Vous êtes sur la page 1sur 15

Engineering Structures 28 (2006) 357371

www.elsevier.com/locate/engstruct

Vulnerability assessment and earthquake damage scenarios of the building


stock of Potenza (Southern Italy) using Italian and Greek methodologies
Mauro Dolce a , Andreas Kappos b, , Angelo Masi a , Gregory Penelis b , Marco Vona a
a Department of Structures, Geotechnics and Geology Applied to Engineering, University of Basilicata, Via dellAteneo Lucano, 85100 - Potenza, Italy
b Department of Civil Engineering, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece

Received 8 November 2004; received in revised form 5 August 2005; accepted 8 August 2005
Available online 5 October 2005

Abstract

The prevailing Italian and Greek methodologies for seismic risk assessment are used herein to construct loss scenarios for the building stock
of a small city (Potenza, Southern Italy). The inventory of buildings of interest is obtained from a survey carried out after the 1990 earthquake that
struck Potenza and its hinterland, subsequently updated in 1999. About 12,000 buildings were surveyed in Potenza, using the Italian first level
survey form for damage and vulnerability evaluation. In the Italian methodology, a hybrid technique is set up to evaluate vulnerability, combining
an analysis of building typologies with expert judgement. The probabilistic distribution of damage is evaluated by assigning Damage Probability
Matrices (DPMs) from the literature. Besides the vulnerability classes A, B and C of the MSK-scale, the class D of the anti-seismic buildings
is considered and the relevant DPM is defined. Damage and economic loss scenarios relevant to dwelling buildings are constructed for three
reference earthquakes. Next, the hybrid methodology for seismic vulnerability assessment of reinforced concrete (R/C) and masonry buildings
developed at the University of Thessaloniki (Greece) is applied to the same building stock. The methodology combines available statistical data of
damage collected after past earthquakes with a systematic nonlinear analysis of various model buildings, representative of several vulnerability
classes. Similarities, as well as discrepancies, between the two methods are discussed in the light of the obtained results, and possible sources for
the discrepancies are suggested.
c 2005 Elsevier Ltd. All rights reserved.

Keywords: Vulnerability; Seismic risk; Assessment; Damage scenario; Nonlinear analysis; Buildings; Concrete; Masonry

1. Introduction provide a complete picture of what happens when such an


earthquake occurs. Combined with GIS technology they are
The last decade or so has witnessed a growing interest in also powerful tools to check and visualise the effects of any
assessing the seismic vulnerability of European cities and the risk mitigation strategy.
associated risk; not surprisingly this interest was stronger in Earthquake damage scenarios can be referred to different
Southern Europe where the largest part of the seismic activity kinds of damage and losses. Consequently, preparing a
in this continent takes place. As a result, a fair number of scenario requires contributions from a wide range of topics
earthquake damage (loss) scenario studies appeared over this and disciplines, spanning from Seismology and Geology to
period [4,5,10,14,15,19,25] wherein some of the most advanced Structural and Geotechnical Engineering, from Urban Planning
techniques have been applied to a number of European cities (or and Transport Engineering to Social and Economical Sciences.
parts thereof). Whereas risk analysis refers to all the possible However, it often happens that each specialist interacts very
arriving earthquakes, estimating the probability of losses over little (if at all) with the other specialists. Furthermore, each
a specified period of time, scenario studies refer to a given country has its own specifics for the numerous aspects involved
earthquake (maximum credible, standard design, frequent) and by seismic problems, such as: characteristics of structures,
seismological features of the territories, socio-economical
Corresponding author. Tel.: +30 31 995 743; fax: +30 31 995 614. conditions, etc. Even in Europe, where common cultural
E-mail addresses: dm142ing@smtp.unibas.it (M. Dolce), roots and similar characteristics of structures could make the
ajkap@civil.auth.gr (A. Kappos). exchanges easier and very fruitful, it is difficult to find a

0141-0296/$ - see front matter 


c 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2005.08.009
358 M. Dolce et al. / Engineering Structures 28 (2006) 357371

common language among researchers from different countries, villages was carried out by local engineers and architects,
to exchange data and, finally, to set up common approaches to under the co-ordination of the Regione Basilicata, with the
risk mitigation. co-operation of the Civil Protection Department and the
To this purpose, earthquake damage scenarios relevant to the Italian National Group for the Protection against Earthquakes
case study of Potenza, a small city in Southern Italy, have been (GNDT). 21 villages, where intensity was estimated between
prepared using different methodologies that can be considered VI and VII (MCS) in the post-event macroseismic survey, were
as the prevailing ones utilised in Italy and in Greece. The main fully surveyed. The surveyors used the first level GNDT90
objectives of this study, part of which was developed in the inspection form,2 for damage and vulnerability evaluation.
framework of the EU-funded ENSeRVES Project [16], are: About 50,000 buildings were surveyed, 12,000 of which were in
to compare vulnerability and damage assessment procedures Potenza. In addition to damage data, geometrical and qualitative
for residential buildings in the two countries; characteristics were collected, such as height, configuration
in plan and elevation, age, type of vertical and horizontal
to identify similarities and differences in databases on
structures, type of foundation and of roof, retrofitting (if any),
damage from past earthquakes;
state of preservation, etc. [14].
to suggest possible improvements in vulnerability assess-
ment procedures for buildings through the integration of dif- In 1999 that inventory was updated with data relevant to
ferent approaches (statistical, mechanical, expert-heuristic, R/C buildings built after 1990. The update was carried out
etc.). by the DiSGG of the University of Basilicata, on the basis
of the technical documentation provided by the Municipality
It is pointed out here that in the remaining of this paper of Potenza. In this case too, the data were collected using
reference to the Italian and Greek methodologies for the first level GNDT90 inspection form. About 300 buildings
vulnerability assessment, should by no means be construed were surveyed, with about 1500,000 m3 total volume. The
as implying that these are the only methods used in either smallest R/C structures, mainly situated in rural zones, were
country; other methods have also been suggested and used, not considered.
particularly in Italy. Having said this, the methodology based More details on the characteristics of the building stock
on damage probability matrices is by far the most commonly obtained from the two surveys can be found in Dolce et al. [14,
used in scenario studies, both in Southern Europe and in most 17], where the database obtained merging the two surveys has
of the remaining world, while the particular DPMs used herein been examined to obtain a complete description of the Potenza
are certainly the ones predominantly used in either country. building stock.
Earthquake damage scenarios relevant only to building The property status, as well the use, of buildings is mostly
damage and the associated economic losses are presented in this
private (about 95%), only a small percentage of buildings being
paper. As is typical in the preparation of earthquake damage
public property. Surprisingly high percentage (about 50%) of
scenarios [13], they are based on information regarding:
buildings used for production activities, as classified in the 1990
(i) the characteristics of ground shaking in the area under
survey, was noted [14]. What really happens, in many cases,
consideration, i.e. the consideration of a maximum probable or
is that they are very small structures separated from the main
maximum credible earthquake based on the results of seismic
building and used as agricultural warehouses or garages.
hazard studies; (ii) the description of the seismic vulnerability
In this paper, only the largest (private and public) buildings
of the buildings of interest, i.e. the probability of damage
are considered, leaving out the smallest rural masonry
to given building types for a given ground shaking; (iii)
structures, thus resulting in an inventory of about 9000
the inventory of the buildings of interest, i.e. location and
buildings.
characteristics of buildings.
The composition of the building stock is very different,
2. Typological analysis of the building stock depending on whether the number or the volume of the
buildings is considered. In terms of number of buildings, the
The preparation of the earthquake damage scenarios is based sample is mostly made of masonry (62%) rather than R/C
on a comprehensive inventory of buildings, obtained from a structures (36%). On the contrary, in terms of volume there is a
survey carried out after the 1990 earthquake that struck Potenza strong prevalence of R/C (66%) over masonry structures (33%).
and its hinterland, whose local magnitude was Ml = 5.2 [3]. It has to be noted that a very limited number of steel and mixed
The maximum intensity felt in nearby villages was VII on (masonry and R/C) buildings is present in the building stock
the MCS scale.1 Following the earthquake, a survey in 41 of Potenza. With regard to the structural type, most masonry
buildings have vertical load bearing elements made up of not
hewn (rubble stone) masonry, called sacco, whereas most
1 The MCS scale is a 12-point scale still in use in Italy, mainly by R/C buildings are frame structures.
seismologists. Set up by Mercalli, Cancani and Sieberg in 1912, it is rather
different from the MM and EMS scales. The following relationship can be
established among MCS and EMS (or MSK) values [27]: 2 The first level GNDT90 inspection form is a vulnerability and damage
IMCS = 1.17 IEMS 0.76 survey tool widely used in Italy. The form contains data on building
identification, dimensions, age, conditions, vertical and horizontal structural
hence, for example VII on the EMS is equivalent to VIIVIII on the MCS. type and roof type, damage to structural and non-structural elements.
M. Dolce et al. / Engineering Structures 28 (2006) 357371 359

Considering the distribution of buildings according to their


height, more than 65% of the masonry buildings have less than
three stories (typical in Southern Italy, as well as in Greece).
On the contrary, R/C buildings are almost uniformly distributed
between one and nine stories, with just a peak for the three-
storey category, which accounts for about 30% of the entire R/C
group.
As regards the distribution of buildings according to the
period of construction, old masonry buildings (pre-60s70s)
prevail over new ones, while post-war R/C buildings account
for almost 100% of the R/C population. It has to be noted that,
because after the 1980 Irpinia Earthquake the area of Potenza
was classified as a seismic zone, all buildings designed after
1980 should be earthquake resistant. Furthermore, after the
1980 earthquake many buildings (mostly masonry ones) have Fig. 1. Distribution of ERD levels in terms of number of buildings.
been retrofitted.
Some remarks can be made by considering masonry and
R/C structures separately, and by categorising them according
to the period of construction, the retrofit (if any), and the
position of buildings in urban or rural zone. Masonry buildings
are mainly present in rural zones (about 75% of the total of
masonry buildings), whereas R/C buildings are more numerous
in urban zones, with a significant presence also in the historical
centre. This distribution of building types in the territory of
Potenza is mainly due to political decisions on the town-
planning development, taken after the Second World War. In
urban zones, new buildings, often public housing, were always
constructed in R/C. Moreover, between 1945 and 1970, many
old masonry buildings in the historical centre were demolished
and replaced by new R/C buildings. On the contrary, in rural
Fig. 2. Distribution of ERD levels in terms of volume of buildings.
zones, masonry was widely used until the 80s.
Finally, a relatively low percentage of earthquake-resistant
data, thus preventing the use of a direct typological technique
(retrofitted and/or post-1980) buildings are present in the
for vulnerability evaluation. For this reason, the vulnerability
Potenza stock. In terms of number of buildings, seismic
evaluation is made by using the Damage Probability Matrices
resistant buildings are only 35% of the total (27% for masonry,
(DPMs) set up by Braga, Dolce and Liberatore after the
54% for R/C). In terms of volume of buildings, seismic resistant
1980 Southern Italy earthquake [7,8]. Besides the vulnerability
buildings are 44% of the total. However, it should be noted
classes A, B and C of the MSK-scale, class D of the
that the actual situation is somewhat better, because the data
EMS98 scale, pertinent to earthquake-resistant buildings, has
regarding the retrofit interventions go back to 1990, whereas
been considered and the relevant DPM has been defined.
in the 90s other buildings, mostly masonry structures, were
Descriptions of vulnerability classes A, B, C and D are provided
retrofitted.
The distribution of the earthquake resistant design (ERD) in [6,17,18]. The general idea is that class A corresponds to the
levels in the building stock is shown in Figs. 1 and 2 (in terms highest vulnerability (e.g. rubble stone or adobe buildings), and
of number and volume of buildings, respectively). D to low vulnerability (e.g. R/C frame or wall systems, with
moderate level of seismic design); EMS also includes classes
3. Earthquake damage scenarios based on the Italian E and F, corresponding to modern R/C or steel earthquake
methodology resistant structures.
Based on an accurate typological analysis of the building
The Italian methodology [16] consists of four main steps: (i) stock, each building is assigned to one of the four considered
analysis of building typology, (ii) recognition of vulnerability vulnerability classes, taking into account the vertical and
classes, (iii) selection of reference earthquakes, (iv) preparation horizontal structural types, the eventual retrofitting, and the
of earthquake damage scenarios. period of construction (before or after the introduction of
The characteristics of the entire building stock have been seismic design in the area).
analysed in previous works by Dolce et al. [14,15,17], where With regard to the selection of reference earthquakes, three
the high number of different building types was highlighted. deterministic events have been considered, having intensity
The analyses showed poor correlation between structural types IMSK equal to VI, VII and VIII. They correspond to events
and damage, mainly because of the low reliability of damage expected in Potenza having return period ranging from 50 yr
360 M. Dolce et al. / Engineering Structures 28 (2006) 357371

Fig. 3. Distribution of classes of vulnerability of building stock.

(probable event) to over 475 yr (strong event), as given in The statistical distributions of the vulnerability classes of all
the New seismic hazard maps of the Italian territory [1]. buildings are shown in Fig. 3. The building stock exhibits a low
Finally, the preparation of earthquake damage scenarios overall vulnerability, as classes C and D account for about 69%
relevant to the three reference intensities is made, both in terms in terms of number of buildings and about 87% in terms of
of damage and of economic losses. The influence of site effects, volume.
discussed in detail elsewhere by Dolce et al. [15,17], is not Comparisons between the damage distributions caused by
considered herein. the three selected earthquake intensities are shown in Figs. 4
and 5 (in terms of number of buildings and volume of
3.1. Seismic vulnerability assessment and damage scenarios buildings, respectively). Damage level is defined according to
The definition of vulnerability classes has been related to the specifications of the EMS98 scale, wherein five levels of
the Damage Probability Matrices (DPMs) set up after the damage, beyond the null (zero) damage, are assumed (Table 1).
1980 Southern Italy earthquake [7,8]. However, the significant Fig. 4 shows that, for the seismic event with 50 yr return
evolution of the building stock in the last 20 years, mainly period (IMSK = VI) a few buildings, about 11%, have damage
due to the post-1980 seismic classification, had to be taken level greater than 1. For the seismic event with IMSK = VII
into account. For this reason, a further class with lower the number of partially or totally collapsed buildings (L d 4)
vulnerability (class D), relevant to earthquake-resistant or is 1.9%, while it increases up to 12.4%, when considering
retrofitted buildings, was derived from the above-mentioned IMSK = VIII.
DPMs accounting for the EMS98 specifications. More details By considering the damage distribution in terms of building
of this procedure are reported by Dolce et al. [17]. volume (Fig. 5), further reductions in the percentages relevant
After setting up the DPMs, the next step has been the to the higher damage levels can be observed. This is a
assignment of a building type to each building in the stock consequence of the prevalence of R/C buildings (assigned
under consideration. By combining the vertical and horizontal to class C or D) when considering the distribution in terms
structural types, each building was classified into one of the of volume: R/C buildings have average volume larger than
four vulnerability classes (A to D). When a variation of the masonry buildings (3650 m3 as opposed to 670 m3 ).
structural characteristics along the height was observed, the To provide a global evaluation of the damage due to a given
most vulnerable vertical and/or horizontal types have been the mean damage index DImed [17] is evaluated as
intensity, 
considered. With regard to age and the eventual retrofitting, DImed = i (L d,i fi )/n, where L d,i is a generic damage level
both seismic-resistant (after 1980) and retrofitted buildings (L d,i = 15) and fi is the relevant frequency. The summation
(both R/C and masonry buildings) were assigned to the is carried out for n = 5, i.e. zero damage level is not included.
lower class of vulnerability. Class C has been assigned to DImed varies between 0 and 1, where DImed = 0 means total
R/C buildings designed and constructed before 1980. These absence of damage and DImed = 1 means total destruction. The
assumptions can underestimate, in some cases, the actual values of DImed , for the three considered seismic intensities, are
vulnerability of retrofitted masonry buildings (assigned to class reported in Table 2.
D), particularly when they had high vulnerability (A) before 3.2. Economic loss scenarios
the retrofitting. Moreover, as regards pre-1980 R/C buildings
(assigned to class C), recent studies by Masi [28] show that In Masi et al. [29] a methodology for the evaluation of eco-
their vulnerability class could be in the range between B and D, nomic losses due to damage in the building stock is presented.
depending on some global or local characteristics (e.g. infilled This methodology is firstly based on an estimation of the re-
or not infilled frames, presence of a soft story, deformed or pair cost conditional upon the suffered damage level and the
smooth reinforcement bars). Further work to better evaluate the building type [11]. The estimation of the repair cost was carried
actual vulnerability of these buildings is needed and is currently out using the data collected with the GNDT-SSN usability sur-
in progress. vey form after the Umbria-Marche 1997 and the Pollino 1998
M. Dolce et al. / Engineering Structures 28 (2006) 357371 361

Fig. 4. Damage distribution relevant to the reference earthquakes (in terms of number of buildings).

Fig. 5. Damage distribution relevant to the reference earthquakes (in terms of volume of buildings).

Table 1
Definition of damage levels

Damage level L d Definition Notes


0 No damage
1 Slight damage No SD, slight N-SD
2 Moderate damage Slight SD, moderate N-SD
3 Substantial to heavy damage Moderate SD, heavy N-SD
4 Very heavy damage Heavy SD, very heavy N-SD
5 Destruction Very heavy SD
SD = Structural Damage; N-SD = Non-Structural Damage.

earthquakes, thus obtaining a database with more than 50,000 Table 2


buildings. Based on the information drawn from this form, typi- Values of the mean damage index relevant to the reference earthquakes
cal repair interventions were selected, their extent and cost were IMSK = VI VII VIII
computed, and the global repair costs were evaluated. DImed (number of buildings) 0.10 0.18 0.35
In Di Pasquale and Goretti [12] curves of repair costs as a DImed (volume) 0.07 0.14 0.29
function of damage level and building vulnerability class are
proposed; an economic damage index Cr,r (relative repair cost)
was evaluated, equal to the ratio of cost of repair to cost of is dependent only on damage level, whereas the dependency
replacement of the building. These curves indicate that Cr,r on the building vulnerability class, i.e. on the type of the
362 M. Dolce et al. / Engineering Structures 28 (2006) 357371

Table 3
Statistical values of the relative repair cost relevant to the damage levels

Ld 0 1 2 3 4 5
0.005 0.035 0.145 0.305 0.800 0.950
0.035 0.043 0.056 0.111 0.113 0.060
q 0.015 0.604 5.587 4.942 9.224 11.585
r 3.046 16.662 32.946 11.262 2.306 0.610

Fig. 8. Values of the total relative repair costs of the entire building stock
(T Cr,r ) relevant to the reference earthquakes.

the probabilities provided in Fig. 7, the total probability for


each considered value of repair cost Cr,r > 0, given a seismic
intensity I , can be obtained as follows:
Prob[Cr,r | I ]
5  D
= Prob L d [Cr,r | L d ] ProbV [L d | V, I ] (1)
Fig. 6. Standard beta density function PDF of the relative repair cost. L d =1 V =A

where the five L d values larger than 0 (15) and the four
vulnerability classes V (A, B, C, D) have been considered.
The probability associated with Cr,r = 0, can be computed as
follows:
Prob[Cr,r = 0 | I ] = ProbV [L d = 0 | V, I ]
5  D
+ ProbL d [Cr,r | L d ] ProbV [L d | V, I ]. (2)
L d =1 V =A

Eqs. (1) and (2) take into account that Cr,r is not dependent
on the vulnerability class of the building.
If an inventory of buildings is available, the total repair cost
T Cr,r caused by a given seismic event, as a fraction of the
total cost of replacement of the entire building stock, can be
computed.
Fig. 7. Cumulative density function CDF of the relative repair cost.
This procedure has been applied to the building stock of
structural system, can be neglected. As Cr,r is a random variable Potenza, thus obtaining the T Cr,r values reported in Fig. 8,
whose values are bounded between 0 and 1, the curves of the where T Cr,r is computed considering the distribution of the
relative repair cost against the damage level (referred to the vulnerability classes both in terms of building number (black
mean damage of the vertical structures) can be described by bars) and of building area (grey bars). It is worth noting that,
a standard beta distribution. The mean values and standard in this case, reference to the area of buildings instead of their
deviations of the distributions [21] are reported in Table 3, volume has been made, as the costs (replacement, repair) are
together with the relevant values of the parameters q and r (of typically provided in terms of unit area.
the beta-distribution) obtained in closed form in [29]. When the computations are based on the number of
Figs. 6 and 7 show, respectively, the standard beta buildings, the total value of the relative repair cost T Cr,r ranges
probability density function (PDF) and the cumulative density from 2.7% (IMSK = VI) to 18.1% (IMSK = VIII) of the
function (CDF) of the relative repair cost, relevant to the replacement cost of the entire building stock. On the other hand,
damage levels L d = 15. It has to be noted that each value of more correct values of T Cr,r can be computed considering the
Cr,r can result from more values of the damage level L d (see for building area, equal to 1.8%, 4.9% and 14.3%, for IMSK =
example the vertical continuous line in Fig. 7 for Cr,r = 0.1). VI, VII and VIII, respectively. Due to the characteristics of
On the other hand, the DPMs give the probability to observe the Potenza building stock, where a high percentage of small
the different damage levels L d for each vulnerability class V , masonry buildings is present, T Cr,r values are overestimated
given a seismic intensity I . Connecting these probabilities to when computed according to the number of buildings.
M. Dolce et al. / Engineering Structures 28 (2006) 357371 363

If the values of the total repair costs T Cr in Me (million purely empirical approach and statistical data from three Greek
of euro) are considered, assuming an average replacement cost earthquakes, and some first comparisons with Italian DPMs
equal to e700/m2 (approx. $840/m2) and a total area of the were made.
Potenza building stock equal to about 4500,000 m2 , they are
found to be rather low, thus confirming the low vulnerability of 4.1.1. Method based on time-history analysis
the Potenza building stock. Considering the building area, the The basic steps in the hybrid methodology for deriving
maximum computed value of T Cr is 458 Me (approx. 550 M$) damage probability matrices (DPMs) based on statistical
equal to 0.035% of the Italian Gross Domestic Product (GDP). data and inelastic time-history analysis of appropriate models
Computing the GDP only for Potenza (which has about 70,000 [22,26] are the following:
inhabitants), T Cr is equal to 35% of the local GDP, i.e. about
Construct the columns (corresponding to specific earthquake
two times the annual total budget of Potenza municipality.
intensities) of each DPM for which statistical data from
past earthquakes are available, using the standard empirical
4. Earthquake damage scenarios based on the Greek
procedure. The direct use of data in terms of repair to
methodology
replacement cost ratios is recommended (for loss assessment),
while quantitative descriptions of the physical damage states,
4.1. Overview of the methodology
with subsequent correlation with cost, is also an option. Both
procedures inherently involve large scatter, as also discussed in
The Greek version of the hybrid approach to seismic
the following.
vulnerability has been developed by Kappos et al. [24,26]
in recognition of the fact that reliable statistical data for Construct the remaining parts of the DPM on the basis of the
seismic damage are quite limited and typically correspond results of inelastic time-history analysis of models simulating
to a very small number of intensities. Within a number of as closely as practicable the behaviour of each building class,
risk studies, a recent one being the Volos project [25], it was and subjected to input motions that have been derived for
realised that damage data collected from Greek earthquakes the site under consideration, also taking local soil conditions
(Kalamata 1986, Pirgos 1993, Patras 1993, Aegion 1995), into account. As dynamic analysis typically provides structural
although valuable, are generally not in a form such that response quantities (such as ductility factors, displacements
economic damage statistics can be assessed for a representative etc.) appropriate models for correlating structural parameters
set of buildings. What usually happens is that the collected to direct loss (expressed in terms of cost of repair) are required
data concerns only buildings that have been inspected for at the post-processing stage. Such models were proposed by
a second time (i.e. after the initial post-earthquake rapid Kappos et al. [26] for correlating the economic damage index
screening) and/or wherein some post-earthquake intervention for R/C members to the ductility demand in each critical region,
(repair, strengthening, etc.) has taken place, while the extent and the economic damage index for masonry infill panels to the
of the geographical area (number of municipalities), hence interstorey drift.
the total building stock to which the data refers, is often If intensities higher than 10 have to be considered, expert
unclear. Moreover, even the good quality data for Thessaloniki judgement should be used to construct the corresponding
correspond to a single intensity (MSK) estimated at about 7 columns of the DPM, as empirical data are extremely scarce,
(estimates for the considered area varied from 6.5 to 7.0). while analytical data tend to be unreliable. It is noted that these
It was then clear that either data from abroad should be intensities are typically of little practical use in developing loss
imported, or the available data should be augmented using scenarios; they are actually ignored in the Italian DPMs [17].
either expert judgement (see ATC 1985 [2]) or an analytical The selection of input motions for carrying out the time-
(mechanical) approach. The Greek hybrid methodology history analysis has to be done in a meaningful way, consistent
was based on a combination of the empirical/statistical and the with the overall approach used. If the DPMs are to be
analytical/mechanical approach. developed in terms of intensity (traditional approach), the
The basics of the mechanical approach to vulnerability input motions used for each set of analyses have to correspond
assessment and the procedure to apply it for deriving to a particular intensity. Clearly the options are numerous and,
damage probability matrices is described in [23,25]. In these in practical terms, the choice largely depends on the scope of
publications the emphasis is on methodological aspects (such the project. Whenever the problem is tackled in the framework
as the crucial model correlating structural to economic damage of microzonation and/or scenario studies, typically a scenario
index), while application of the method to deriving damage earthquake input motion at bedrock is first estimated and then
and loss scenarios is of essentially pilot character. Moreover, it is filtered through the various soil layers in each zone of the
the analytical component of the method includes inelastic area studied, to arrive at surface motions that can be used for
time-history analysis of the response to a number of input the time-history analysis. In practice the analyst (dealing with
motions, and is also restricted to R/C buildings. An expanded the vulnerability analysis) usually has at his/her disposal a set of
version of the method for application in an actual regional loss motions corresponding to one or two scenarios. To avoid the use
assessment study (Volos project) was recently presented by of complicated and cumbersome procedures, one then has to
Kappos et al. [25]. DPMs for masonry buildings (brick or stone make recourse to the convenient, albeit simplistic, PGA scaling
masonry) were recently derived by Penelis et al. [30] using a of the surface motions, which inevitably leads to the empirical
364 M. Dolce et al. / Engineering Structures 28 (2006) 357371

Fig. 9. Pushover curves for one-, two- and three-storey masonry buildings.

PGAintensity correlations, which are associated with a large For the purposes of vulnerability assessment a total
scatter. of 36 typical masonry building configurations have been
Clearly, the alternative (modern) approach of developing modelled [31], with the following main parameters:
DPMs (and/or fragility curves) in terms of spectral quantities
Number of storeys (one to three)
can lead to a more convenient selection of the input motions;
Materials (low/high quality, shear strength 100 and 200 kPa,
however, the problem then arises that for the hybrid method
respectively)
to be used, empirical data in terms of spectral quantities are
Dimension of piers (large/small, depending on openings
necessary (first step of the method), which are hardly available
size)
to date, hence recourse to empirical correlations between
Dimension of spandrels (large/small)
intensity and spectral quantities (or PGA) becomes mandatory
again. The pushover curves (normalised base shear vs. top storey
drift) for single-storey, two-storey, and three-storey buildings,
4.1.2. Method based on pushover analysis are shown in Fig. 9, in their bilinearized form (which is useful
The main advantage of this procedure compared to time- when seismic demand is calculated from response spectra).
history analysis is that there is no need to select an appropriate As expected, normalised yield strength V /W (where V is the
number and type of input motions, and no need to consider base shear and W the weight of the structure) increases as
hysteretic models for the members; most other issues and the number of storeys decreases, but the corresponding yield
problems are similar in both static and dynamic inelastic displacement decreases. It is interesting to note that the capacity
analysis. On the other hand, the definition of earthquake curves suggested in HAZUS for typical URM buildings (a
input in pushover analysis is clearly a problem to which a single-storey and a three-storey one), are elastoplastic with
number of different solutions have been suggested. Pushover maximum strengths V /W of 0.20 and 0.17, respectively,
analysis is seen by the authors mainly as a viable procedure which are not very different (about 30%) from those
to tackle the assessment of structures for which time-history in Fig. 10, but corresponding ultimate displacements of
analysis is cumbersome, if at all feasible, such as masonry 1.8% and 0.6%, which are significantly larger than those
buildings in general, and R/C buildings for which a spatial predicted by Penelis et al., and also in tests of large-scale URM
(3D) model is necessary for their assessment. It is noted, specimens [9]. On the other hand, one should take into account
though, that application of pushover analysis to irregular and/or that HAZUS adopts very large ultimate displacements with
higher-mode-sensitive structures is still the subject of ongoing a view to defining damage states corresponding to partial
research. collapse, not simply exceedance of the deformation capacity of
The cornerstone of this assessment procedure is the a few members.
construction of a representative pushover curve for each of the For the development of fragility curves using the pushover
building classes considered. If a refined scheme, such as that analysis based procedure, the seismic demand (to be compared
in HAZUS [20] is envisaged, a substantial number of curves with the capacity of the structure) has to be expressed in terms
are required, and feasible but realistic inelastic models are of an inelastic response spectrum (or a corresponding elastic
clearly preferred. The Greek group has investigated a number one for equivalent damping >5%). The degree of damage in
of alternative models for elastic as well as inelastic masonry masonry buildings is then defined by comparing these demands
buildings subjected to horizontal loading and compared results with the capacity curve for each building type and applying the
with available experimental evidence [23]. It appears that a criteria shown in Table 4, wherein the damage index varies from
model based on equivalent frame discretization and the familiar 0% to 100% and in this study is taken equal to the loss index,
concept of idealized plastic hinges in the critical regions of piers i.e. (repair cost)/(replacement cost). Note that damage states are
and spandrels is the appropriate choice whenever the expected defined in terms of top storey drift (top displacement divided by
type of failure and/or damage is of the in-plane type. total height), which is compared with appropriate fractions of
M. Dolce et al. / Engineering Structures 28 (2006) 357371 365

Fig. 10. Damage scenarios for Potenza, using the Italian and Greek methodologies (in terms of number of buildings).

the drift at yield ( y ) and at failure (u ), rather than with fixed the information on the (generic) structure is available in terms
values of interstorey drift (as done in most previous studies). of a global pushover curve, rather than as a series of storey shear
This is much better suited to the procedure used herein, since vs. storey drift curves; obviously, depending on the building
366 M. Dolce et al. / Engineering Structures 28 (2006) 357371

Table 4
Definition of damage states in masonry buildings

Damage state Range of damage index Definition Top storey drift


0 0 No damage < 0.7y
1 >05 Slight damage 0.7 y < < 0.7 y + 5(0.9u 0.7 y )/100
2 >520 Moderate damage 0.7 y + (5(0.9u 0.7 y ))/100 < < 0.7 y + (20(0.9u 0.7 y ))/100
3 >2050 Substantial to heavy damage 0.7 y + (20(0.9u 0.7 y ))/100 < < 0.7 y + (50(0.9u 0.7 y ))/100
4 >5095 Very heavy damage 0.7 y + (50(0.9u 0.7 y ))/100 < < 0.7 y + (100(0.9u 0.7 y ))/100
5 >95100 Destruction > u

height and failure mechanism, y and u values vary for each was summarized in the previous section of this paper and is
building type (see Fig. 9). presented in more detail elsewhere [30,31].
Based on the foregoing considerations, R/C buildings in
4.2. Application of the methodology to the Potenza building Potenza were classified into one of the categories shown in
stock Table 5. The information from the database with respect to R/C
structural systems was either R/C walls or R/C frames.
4.2.1. Assumptions and input data As in the city studied the former class was deemed generally
The starting point of the study was the database of 9281 closer to a dual system than to a pure wall system, it was
buildings (private and public buildings in Potenza) described classified as RC1 (dual system, i.e. combination of frames and
in Section 2. The approach adopted by the Italian group was R/C walls), rather than RC3 (wall system). R/C frames were
to assign each building, according to the combination of its assigned to the RC2 class (infilled frames, i.e. R/C frames with
horizontal and vertical system, to one of the vulnerability masonry walls used as cladding and/or partitions). Depending
classes (A to D), for which empirical DPMs were already on whether the building was pre-1980 or post-1980, it was sub-
available for Italy [7,8] or were derived specifically for this classified as I or E, respectively (Inferior or Enhanced seismic
study (for class D), by interpreting in a quantitative way design); the implicit assumption is that post-1980 buildings
the qualitative (linguistic) description given in the EMS in Italy have similar performance to post-1985 buildings in
table [18]. Greece (1985 is the year of the first major revision of the
The Potenza inventory was processed by the Greek group Greek seismic code). Finally, according to its height, each
with a view to setting up classes for which Greek DPMs and R/C building was classified as L (one to three storeys), M
fragility curves were already available. It is emphasised that (four to seven storeys), or H (eight or more storeys). It is
it was not within the scope of this exercise to develop new seen from Table 5 that, in terms of volume, the most common
DPMs for the Italian buildings using the hybrid methodology R/C building types in Potenza are RC2I-M and RC2I-H
described in Section 4.1. (medium-rise and high-rise frame systems, with inferior
The building stock in Potenza consists of the following seismic design).
categories, using the construction material of the main For masonry buildings, the recently derived DPMs [30,31]
structural system as the criterion: based on a combination of statistical damage data from
the Thessaloniki 1978 and Aegion 1995 earthquakes, and
R/C: 72.8% of volume (36.3% of number of buildings)
nonlinear analyses of typical building configurations (as
URM: 26.5% of volume (62.1% of number of buildings)
described in Section 4.1 of this paper) were used. Two classes
Other (steel): 0.6% of volume (1.3% of number of buildings)
are distinguished with regard to type of masonry, i.e. stone
As pointed out earlier, it is important to distinguish between masonry SM (in Greece this typically corresponds to pre-1940
the description of the stock in terms of built volume, in which construction) and brick masonry BM (post-war construction,
case R/C construction clearly dominates, or in terms of number typically with R/C floors). Within each class, three different
of buildings, in which case masonry buildings are by far the DPMs are set up, based on the height range: L (one or two
most numerous. In either case, though, it is clear that URM storeys), M (three or four storeys), and H (five or more storeys).
construction is an important component of the existing stock. Note that the height classification is different from that used
It is noted that, as a rule, this is not the case in Greek in R/C construction, and that the two DPMs for class H
cities (as opposed to villages), where post-war structures are are tentative ones (mainly based on ATC-13 data), since tall
predominantly built in R/C and (at least in terms of volume) masonry buildings are extremely rare in Greece and data on
clearly dominate the entire stock (see the Volos case in [25]). their seismic damage is not available. In the absence of more
As a result of this situation, differentiation in the vulnerability detailed information, masonry buildings reported in the Potenza
of different R/C building types has long been a major concern database as good quality masonry and/or with R/C floors
in Greece. On the other hand, work on the vulnerability of were assigned to the Greek BM class, and all others to the SM
masonry buildings has been considerably less so far, and class. From the aforementioned categories (rows) of Table 5 it
has focussed more on the assessment of specific important is seen that low-rise masonry buildings are substantially more
and/or monumental structures, rather than on ordinary masonry than medium and high-rise ones, in terms of number (2:1) but
buildings. Recent work by the Greek group on such buildings not in terms of volume (about 0.3:1).
M. Dolce et al. / Engineering Structures 28 (2006) 357371 367

Table 5
Classification of Potenza buildings by AUTh group

Type Description Volume (m3 ) % of total Number % of total


RC1I-L dual, poor design, low 264,757 1.8% 425 4.6%
RC1E-L dual, good design, low 87,741 0.6% 124 1.3%
RC1I-M dual, poor design, medium 257,507 1.7% 74 0.8%
RC1E-M dual, good design, medium 55,188 0.4% 16 0.2%
RC1I-H dual, poor design, tall 201,493 1.4% 7 0.1%
RC1E-H dual, good design, tall 61,754 0.4% 3 0.0%
RC2I-L frame, poor design, low 860,236 5.8% 606 6.5%
RC2E-L frame, good design, low 779,710 5.2% 728 7.8%
RC2I-M frame, poor design, medium 4,137,942 27.8% 827 8.9%
RC2E-M frame, good design, medium 1,489,547 10.0% 319 3.4%
RC2I-H frame, poor design, tall 2,027,789 13.6% 188 2.0%
RC2E-H frame, good design, tall 608,519 4.1% 65 0.7%
SM-L stone masonry, low 429,818 2.9% 1263 13.6%
SM-M stone masonry, medium 883,719 5.9% 912 9.8%
SM-H stone masonry, high 982,396 6.6% 267 2.9%
BM-L brick masonry, low 662,169 4.5% 2591 27.9%
BM-M brick masonry, medium 611,158 4.1% 689 7.4%
BM-H brick masonry, high 370,400 2.5% 51 0.6%
Steel steel structures of any type 66,281 0.4% 106 1.1%
Other other types of buildings 32,127 0.2% 20 0.3%

Total 14,870,251 100% 9281 100%

Table 6
Definition of damage states in each method

Damage state Definition Structural and non-structural damage Assumption A Assumption B


DIa range to be considered DIa range to be considered
0 No damage 0 0
1 Slight damage No SD slight N-SD >010 >01
2 Moderate damage Slight SD moderate N-SD 1030 130
3 Substantial to heavy damage Moderate SD heavy N-SD 3060 3060
4 Very heavy damage Heavy SD very heavy N-SD 60100 60100
5 Destruction Very heavy SD 100 100
a DI: Damage Index; SD = Structural Damage; N-SD = Non-Structural Damage.

Having selected DPMs for each category, a further issue was In the scenario development, both (structural) damage
to establish an appropriate correspondence between the damage assessment and loss assessment were carried out, in line with
state definition used in each methodology. In the Italian [14] the assessments made by the Italian group.
and EMS [18] classification, a total of six damage states
(levels), 05, as shown in the first column of Table 6 were used 4.2.2. Key results
(defined as shown in the second and third columns). The Greek Shown in Fig. 10 are three damage scenarios for the entire
DPMs for R/C structures were originally developed following building stock, corresponding to three different intensities con-
the ATC damage classification scheme, having seven damage sidered (VI, VII, VIII); it is recalled that the scenario for
states. The correspondence between the lowest (0) and the three I = VI corresponds to the reference earthquake with return
highest (35) damage states is relatively straightforward, but period of 50 yr, while the reference earthquake with return pe-
levels 1 and 2 are a bit ambiguous, as ATC defines a slight riod of 475 yr corresponds to an intensity somewhat lower than
damage state with damage index (DI) between 0% and 1% (of VIII [14], but for clarity of comparisons, two different scenarios
replacement cost), and light damage state, with DI between for uniform I = VII and VIII are considered. A similar damage
1% and 10%; moderate damage corresponds to DI between scenario in terms of building volume is shown in Fig. 11.
10% and 30%. To see the sensitivity of results to the different A number of similarities, but also some notable differences
definitions, two alternative assumptions were made, as shown are seen in the scenarios of Figs. 10 and 11. It is easily noted
in Table 6; it is noted that such assumptions are critical for that there is significant difference between the graphs produced
loss scenarios, since levels 02 are very commonly predicted based on the number of buildings (Fig. 10) and the ones based
in risk studies. The Greek DPMs for masonry buildings were on the volume (Fig. 11) which can be mainly attributed to the
developed for six damage states (as shown in Table 4), hence fact that masonry buildings which are a large percentage of the
correspondence with the Italian and EMS classification was total building stock in number, are significantly less in volume.
straightforward. Considering the graphs of Fig. 10 (number of buildings) it is
368 M. Dolce et al. / Engineering Structures 28 (2006) 357371

Fig. 11. Damage scenarios for Potenza, using the Italian and Greek methodologies (in terms of building volume).

easily noted that the Greek method results in a scenario which buildings are in damage states (0) and (1) using the Greek
involves less damage than the one derived using the Italian approach, while this percentage is reduced to only 70% using
method. For example in the case of I = VII, 85% of the the Italian approach resulting in higher number of buildings
M. Dolce et al. / Engineering Structures 28 (2006) 357371 369

Fig. 12. Comparison of relative repair cost values of the entire building stock
(T Cr,r ) relevant to the reference earthquakes based on building area.

suffering higher degrees of damage in the latter case. Similar


trends are identified for I = VIII.
Fig. 12 compares the estimated damage, calculated using the
Italian and Greek approach, for the three intensity scenarios
in terms of relative repair cost of the entire building stock
(%), using the buildings area. Monetary values in the Greek
Fig. 13. Comparison of vulnerability curves: Class A of the Italian
methodology have also been calculated assuming that the methodology with low quality (stone masonry) medium height URM buildings.
average cost of replacement for Greek practice would be
e600/m2 ($720/m2), as opposed to the e700/m2 used in
the Italian approach. Consistently with the damage scenarios
summarised in Figs. 10 and 11, the Italian procedure predicts
higher cost of repair of damaged buildings for intensity VIII,
but for intensity VII it gives slightly lower predictions than the
Greek methodology.

5. Comparison among Italian and Greek results

The fact that the Italian approach produces more destruc-


tive scenarios means that the Italian DPMs, for a given inten-
sity and for a specific building type, predict higher vulnerabil-
ity than the Greek ones, which suggests that either Italian con-
struction practices (for the considered building stock) are infe-
rior to Greek ones, or that the intensities attributed to specific
events from which the DPMs have been derived are defined in
an inconsistent way in the two studies, or both. The latter has
been demonstrated in other studies [30] for URM buildings, in
which the Thessaloniki and Aegion events were suggested to
have lower intensities than the ones initially attributed to them.
Of course, several other parameters interfere, such as
differences in the correspondence of building types, in the
interpretation of damage levels, in local construction practices
(and their effect on the quality of a particular building type),
and (at least for cases other than I = VII) to the fact that Fig. 14. Comparison of vulnerability curves: Class B of the Italian
methodology with good quality (brick masonry) medium height URM
Greek DPMs for I > VII are based on theoretical analysis (see buildings.
Section 4.1) rather than on statistical data.
In Figs. 13 and 14 the vulnerability curves of Italian (and for the same type of building the Italian methodology considers
EMS) Class A and B are plotted against the vulnerability higher damage than the Greek one for a given intensity.
curves derived within the Greek methodology for low quality 6. Summary and conclusions
and good quality medium-rise URM buildings, respectively.
Based on these (and similar) comparisons, the reasons for the The work presented herein is a first attempt for a systematic
aforementioned observation become clear, since it is shown that comparison of the methodologies used for vulnerability
370 M. Dolce et al. / Engineering Structures 28 (2006) 357371

assessment and earthquake damage scenario development in and appropriate models for correlating structural damage
the two European countries that have the highest seismic risk indices (ductility factors, interstorey drifts) to loss. It is noted
(Italy and Greece). Comparisons are focussed on a particular that, in their form used in this study, the Greek DPMs were
case study (so that more practical conclusions can be drawn) not derived from corresponding fragility (vulnerability) curves
but more general issues like the composition of the building based on statistical distributions (like the beta distribution used
stock, the (structural) typology of buildings and schemes in the Italian DPMs or the lognormal distribution used in the
for its classification, and the relative merits of alternative HAZUS fragility curves), hence uncertainty was not treated in a
methods for deriving Damage Probability Matrices (DPMs) and consistent way, i.e. by differentiating between different sources
vulnerability curves, are also tackled. of uncertainty (as done e.g. in HAZUS). A new generation
The availability of a large amount of data regarding of Greek DPMs is currently being finalised, wherein they are
the building stock has given us the possibility to develop derived by the corresponding lognormal fragility curves. On the
earthquake damage scenarios for the city of Potenza, using other hand, the Greek DPMs for R/C structures correspond to
methodologies typically used in Italy and in Greece. The significantly more typologies than currently used in Italy, and
inventory of the building stock has been obtained from a survey this should be attributed both to the method adopted and to the
carried out after the 1990 Potenza earthquake and in 1999 different composition of the building stock in the cities of the
this inventory was updated to include the reinforced concrete two countries.
buildings built after 1990. The comparative study of Potenza has shown similarities but
In the Italian methodology, a hybrid technique has been also differences between the predictions of the Italian and the
used to evaluate the absolute vulnerability of the buildings of Greek DPMs. In this respect one should recall that statistical
interest, where typological analyses and expert judgement are data of earthquake damage is less rich in Greece, and this
combined. After a unification of the large number of vertical should be one of the key reasons why the estimated losses in the
and horizontal structural types contained in the GNDT90 case study of Potenza are lower for high intensities, compared
survey form, a limited number of different buildings types to the predictions of the Italian methodology. Following the
has been obtained and each building has been assigned 1999 Athens earthquake in Greece, a significant amount
to the relevant vulnerability class. Beyond the classes of of damage data has been gathered and is currently being
vulnerability A, B and C of the MSK-scale, class D of the incorporated in revised versions of the hybrid methodology,
1998 European Macroseismic Scale, for the less vulnerable to produce the aforementioned new generation of Greek
buildings, has been considered. The DPMs found from the fragility curves and DPMs. Other possible factors to which the
damage data of the 1980 earthquake were used to define the discrepancies between the two methodologies can be attributed,
absolute vulnerability of the different vulnerability classes, are different construction practices, different trends in assigning
i.e. the probabilistic damage distributions for given earthquake intensities, uncertainty in correlating intensity with level of
intensities. However, as the DPM of class D was not available, seismic action used in the analyses, and differences in the
it was extrapolated from the class C with a purposely set definition of the building types as well as of the damage
up procedure with reference to the EMS98 scale. As regards states.
the selection of the reference earthquakes, three events have Finally, the comparative study presented herein, despite its
been considered as representative of probable or strong events relatively limited scope, clearly points to the importance of
expected in Potenza. Their intensities have been obtained by a consistent definition of the damage state and the potential
recent seismic hazard studies of the Italian territory. Three advantages from expressing them in economic rather than
earthquake damage scenarios relevant to dwelling buildings linguistic terms.
have been prepared both in terms of damage and of economic
losses, involving the selected reference earthquakes. They show References
a generally low vulnerability and, consequently, a limited [1] Albarello D, Bosi V, Bramerini F, Lucantoni A, Naso G, Peruzza L et al.
number of damaged buildings for the lower intensity, and of New seismic hazard maps of the italian territory. Italian National Seismic
partially or totally collapsed buildings, for the higher intensity Service and Italian National Group for the Protection against Earthquakes.
earthquake. web site http://www.dstn.it/ssn/PROG/2000/carte pericolosita; 1998.
[2] ATC. Earthquake damage evaluation data for California (ATC-13).
In the Greek methodology, a hybrid technique was also used, Redwood City (California): Appl. Technology Council; 1985.
but in this case inelastic analysis of generic structures, deemed [3] Azzara R, Basili A, Beranzoli L, Chiarabba C, Di Giovanbattista R,
to be representative of the corresponding building typology, Selvaggi G. The seismic sequence of Potenza (May 1990). Annali di
was used in lieu of expert judgement. The other (most crucial) Geofisica (Rome) 1993;XXXVI(1).
[4] Barbat AH, Moya FY, Canas JA. Damage scenarios simulation for
component of the hybrid approach, statistical data of earthquake
seismic risk assessment in urban zones. Earthquake Spectra 1996;12(3):
damage, is essentially the same as in the Italian method. 37194.
According to the Greek approach, statistical data (preferably, [5] Bard PY et al. Seismic zonation methodology for the city of Nice-Progress
albeit not necessarily, in terms of the loss index, i.e. repair report. In: Proceedings 3rd intern. conf. on seismic zonation, vol. III,
cost to replacement cost ratio) is directly used for constructing p. 174984.
[6] Bernardini A. Qualitative and quantitative measures in seismic damage
the DPM columns for intensities for which data is considered assessment and forecasting of masonry buildings. In: Proc. of the
adequate, while the remaining parts of the DPM are constructed international workshop on measures of seismic damage to masonry
using inelastic dynamic or static analysis of generic buildings buildings; 1998.
M. Dolce et al. / Engineering Structures 28 (2006) 357371 371

[7] Braga F, Dolce M, Liberatore D. Southern Italy November 23, 1980 32743.
Earthquake: A statistical study on damaged buildings and an ensuing [20] FEMA-NIBS. Earthquake loss estimation methodology - HAZUS99
review of the M.S.K.-76 scale. CNR-PFG n.503. Roma; 1982. Technical Manual, vols. 13. Washington DC; 1999.
[8] Braga F, Dolce M, Liberatore D. A Statistical study on damaged [21] Goretti A, Di Pasquale G. Perceptibility of seismic events at national
buildings and an ensuing review of the M.S.K.76 scale. In: 7th European scale. In: Proceed. of the 10th Italian national conference on earthquake
conference on earthquake engineering; 1985. engineering; 2001 [in Italian] [CD ROM].
[9] Calvi GM, Kingsley GR, Magenes G. Testing of masonry structures for [22] Kappos AJ. Seismic vulnerability assessment of existing buildings in
seismic assessment. Earthquake Spectra 1996;12(1):14562. Southern Europe, Keynote lecture. In: Proceed. of the 10th Italian national
[10] DAyala DF, Spence RJS, Oliveira CS, Silva P. Vulnerability of conference on earthquake engineering; 2001 [CD ROM].
buildings in historic town centres: A limit-state approach. In: 11th world [23] Kappos AJ, Penelis GrG, Drakopoulos C. Evaluation of simplified models
conference on earthquake engineering. Paper No. 864, Pergamon [CD for the analysis of unreinforced masonry (URM) buildings. Journal of
ROM Proceedings]; 1996. Structural Engineering, ASCE 2002;128(7):8907.
[11] Di Pasquale G, Orsini G, Serra C. Assessment of the economic losses [24] Kappos A, Pitilakis K, Stylianidis K, Morfidis K, Asimakopoulos D. Cost-
from the GNDT-DPC-SSN safety evaluation forms. In: Proc. of the benefit analysis for the seismic rehabilitation of buildings in Thessaloniki,
international workshop on measures of seismic damage to masonry based on a hybrid method of vulnerability assessment. In: Proceedings of
buildings; 1998. 3rd intern. conf. on seismic zonation, vol. I. 1995. p. 40613.
[12] Di Pasquale G, Goretti A. Economic and functional vulnerability of [25] Kappos A, Pitilakis K, Morfidis K, Hatzinikolaou N. Vulnerability
residential buildings stricken by Italian recent seismic events. In: Proceed. and risk study of Volos (Greece) metropolitan area. In: 2th European
of the 10th Italian national conference on earthquake engineering; 2001 conference on earthquake engineering, CD Rom Proceedings, Paper 074;
[in Italian] [CD ROM]. 2002.
[13] Dolce M. Seismic vulnerability evaluation and damage scenarios. In: US [26] Kappos AJ, Stylianidis KC, Pitilakis K. Development of seismic risk
Italian workshop seismic evaluation and retrofit. New York City: scenarios based on a hybrid method of vulnerability assessment. Natural
Columbia University; 1996. Hazards 1998;17(2):17792.
[14] Dolce M, Marino M, Masi A, Vona M. Seismic vulnerability analysis and [27] Margottini C, Molin D, Narcisi B, Serva L. Intensity vs. acceleration:
damage scenarios of Potenza. In: International workshop on seismic risk italian data. In: Proceedings of workshop on historical seismicity of
and earthquake damage scenarios of Potenza; 2000. central-eastern mediterranean region; 1987.
[15] Dolce M, Marino M, Masi A, Vona M. Evaluation of site effects for [28] Masi A. Seismic vulnerability assessment of gravity load designed R/C
seismic microzonation. In: International workshop on seismic risk and frames. Bulletin of Earthquake Engineering 2003;1(3):37195.
earthquake damage scenarios of Potenza; 2000. [29] Masi A, Dolce M, Vona M. A procedure to estimate economic losses
[16] Dolce M, Masi A, Marino M. EAEE-ESC Task Group 3 Seismic risk due to damage at residential buildings. In: Proc. DiSGG, no. 5; 2002 [in
and earthquake damage scenarios report of the activities. In: Proceed. of Italian].
12th ECEE; 2002 [CD ROM]. [30] Penelis GrG, Kappos AJ, Stylianidis KC, Lagomarsino S. Statistical
[17] Dolce M, Masi A, Marino M, Vona M. Earthquake damage scenarios assessment of the vulnerability of unreinforced masonry buildings. In:
of Potenza town (Southern Italy) including site effects. Bulletin of International conference earthquake loss estimation and risk reduction;
Earthquake Engineering 2003;1(1):11540. 2002.
[18] ESC Working Group Macroseismic Scales. European Macroseismic Scale [31] Penelis GrG, Kappos AJ, Stylianidis KC. Assessment of the seismic
1998. GeoForschungsZentrum Potsdam, Germany, 1998. vulnerability of unreinforced masonry buildings. In: STREMAH 2003,
[19] Faccioli E, Pessina V, Calvi GM, Borzi B. A study on damage scenarios 8th international conference on structural studies, repairs and maintenance
for residential buildings in Catania city. Journal of Seismology 1999;3(3): of heritage architecture. WIT Press; 2003. p. 57584.

Vous aimerez peut-être aussi