre di pezzature pi regolare, bene intessuta e priva di
collegamento tra i due fogli, oppure come sopra con spigoli, mazzette e/o ricorsi in pietra
squadrata o mattoni pieni.
C : Muratura di pietra sbozzata di cattiva qualit in presenza di irregolarit.
D : Muratura di pietra sbozzata con spigoli, mazzette e/o ricorsi in mattoni pieni e/o pietra
squadrata.
E : Muratura di pietra arrotondata o ciottoli di fiume di pezzatura varia senza mazzette e/o ricorsi
in pietra squadrata e/o mattoni pieni.
F : Come sopra con spigoli, mazzette e/o ricorsi in pietra squadrata e/o mattoni pieni.
G : Muratura in blocchetti di tufo o pietra da taglio di dimensioni costanti.
H : Muratura in blocchetti di calcestruzzo prefabbicati, con inerti ordinari.
I : Muratura in blocchetti di calcestruzzo prefabbicati con inerti leggeri (argilla espansa, ecc.)
omogenei in tutta lestensione.
L : Muratura in laterizio di buona qualit, pieno o semipieno (% foratura 45%).
M : Muratura in laterizio con foratura > 45%.
N : Pareti in calcestruzzo non armato
O : Pareti in calcestruzzo armato e ben collegato; nel caso non sia possibile rilevare o avere notizie
sulla presenza di armatura, assegnare la tipologia N.
Strutture orizzontali
A : Solai con orditura principale e secondaria in legno.

B : Come sopra con catene o tiranti.
C : Solai con putrelle metalliche e voltine o tavelloni.
D : Come sopra con catene o tiranti.
E : Solai laterocementizi o solette in calcestruzzo con cordolo ammorsato o inserito nella
muratura.
F : Volte (a botte, a padiglione, ecc.) senza catene.
G : Come sopra con catene.
H : Combinazione di volte spingenti (tipo F) con solai piani.
I : Combinazione di volte non spingenti (tipo G) con solai piani.
L : Altra tipologia strutturale per le strutture orizontali.
Coperture
M : Coperture in legno spingenti per i soli carichi verticali, generalmente costituite da due falde
inclinate senza trave di colmo o con trave di colmo deformabile (rapporto luce/altezza l/h > 20) e
senza catene o tiranti (Fig. 7a, 7b).
N : Coperture in legno con travi inclinate ad una falda oppure a due falde e con muro di spina (Fig.
7c, 7d) oppure con trave di colmo poco deformabile (rapporto l/h < 20) (Fig 7e).
O : Coperture in legno a spinta eliminata (Fig. 7f) o con travi orizzontali (Fig. 7g).
A2.6
Annex 2
P : Coperture con solai laterocementizi o solette in calcestruzzo.

Q : Coperture con orditura principale in putrelle metalliche, spingente (analoga al tipo M per il
legno).
R : Coperture con orditura principale in putrelle metalliche, non spingente (analoga ai tipi N ed O
per il legno).
S : Combinazione dei tipi suddetti con almeno un tipo spingente (M, Q).
T : Combinazione dei tipi non spingenti suddetti (N, O, P, R).
U : Altra tipologia strutturale per le coperture.
l = luce trave
l/h > 20
Fig. 7a Fig. 7b Fig. 7c Fig. 7d
l = luce trave
l/h < 20
Fig. 7e Fig. 7f Fig. 7g
Cordoli di piano (caselle 149 a 156): Si riporta se il edificio presenta cordoli perimetrali in
calcestruzzo ed il numero di piani per i quali si ripete.
8. SEZIONE 7: CARATTERISTICHE GENERALI

DELLEDIFICIO
Regolarit pianta (caselle 159 a 166): Lassegnazione di un edificio alle varie classi aviene sulla
base della pi sfavorevole delle condizioni poste dai parametri B1 e B2 nel modo seguente:
B1 = a / l *100 B2 = b / l * 100
a a
a
b b
l l l
a a
b b
l l
A2.7
Annex 2
Orditura solai (caselle 169 a 176): Si riporta la direzione prevalente dellorditura del solai per ogni
piano.
Corpi aggiunti (caselle 179 a 186): Si intende riportare la esistenza di corpi precari aggiunti
successivamente, magari di altezza inferiore e male ammorsati alledificio originale.
Restringimenti (caselle 189 a 196): Si riportano eventuali restrizioni nellaltezza del edificio in ogni
piano.
Sopraelevazione precarie (caselle 199 a 206): Si riporta se esistono elementi aggiunti recentemente
(ultimi 50 anni), non ben integrati nel sistema strutturale.
Solai sfalsati (caselle 209 a 216): Si intende riportare quando il solaio di uno degli edifici adiacenti
nel terzo centrale dellinterpiano.
Demolizione elementi portanti (caselle 219 a 226): Si riportano eventuali interruzioni a piano terra
dei maschi murari in facciata o eliminazione di setti portanti allinterno (normalmente a beneficio di
esercizi commerciali).
Aggetti ed elementi svettanti (caselle 229 a 236): Si riportano elementi pericolanti come balconi in
pietra degradati, piccoli corpi abusivi, cornicioni particolarmente aggettanti, camini in cattivo stato di
manutenzione, ogni altro elemento ritenuto pericoloso.
Pianta schematica tipica: Si deve disegnare la pianta tipica delledificio schematicamente,

identificando con un indice le pareti. Non necessario disegnare ledificio in scala. Indicare il piano
asunto come piano tipico. Conviene trascurare gli elementi accessori delledificio, quali portici,
garage, ecc., che non si sviluppano per lintera altezza del corpo principale.
Pareti perimetrali: Per ogni parete perimetrale indivuata nella pianta schematica si deve riportare:
Indice parete (colonna 239): Si riporta lindice della parete secondo la pianta schematica.
Spessore t e rapporto Lm/Lo(colonne 240 a 255): Si riporta lo spessore de la parete t ed il
rapporto tra la lunghezza totale dei maschi Lm ed la lunghezza totale della parete Lo in ogni
livello.
Direzione (colonna 256): Si riporta X o Y della parete tenendo conto se la parete parallela
oppure perpendicolare alla facciata.
Lunghezza Lo (colonne 257 a 259): Si riporta la lunghezza totale della parete.
Piani edificio contiguo (colonna 260): Si riporta il numero minimo di piani degli edifici
contigui.
# piani a pareti uguali (colonna 261): Si riporta il numero di piani per cui la pianta uguale.
Distribuzione regolare di catene (colonna 262): Si riporta se le catene sono regolarmente
distribuite nella lunghezza della parete.
Numero di pareti interne collegate (colonna 263): Si riporta quanti pareti interne sono
collegate alla parete perimetrale.
Buon collegamento con pareti laterali (colonna 264): Si riporta se c un buon
collegamento tra la parete perimetrale e le parete laterale in modo tal che sia possibile
ipotizzare un comportamento unitario tra le pareti collegate.
Numero di aperture allineate verticalmente con distanza L1 < L2 (colonna 265): Si
riporta il numero di colonne di aperture allineate verticalmente, per cui la distanza L1 < L2
secondo il seguente schema.
Distanza Lvar (colonne 266 a 268): Si riporta la distanza Lvar secondo il seguente schema:
A2.8
Annex 2
Lvar Lvar Lvar
L1 L2 L1 L2
Numero di aperture allineate Numero di aperture allineate Numero di aperture allineate

verticalmente con distanza verticalmente con distanza verticalmente con distanza
L1 < L2 = 0 L1 < L2 = 0 L1 < L2 = 3
Superficie media coperta (caselle 488 a 507): Si definisce superficie media coperta larea coperta al
lordo delle pareti dai seguenti elementi strutturali: solai, coperture, volte, porticati, scale esterne.
Altezza media interpiano (caselle 510 a 521): La misura deve essere comprensiva di met spessore
dei due solai nel caso della altezza media interpiano.
Altezza aperture (caselle 524 a 535): Si riporta la altezza media delle aperture nel piano (finestre,
porte, ecc.).
Area totale degli elementi resistenti (caselle 538 a 565): Larea degli elementi resistenti misurata
tra gli interassi dei muri ortogonali di intersezione. Larea degli elementi inclinati di un angolo
rispetto alla direzione considerata va moltiplicata per Cos2.
9. SEZIONE 8: MECCANISMI DI RIBALTAMENTO

ASSOCIATI A DANNI ESISTENTI
Meccanismo (caselle 569 a 576): Si reporta il meccanismo di ribaltamento associato al danno

esistente, tenendo conto dei dieci meccanismi in Fig. 9.
Lunghezza (caselle 577 a 588): Si riporta la lunghezza L del meccanismo associato tenendo conto
della Fig. 9.
Altezza (caselle 589 a 600): Si riporta la altezza h del meccanismo associato tenendo conto della Fig.
9.
Spessore (caselle 601 a 608): Si riporta lo spessore t del meccanismo associato tenendo conto la
Fig. 9.
A2.9
Annex 2
t t
h h
1 L
2 L
2/3h
h
h
L
t
3 4 L
t
t
h
h
5 L
6 L
t t
L
h
L
h
7 8
A2.10
Annex 2
t
t
h h
9 10
Fig. 9
10. SEZIONE 9: SEZIONE DI CONTROLLO
Scheda controllata (caselle 609 a 614): Si riporta la data in cui la scheda stata controllata e
verificata la non mancanza de nessuna informazione fondamentale. Si riporta anche il nome e
cognome della persona che lha controllata.
Scheda nella database (caselle 615 a 620): Si riporta la data in cui la scheda stata inserita nella
database ed anche il nome e cognome della persona che lha inserita.
A2.11
Annex 2
A2.12
Annex 2
A2.13
Annex 2
A2.14

A Simplifies Mechanics-Based Procedure For The Seismic Risk Assessment of Unreinforced Masonry Buildings

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

A Simplifies Mechanics-Based Procedure For The Seismic Risk Assessment of Unreinforced Masonry Buildings

Diunggah oleh

Hak Cipta:

Format Tersedia

I.U.S.S.

EUROPEAN SCHOOL OF ADVANCED STUDIES IN

A SIMPLIFIED MECHANICS BASED PROCEDURE

An Individual Study Submitted in Partial

LUIS FERNANDO RESTREPO VELEZ

Supervisor: Prof. GUIDO MAGENES

Paolo Emilio Pinto

A SIMPLIFIED MECHANICS BASED PROCEDURE

LIST OF FIGURES .. vii

2. VULNERABILITY AND LOSS ESTIMATION PROCEDURES .. 3

2.1 DAMAGE PROBABILITY MATRIX ..... 4

2.2 GNDT I AND II LEVEL APPROACHES .... 6

2.3 VULNUS ... 8

2.5 FaMIVE .... 14

2.6 SIMPLIFIED DISPLACEMENT BASED PROCEDURE FOR CLASESS OF

2.7 GENERAL REMARKS .... 18

3. MeBaSe PROCEDURE .... 19

3.1 PROCEDURE FOR IN-PLANE FAILURE MECHANISM .................................... 19

3.1.1 Demand ................................................................. 19

3.2 PROCEDURE FOR OUT-OF-PLANE FAILURE MECHANISM . 29

3.2.1 Demand ................................................................. 30

3.3 PROBABILISTIC CONSIDERATIONS ...... 36

3.3.1 Uncertainties ......................................................... 37

3.4 DAMAGE PROBABILITY AT URBAN OR REGIONAL SCALE 50

4.1 PROBABILITY OF FAILURE FOR IN-PLANE RESPONSE OF A 4-STORIES

4.2 PROBABILITY OF FAILURE FOR OUT-OF-PLANE RESPONSE OF SINGLE

5. SURVEY DATA FORM 63

6.1 SUMMARY ....................................................... 68

6.2 FINAL REMARKS ........................................... 70

6.3 FUTURE RESEARCH .. 70

ANNEX 1. FORTRAN Code for the MeBaSe Procedure

ANNEX 2. Survey data form

2.1 Classification of vulnerability assessment methods, according to input, methodology and

2.1 Differentiation of structures (buildings) into vulnerability classes [Gruntal, 1998].

4.3 Cumulative density functions CDF for demand FD.

Additionally, large portions of buildings in non-developed countries, and an important portion of

2. VULNERABILITY AND LOSS ESTIMATION

Conventional: these correspond to methodologies based on judgements given by experts. The

Purpose and regional scale of the damage scenario.

Accuracy of the earthquake scenario used for the damage scenario.

Existing information, either from government bodies, databases, libraries, etc.

2.1 DAMAGE PROBABILITY MATRIX

DPM (DV , I , T ) = P(DV | I , T ) (2.1)

Table 2.2. European Macroseismic Scale EMS [Gruntal, 1998]

2.2 GNDT I AND II LEVEL APPROACHES

Table 2.4. Scores and relative weights to compute Iv

{Vx ,V y } = {Fx , Fy } f t 1 + ( W ) (2.4)

Figure 2.3. Definition of the safety criterion [Bernardini, 2000]

Table 2.5. Classification and corresponding values of the vulnerability factors

Table 2.6. Vulnerability factors related to qualitative judgements and their

state and is the standard normal cumulative density function.

ds = (CONV [ C , D ])2 + ( T ,ds )2 (2.8)

Figure 2.5. Example of fragility curves for slight, moderate, extensive

Figure 2.6. Example of building damage estimation process [FEMA, 1999]

2.6 SIMPLIFIED DISPLACEMENT BASED PROCEDURE FOR CLASSES OF

In general, the steps suggested in the procedure are:

Table 2.8. Limit states for masonry structures, Calvi [1999]

LS = 0.67hs n y + ( LS y )hs (2.10)

2.7 GENERAL REMARKS

To consider the uncertainty on the demand, recognized to be one of the components of

Conversely, the problems identified to be overcome are:

Complexity in the application of the procedure in real cases.

3.1 PROCEDURE FOR IN-PLANE FAILURE MECHANISM