You are on page 1of 38

Journal of Earthquake Engineering

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ueqe20

Displacement-Based Simplified Seismic Loss


Assessment of Masonry Buildings

Daria Ottonelli , Serena Cattari & Sergio Lagomarsino

To cite this article: Daria Ottonelli , Serena Cattari & Sergio Lagomarsino (2020) Displacement-
Based Simplified Seismic Loss Assessment of Masonry Buildings, Journal of Earthquake
Engineering, 24:sup1, 23-59, DOI: 10.1080/13632469.2020.1755747

To link to this article: https://doi.org/10.1080/13632469.2020.1755747

Published online: 08 Jun 2020.

Submit your article to this journal

Article views: 39

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=ueqe20
JOURNAL OF EARTHQUAKE ENGINEERING
2020, VOL. 24, NO. S1, 23–59
https://doi.org/10.1080/13632469.2020.1755747

Displacement-Based Simplified Seismic Loss Assessment of


Masonry Buildings
Daria Ottonelli , Serena Cattari , and Sergio Lagomarsino
Department of Civil, Chemical and Environmental Engineering (DICCA), Polytechnic School, University of
Genoa, Genova, Italy

ABSTRACT ARTICLE HISTORY


In the framework of the evaluation of the consequences of an earth- Received 30 July 2019
quake, the paper provides a procedure for the economic loss assess- Accepted 10 April 2020
ment in unreinforced masonry (URM) buildings, proposing an KEYWORDS
analytical approach based on a cost function directly dependent on Loss analysis; masonry
the damage level associated to different structural elements and structures; cost function;
identified by specific engineering demand parameters. The proce- consequence function;
dure follows a component-based approach, to be implemented equivalent frame model;
within modern probabilistic seismic risk analyses, and it includes probabilistic seismic
both the in-plane and out-of-plane damage modes that can be assessment
activated in a URM structure. It is applied to a three-storey building,
examined varying some configurations of structural details, aimed to
simulate two different in-plane collapse mechanisms (with damage
concentrated mostly on piers or also in spandrels), or to allow, in one
case, also the activation of local out-of-plane mechanisms.

1. Introduction
The seismic risk is defined as the potential for negative consequences due to a hazardous
event in a specific area (a city, a region, a state, or a nation) and a given period of time,
determined by the probabilistic convolution of hazard, exposure, and vulnerability. In
general, the consequences of an earthquake are the physical damage to buildings and other
facilities, the casualties, the potential economic losses due to the direct cost of damage and
to indirect economic impacts, the loss of function in lifelines and critical facilities and also
the social, organizational, and institutional impacts. Among all these aspects, in this paper
the attention is focused only to the potential economic losses due to the direct cost of
damage. They are synthetically expressed in the parameter called Expected Annual Loss
(EAL) that measures the average yearly loss by considering the frequency and severity of
all possible earthquake losses and it is seen as fraction of the overall value of the building
replacement cost (Porter, Beck, and Shaikhutdinov 2004).
Nowadays, the literature shows great interest in this aspect, since seismic risk analysis
developed in economic terms could be effective for different scopes:

(i) cost-benefit evaluations, comparing the costs of a mitigation scheme, such as


retrofit, to the benefits achieved from improved seismic performance;

CONTACT Daria Ottonelli daria.ottonelli@unige.it Department of Civil, Chemical and Environmental Engineering
(DICCA), Polytechnic School, University of Genoa, Via Montallegro 1, Genova 16145, Italy
© 2020 Taylor & Francis Group, LLC
24 D. OTTONELLI ET AL.

(ii) the knowledge of the economic impact and the resilience of a society after the
occurrence of an earthquake;
(iii) the calibration of insurance premia;
(iv) the “Building Seismic Performance Classification,” as proposed in Italy in the
Ministerial Decree of February 27, 2017, integrated with the changes of March 7,
2017 (Ministry Decree n.65 7/3/2017 2017) and, firstly, already proposed in Calvi,
Sullivan, and Welch (2014).

It is important to note that the cost-benefit evaluations could orient a new approach for
the future of the design, implying a transition from Performance-Based Design (PBD),
essentially based on the correlation of the structural response to specific Performance
Levels (PL), also called Limit States (LS) in codes, to the EAL design, as recently
proposed in O’Reilly and Calvi (2019). PBD represents the current approach adopted
in almost all codes at national and international levels. Indeed, some codes mention the
concept of economic losses, for example, the significant damage PL is defined as “… is
uneconomic to repair”; however, it remains a purely qualitative concept, being then the
quantitative criteria for the identification of the PL only function of the structural
response parameters. The CNR DT 212 (2014) document, issued by the National
Council of Research, rather than referring to the more traditional life safety PL, has
proposed the introduction of the severe damage PL that is related to a state of damage
uneconomic to repair associated to the loss of use of non-structural systems and
a residual capacity of main structure of resisting to horizontal actions. In this docu-
ment, for the first time, the criterion for the attainment of the severe damage PL is
explicitly based on a cost function, aimed to quantify the economic losses.
In the analysis of seismic consequences, the lack of literature and tools is particularly
noticeable, especially for masonry buildings within the European built environment. This
is due to the fact that the methodology for seismic loss evaluation has been developed
mainly in the USA, where other structural typologies (like as steel and reinforced concrete
ones) are most spread.
Loss estimation is usually based on the definition of consequence functions, represent-
ing, for example, the distribution of the loss on a set of discrete performance or damage
levels (DL). These functions can be derived from different approaches: i) experimental
tests; ii) empirical data from post-earthquake survey; iii) analytical data related to the
building structural response.
Experimental tests provide valuable and accurate data in particular to correlate the
damage levels to the progression of nonlinear response at the scale of single element
(structural or nonstructural). In this framework, for example, recent literature has focused
great attention to experimental campaigns of masonry infills for which, on the one hand,
increasing is the number of tests that provide relationships between cracks and DLs (e.g., in
Akhoundi, Vasconcelos, and Lourenço 2018; Morandi, Hak, and Magenes 2018a) and, on the
other one, correlations between DLs and losses have been proposed (e.g., in Cardone and
Perrone 2015; Sassun et al. 2016). Indeed, in the case of masonry structures, while great effort
has been devoted to the first step relationship between cracks and DLs, e.g., in Vanin et al.
(2017), Kržan et al. (2015), Petry and Beyer (2014), the second one is completely missing.
In the case of the empirical approach, it is important to note that in literature most of
the data are available in damage terms and not in economic ones. To this specific aim, the
JOURNAL OF EARTHQUAKE ENGINEERING 25

consequence functions can be obtained processing the data available from the reconstruc-
tion phase on basis of the analysis of funding necessary for repairing and rebuilding
damaged structures. However, nowadays such type of data is very limited in literature; in
Italy, some data from Pollino earthquake 1998 (Di Pasquale and Goretti 2001) and
L’Aquila earthquake 2009 (Del Vecchio et al. 2018; Di Ludovico et al. 2017a, 2017b;
Dolce and Manfedi 2015) are available. A critical issue in this approach is the definition of
the damage metric that has to be consistent between the one adopted for interpreting the
structural response and the one used in the consequences functions. It is evident that
empirical data are valuable because they are directly correlated to the actual data, but
when considering masonry structures, they are strongly influenced by the types of build-
ings investigated, since the characteristics of masonry buildings are dependent from the
local seismic culture and the available materials in the area. Thus, the extrapolation of
empirical consequence functions for traditional masonry buildings to other geographic
areas can be questionable.
Therefore, for masonry structures, the most advisable approach is the analytical one,
which is based on the definition of a cost function directly dependent on the damage level
identified by specific Engineering Demand Parameters (EDP), related to the response of
different structural elements (as proposed, for example, in CNR-DT 212/2013(2014) for
the significant damage PL). Within this context, this paper proposes a loss estimation
analytical model, that can be applied to any single building, according to the component-
based approach (Mitrani-Reiser and Beck 2007), as described in Section 2. The compo-
nents are the parts (structural and non-structural) that all together comprise a building.
According to this approach, the direct loss is calculated by summing the losses over all
damageable components in the building.
The proposed procedure complies with the aims and general framework of FEMA
P-58-1 (2012) developing the traditional steps of the seismic risk (hazard, exposure, and
vulnerability analysis) by using an accurate and detailed modelling and analysis of the
structural response. The vulnerability part entails the set-up of a 3D structural model and
the execution of Non-Linear Static (NLSA) and Dynamic Analyses (NLDA), following the
Multiple Stripe Analysis (MSA) method proposed in Cornell et al. (2002). For the hazard,
it is necessary to have: the seismic hazard curve discretized by given values of the intensity
measure IM (Fig. 1a), and, for each IM value, the corresponding sets of time-histories
(with the aim of executing NLDA). For what concerns the exposure, the quantification of
economic contribution of the different components (structural elements, non-structural,

Figure 1. Steps of the loss estimation procedure developed for masonry buildings in seismic area: (a)
hazard, (b) vulnerability, and (c) loss curve.
26 D. OTTONELLI ET AL.

and contents) in a masonry building has been detailed in the procedure (Section 2.1). The
non-linear analyses executed allow (Section 2.2):

● to determine the vulnerability curve (relationship between loss ratio, LR, defined as
the percentage of the replacement cost, and IM, Fig. 1b);
● to draw, from the data of the vulnerability curve, the loss curve as a series of points
associated to the number of intensities assumed to cover the hazard curve for the site.
The loss curve is thus described in terms of mean annual frequency, λ(ΙΜ), and loss
ratio (LR) (Fig. 1c);
● finally, to assess the EAL as the area under the loss curve: in that way, the expected
losses at each intensity level are combined with the expected annual probability
(Solberg et al. 2008).

With these data, an intensity-based and scenario-based assessment according to FEMA P-58
(2012) can be performed. In the paper, the vulnerability step has been developed for the
different collapse mechanisms observed in a masonry structure, which can be traced back to
two main groups: the global response characterized by prevailing in-plane damage modes of
walls (Section 2.2), and the out-of-plane mechanisms usually involving only local portions of
the structure (Section 2.2.3). Both can be referred to the structural and nonstructural
components of the building. The proposed procedure has been applied to a case study
(Section 3) varying some configurations of structural details, aimed to simulate two different
in-plane collapse mechanisms (with damage concentrated mostly on piers or also in span-
drels), or to allow, in one case, also the activation of local out-of-plane mechanisms.

2. DBLA Framework and Specific Considerations for Masonry Buildings


This section describes the proposal of the loss assessment methodology for a single
masonry building, aligned to the PEER-PBEE (Performance-Based Earthquake
Engineering) procedure (Porter 2003) and to FEMA P58-1 (2012). The PEER frame-
work (Moehle and Deirelein 2004; Porter 2003) applies the total probability theorem to
predict earthquake consequences in terms of the probability of incurring specific values
of performance measures or outcomes including casualties, repair costs, and downtime.
According to the component-based approach (Mitrani-Reiser and Beck 2007), the total
EAL in a building is equal to the sum of the EALj of each j-th component (Equation 1)
(j = 1 … N) that derives from the repair and replacement costs of those damaged during
the seismic event, properly weighted:

X
N
EAL ¼ EALj αj (1)
j¼1

where αj is the economic weight of each component in a masonry building, determined


during the exposure analysis (Section 2.1).
JOURNAL OF EARTHQUAKE ENGINEERING 27

2.1. Exposure Analysis


The exposure analysis (or inventory of the building) entails the identification of each
structural element and non-structural component that may be damaged, thus imply losses
in the building. The components to be considered for a typical masonry building are listed
in Table 1.
In the application of the proposed procedure, the reference structural model is the
equivalent frame model according to which the walls are discretized in piers and span-
drels, where the non-linear response is concentrated, connected by rigid areas (nodes).
Piers are the main vertical resistant elements carrying both vertical and lateral loads;
spandrel elements, which are intended to be those parts of walls between two vertically
aligned openings, are secondary horizontal elements for what concerns vertical loads,
which couple the response of adjacent piers in the case of lateral loads. For this reason, the
structural components are divided in piers and spandrels; if other modeling strategies are
adopted the components of the structural part may change.
For each component, the following parameters are reported: the EDP assumed as repre-
sentative for its response (for example, the piers drift – θp -, the spandrels drift – θs -, the
angular deformation of the horizontal diaphragms – θd -, the drift of the partitions – θl -, the
Peak Floor Acceleration – PFA– for ceilings and contents as introduced in Table 1); the unit of
measure for the costs, in terms of volume of piers or partitions (obtained from the area Ap or
Apa multiplied by the interstory height hl), area of diaphragms (Ad) or ceilings (Ac,l), and
number of spandrels (Ns); the cost (C) for each element (Cp for piers, Cs for spandrels, Cd for
diaphragms, Cpa for partitions, Cc for ceilings, and Cco for contents) according to the units in
Table 1; the total loss of each component (Lw for walls, Ld for diaphragms, Lpa for partitions, Lc
for ceilings, and Lco for contents). The choice of the EDPs to be adopted as reference could be
changed as a function of the component typology and of their role in the structural typology
examined. By way of example, in the seismic response of an unreinforced masonry building,
thin masonry walls (with thickness about 10–15 cm) are usually considered partitions and
their seismic response is neglected, even if they are part of the whole structural system. Like
any internal wall in a URM building, they are vertically compressed and moreover benefit of
the confinement effect provided by diaphragms; therefore, they are mainly subjected to the in-
plane response. For this reason, they are considered as drift sensitive elements as also
proposed in the literature, above all related to the PBEE approach followed in the paper
(FEMA P58-1 2012; Taghavi and Miranda 2003). Despite that, it is not the only possibility: for
example, the infill elements in reinforced concrete (r.c.) structures are often subjected to the
out-of-plane response, being particularly sensitive to the amplification phenomena of the

Table 1. Inventory of the components for a masonry building.


N° ID Components EDPs Unit of measure Cost Loss
1. Structural Piers θp Ap (p = 1,Np*) Cp Lw
hl (l = 1,Nl**)
Spandrels θs Ns Cs
Diaphragms θd Ad (d = 1,Nd***) Cd Ld
2. Non structural Partitions θl Apa,l Cpa Lpa
hl (l = 1,Nl)
Ceilings PFA Ac,l (l = 1,Nl+1) Cc Lc
3. Contents PFA - Cco Lco
*Np: Number of piers; **Nl: Number of levels/storeys; ***Nd: Number of diaphragms.
28 D. OTTONELLI ET AL.

seismic input due to the filtering effect provided by the main structure. In the latter case, they
have more appropriately to be considered as peak floor acceleration (PFA) sensitive. The value
C for each element correspond to the maximum repair value, equivalent to the reconstruction
value. In the last column of Table 1, the results that can be obtained from the procedure are
listed.

2.2. Vulnerability Analysis


This part aims to quantify the vulnerability of each damageable component, and the
corresponding consequences in terms of repairs costs. Figure 2 provides an overview of
all components (structural and non-structural) and failure modes (both associated to the
global in-plane response and to the local mechanisms) considered in the proposed
procedure. As already mentioned, a crucial step is to define the loss curve (λ – LR) through
the quantification of reliable vulnerability curves (LR-IM). The latter ones can be defined:

Figure 2. Identification of the components in a masonry building as a function of their seismic response
and steps for the loss assessment.
JOURNAL OF EARTHQUAKE ENGINEERING 29

a) for discrete generic values of the intensity measure selected (IMi); or b) only for specific
values of IM, in general those associated to specific performance levels (IMPL). It is evident
that the first approach (a) allows a more accurate discretization of the loss curve and thus,
in principle, a more detailed estimate of the final EAL. However, on the one hand, it
usually requires a higher computational effort and, on the other one, it implies an
analytical correlation between the response of the component under examination and
the consequent losses, which is not available in the literature for all the components that
concur to the quantification of the EAL.
In the procedure that has been developed, the first approach (a) is followed for
assessing the losses due to the in-plane response, while the second (b) is adopted for the
non-structural components and the out-of-plane local response. Moreover, by way of
example and to compare the results achievable from a more or less discretized loss curve,
also in the case of the in-plane response an approach based on the definition of PLs has
been applied (approach b, see Section 3.3).
For the in-plane response, a specific procedure has been developed that constitutes one
of the main novelties of the research illustrated allowing the quantification of the vulner-
ability curve for the specific masonry building under examination. The procedure aims to
be flexible to capture the differences which derive from the large variety of structural
details that can characterize this structural typology.
Figure 3 summarizes the approach adopted for defining the vulnerability curve asso-
ciated with the generic component that contributes to the in-plane response. As already
mentioned, it implies the execution of NLDA. For each record (the n-th of the whole set
selected) of every considered intensity measure (IMi), the loss ratio LR(IMi) is assessed for
the wall and diaphragm components according to the analytical procedure illustrated in
detail in Section 2.2.1. The latter is based, in the case of vertical structural elements, on an
analytical cost function dependent on the drift of piers and spandrels (Section 2.2.1.1),
that are then combined at the scale of the wall as a whole (Section 2.2.1.2); instead, for the
diaphragms, the variable assumed for assessing the replacement cost is the angular
deformation. Given a certain value of IM, the set of results from the accelerograms can
be then interpreted according to an appropriate probability distribution function in order
to define the points associated to the 16%, 50%, and 84% quantiles. By analogously
proceeding for all the IMi values, the final vulnerability curve is defined (Fig. 3).
For the second approach (b), a fundamental step is the identification and quantification of
the PLs. Usually, codes (e.g., NTC08, ASCE/SEI 41-13 2014; EC8-part 1 EN1998 2005) refer
for the structural response to four PLs: operational (O), damage limitation (DL), life safety (LS)
and near collapse (NC) (as per the notation adopted by the NTC08). Indeed, for the purposes

Figure 3. Main steps required for the definition of the vulnerability curve from NLDA.
30 D. OTTONELLI ET AL.

of the loss evaluation, it is necessary to add the so-called zero loss (ZL) PL, as also proposed in
the Ministry Decree n.65 7/3/2017. This is motivated by the need to identify a limit state that
does not require interventions and therefore does not contribute to the economic losses (the
starting point of the loss curve, as showing Fig. 9a). In fact, the first applications of the
procedure (Section 4) outlined that the operational PL (the first PL traditionally considered in
codes) already generates considerable losses, being not suitable to be adopted as the anchor
point of the loss curve. Usually, the losses associated with the ZL PL are assumed equal to 0
while those associated to the NC PL are assumed equal to 1. Moreover, for the same reasons, it
is worth highlighting that even the first approach (a) requires the definition of the ZL PL to cut
and close the loss curve. For the definition of the ZL PL, the Ministry Decree n.65 7/3/2017
proposes to conventionally assume a return period (TR) equal to 10 years, which corresponds
to θZL = 0.1. For the Authors, having an a-priori fixed value of TR is too conventional
considering that the seismic hazard can be very different at national scale (e.g., the hazard
varies significantly in Italy when passing from Milan to L’Aquila). Thus, in this paper,
a specific procedure to quantify the ZL PL (Ottonelli 2016) is applied which is based on the
multiscale approach originally proposed in Lagomarsino and Cattari (2015a, 2015b). The
latter allows to define the PLs – through the execution of either NLSA or NLDA – by
considering the behavior of single elements (local damage in piers and spandrels), macroele-
ments (drift in masonry walls and horizontal diaphragms) and of the entire building (normal-
ized total base shear, from the global pushover curve). Within this general approach, the
proposal for the ZL PL considers only the structural elements scale, and in particular the piers
(Ottonelli 2016). In particular, for this aim, reference is made only to piers since they are the
main structural elements aimed to support both the vertical and horizontal actions; indeed,
spandrels in existing building are often slightly damaged due to ordinary loads and usually are
not repaired, by considering “light cracks” as quite physiological (apart from specific cases, for
example, due to well-proven settlement problems). In Section 3.3, an application of such way
to define the ZL PL is presented.

2.2.1. Loss Assessment Procedure for Structural Elements Due to the In-plane
Response
The present section describes the analytical loss model proposed for the structural
elements in a masonry building. It is based on different stages at different scales: the
development of a cost function at the element scale, which is assumed as the minimum
scale useful to start with defining the repair (Section 2.2.1.1); the definition of
a methodology to combine the costs deriving from the damage diffusion on single
elements and to pass to the macroelement scale (wall) and to evaluate the costs associated
with the damage on diaphragms (Section 2.2.1.2); the computation of the loss ratio and
the EAL at the global scale of the building (Section 2.2.1.2).

2.2.1.1. Element Scale: The Analytical Cost Function. According to the equivalent frame
idealization and as introduced at Section 2.1, the reference structural elements are: the
piers and spandrels in the vertical walls; and the horizontal diaphragms. The normalized
repair cost function of a single element is defined as c(EDP) where the EDP is representa-
tive of the specific element (see Fig. 4a) and Fig. 4b) and represents the variable adopted at
the scale of the constitutive model of the elements correlated to the progressing of non-
linear response (i.e., with the attainment of higher DLEi) (Fig. 4c). This function can vary
JOURNAL OF EARTHQUAKE ENGINEERING 31

DIAPHRAGMS WALLS
DLE0 DLE1 DLE2 DLE3 DLE4 DLE5
1.00 1.00

C
C

V
0.00 0.00 θ1 θ2 θ3 θ4 θ5
EDPC1 EDPC0 EDPC1
EDP EDP = θ
EDP

a. b. c.

Figure 4. (a) Repair cost function c(EDP) for the diaphragms; (b) Repair cost function c(EDP) for
spandrels and piers; (c) Constitutive models adopted for masonry elements as implemented in the
Tremuri software adopted for the application (Cattari et al. 2018).

with a linear or non-linear trend between the zero value – set for a lower threshold EDPC0
(which can be zero) – and the value 1 set for an EDP equal or greater than a higher
threshold EDPC1. In the case of three structural elements herein considered, the repre-
sentative EDPs are: the drift for the piers (θp) and spandrels (θs) and the angular
deformation for the diaphragms (γd).
For the purpose of the cost function, in the case of the vertical elements, the threshold
EDPC0 is the drift for which a slight damage occurs (θ1 referred to the DLE1 of the
element), while the threshold EDPC1 defines the state beyond which it is not convenient
repair the damage (θ4). The latter is referred to the DLE4 of piers and spandrels, with the
exception of spandrel with arch lintel for which the limit is anticipated to the attainment
of DLE3 (θ3); this anticipation is motivated by the quite brittle behavior and very moderate
residual strength evidenced by experimental tests in such a case (Beyer and Dazio 2012).
In particular, as a simplified approach for the vertical elements, the cost function can be
assumed linear.
8
< 0 if 0  EDPe < EDPC0
ce ¼ EDPEDPe
if EDPC0  EDPe < EDPC1 (2)
: C1
1 if EDPe > EDPC1
In order to convert the normalized cost function with the simplified linear trend into
a function based on the actual costs, the real costs of different possible strengthening
techniques for piers (Table 2) and spandrels (Table 3) have been examined by consulting
both: regional price lists of repair actions (Abruzzo 2001, Emilia Romagna 2015); and the
costs resulting from the actual strengthening projects carried out on masonry buildings
damaged by recent Italian earthquakes (Dolce and Manfedi 2015). In particular, in order
to construct the cost functions and consequently define the black markers in Fig. 5, three
steps were performed: 1) to associate the execution of different types of interventions to
specific damage levels; 2) to define the costs of these interventions through the aforemen-
tioned price lists; 3) to determine the average cost for having a single reference value for
each damage level. Step 1) was supported by the expert judgment acquired from the
analysis of the recurrent types of interventions realized after various Italian earthquakes,
above all Umbria and Marche 1997 and L’Aquila 2009 (Di Ludovico et al. 2017a, 2017b).
Furthermore, the definition of this function accounts for the fact that the choice of the
most suitable strengthening intervention varies also with the actual damage reached: it
means that as the DLE attained progresses, the main technique to be adopted is expected
32 D. OTTONELLI ET AL.

Table 2. Actual costs for the piers repair.


Repair techniques for PIERS Cost U.M. DLEi
Filling of cracks with insertion of slivers 8.50 €/m2 1
Filling of cracks with mortar injection 9.50 €/m2 1
Repair of single cracks with mortar injection 12 €/m2 1
Filling of cracks with mortar injection, insertion of slivers and cuci-scuci over 10% of the surface 34 €/m2 2
Digging out of mortar from joints and deep repointing 36 €/m2 2
Strengthening with FRP stripes 60 €/m2 2
Repair of single cracks with reinforced concrete jacketing 115 €/m2 2
Strengthening of the lintel by FRP* 170 €/m2 3
Substitution of the lintel and rebuilding of the spandrel 195 €/m2 4
*In the case of spandrels with arched lintel, the demolition and rebuilding is already considered for DLE3

Table 3. Actual costs for the spandrels repairs.


Components Description of the strengthening techniques Costs - C Total
Partition Brick wall reconstruction 200 €/m3 805 €/m3
Prepare masonry surface (brushing, low-pressure water cleaning, etc.) 1.29 €/m2*
Patch new plaster 25.50 €/m2*
Paint 20 €/m2*
Ceiling Provision and installation of new ceiling, considering all costs necessary 32 €/m2
to give the work done in a professional manner.
*Assuming a thickness of 0.15 m

400 400
Euro per m 2

Euro per m 2

350 350
300 300
250 250
200 200
150 150
100 100
Flexural Arch Lintel
50 50
Shear Wooden Lintel
0 0
0.0% 0.2% 0.4% 0.6% 0.8% 1.0% 1.2% 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% 1.2%
θ θ
a. b.

Figure 5. Cost function of: (a) piers; (b) spandrels.

to be different. It is important to note that the repair description must include all works
associated with the repair of the actual damage as well as collateral work associated with
removal and replacement of finishes, relocation of utilities, provision of scaffolding, and
similar actions. In the case of piers, the cost function has been differentiated with the
prevailing failure mode occurred, if associated to a flexural or shear response (Fig. 5a).
This is justified by the fact that the spread of damage is different in these two cases: mostly
concentrated at the end sections in the flexural response and more spread on the whole
panel in the case of the shear (as shown in the images of Fig. 6). On the contrary, in the
cost function of the spandrels, a significant role is attributed to the type of lintel (Fig. 5b).
In fact, as testified by various experimental campaigns (Beyer and Dazio 2012; Graziotti,
Magenes, and Penna 2012), when an arch architrave is present the strength decay usually
JOURNAL OF EARTHQUAKE ENGINEERING 33

Figure 6. The two collapse mechanisms involved in the cost function: (a) shear; (b) flexural.

occurs more suddenly and with a higher entity than the case in which timber or steel
lintels are coupled to the spandrel (Fig. 7).
For what concerns the diaphragms, a very simplified cost function such as that
proposed in Fig. 4a has been assumed where losses pass directly from 0 to 1 when the
angular deformation exceeds a specific threshold (EDPC1). The value of 1 corresponds to
the reconstruction cost. Of course, similarly to what done for masonry elements, this cost
function ought to be improved in the future by establishing a correlation between the
progressing of non-linear response of diaphragms (as already investigated by experimental
campaigns like those of Brignola, Pampanin, and Podestà (2012) and Giongo et al. (2014))
and the repair costs.

2.2.1.2. Macroelement and Global Scale: The Loss Ratio Evaluation. In the following
two variables representative of the total repair cost of the walls (Lw) and diaphragms (Ld)
are introduced. They are obtained by properly combining the cost functions defined at the
scale of single structural element c(EDP), in particular cp(θp,l,w) for piers, cs(θs,l,w) for
spandrels, cd(θd,l) for diaphragms:

Figure 7. The two mechanisms with two different types of lintel involved in the cost function: (a)
wooden (Graziotti, Magenes, and Penna 2012); (b) arch (Beyer and Dazio 2012).
34 D. OTTONELLI ET AL.

X
Nw X X
Nl;w Np;l;w
  X
Nw X X
Nl;w N s;l;w
 
Lw ¼ Ap;l;w hl Cp cp θp;l;w þ Cs cs θs;l;w (3)
w¼1 l¼1 p¼1 w¼1 l¼1 s¼1

XNl XNd;l  
Ld ¼ l¼1 d¼1
A d;l C d c d γd;l (4)

where the sums are extended over all piers (Np = ΣNp,l,w), spandrels (Ns = ΣNs,l,w), walls
(Nw), diaphragms (Nd = ΣNd,l) and levels (Nl). The other terms are clarified in Table 1.
It is important to note that the total repair cost of the structural elements is also
influenced by the global response of the building: for example, when the soft-story
mechanism occurs the total repair cost of the walls Lw increases (Fig. 8a). More precisely,
when in a given wall at the generic level l*, the interstory drift of the walls (θl,w) exceeds
a given threshold, the total repair cost functions of the walls and diaphragms have
a sudden growth (as illustrated by the dashed line in Fig. 8b). It is justified by the fact
that all piers, spandrels, and diaphragms located in that wall in the higher levels are
considered to be rebuilt, that is they assume the value of the cost function equal to 1.
According to that, the increase of the repair cost is estimated according to the following
formula.

X X
Nw N X
l ;w N p;l;w
  XNw X
Nl;w X
Np;l;w
Lw ¼ Ap;l;w hl Cp cp θp;l;w þ Ap;l;w hl Cp
w¼1 l¼1 p¼1 w¼1 l¼Nl ;w þ1 p¼1
(5)
X X
Nw N X
l ;w Ns;l;w
  X
Nw X
Nl;w
þ Cs cs θs;l;w þ Ns;l;w Cs
w¼1 l¼1 s¼1 w¼1 l¼Nl ;w þ1

Nl X
X Nd;l   X
Nl X
Nd;l
Ld ¼ Ad;l Cd cd γd;l þ Ad;l Cd (6)
l¼1 d¼1 l¼Nl þ1 d¼1

where Nd,l* is the number of diaphragms supported by the wall subjected to the soft-story
mechanism and Nl* is the level of the soft storey, if it does not happen Nl* is equal Nl.
The maximum repair costs for walls and diaphragms, associated with the complete
damage of all elements are given by:

1.00 1.00
Lw

Lw

c (Structural Elements)
c (Structural Elements)
c (Soft Storey)
0.00 0.00

t t

a. b.

Figure 8. (a) Trend of the losses of the walls for a generic analysis (solid line); (b) Increment of the wall
loss due to the occurrence of the soft-storey mechanism (dashed line).
JOURNAL OF EARTHQUAKE ENGINEERING 35

X
Nw X X
Nl;w N p;l;w X
Nw X X
Nl;w N s;l;w

LREPAIR;w ¼ Ap;l;w hl Cp þ Cs (7)


w¼1 l¼1 p¼1 w¼1 l¼1 s¼1

X
Nl X
Nd;l
LREPAIR;d ¼ Ad;l Cd (8)
l¼1 d¼1

The repair loss ratio (lw for the walls and ld for the diaphragms) can be obtained by normalizing
Equations (3) and (4) through the maximum repair cost (Equations 7 and 8), considering also
the increase of the repair cost due to the occurrence of the soft-story mechanism. Under the
hypothesis that Cp, Cs, and Cd are constant all over the building, the following formulas
come out:
2P   PNw PNl;w 3
Nw PNl ;w PNp;l;w PNp;l;w
A h C θ
p;l;w l p p;l;w þ l¼Nl ;w þ1 A h
p;l;w l
lw ¼ αp 4 5
w¼1 l¼1 p¼1 w¼1 p¼1
PNw PNl;w PNp;l;w
w¼1 l¼1 p¼1 A p;l;w h l
2 3 (9)
  1 X Nw NX X
l ;w N s;l;w
  X Nw X
Nl;w
þ 1  αp 4 cs θs;l;w þ Ns;l;w 5
Ns w¼1 l¼1 s¼1 l¼1 l¼N  þ1
l ;w

PNl PNd;l   P PNd;l


l¼1 d¼1 Ad;l cd γd;l þ Nl¼N l
l  þ1 d¼1 Ad;l
ld ¼ PNl PNd;l (10)
l¼1 d¼1 Ad;l

where θp is the weight of the piers on the vertical walls of the building, computed in terms of
ratio between the pier transversal section over the overall one.
The loss ratio lr of the whole set of structural elements in the building (piers, spandrels, and
diaphragms) is obtained by normalizing the total repair costs of walls and diaphragms (Lw+Ld)
by the total replacement cost of the structural elements (LREBUILD); the latter is given by the
rebuilding cost if this is lower than repairing cost of whole elements (LREPAIR,w+ LREPAIR,d).
Also in previous researches on the same topic (Mitrani-Reiser and Beck 2007; Porter, Beck,
and Shaikhutdinov 2004), losses are expressed in terms of a mean damage factor (MDF) aimed
to represent the percentage of the replacement cost of the buildings (Welch, Sullivan, and
Calvi 2014). The loss ratio lr is given by:
lw LREPAIR;w þ ld LREPAIR;d
lr ¼ ¼ χ ½αw lw þ ð1  αw Þld  (11)
LREBUILD

LREPAIR;w þ LREPAIR;d
χ¼ (12)
LREBUILD

LREPAIR;w
αw ¼ (13)
LREPAIR;w þ LREPAIR;d

2.2.2. Loss Assessment Procedure for Nonstructural Elements


As introduced in Section 2.2, unlike the structural elements, the losses associated to the
non-structural components damage are herein determined passing through the PLs or
36 D. OTTONELLI ET AL.

DLs, whose attainment is associated with that of specific damage levels. The attainment of
the given DLs is monitored by the EDP selected as the most sensitive to the seismic
response of the non-structural element under examination. Then, the consequence func-
tions (DLs – LR,NON-STRUCTURAL) have to be introduced to estimate the repair costs. These
latter are selected from literature, together with the values of EDP capacity for each DL. In
general, they have a probabilistic characterization that is usually assumed according to the
lognormal distribution (with median EDP50%,C and dispersion βC). The EDP demand
value is recorded from the structural analysis, from each time-history of the NLD analyses
executed.
For the computation of the probability of occurrence of each λDL, it is particularly
convenient referring to the closed-form solution adopted in the SAC/FEMA approach
(Cornell et al. 2002) according to the power-law relationship introduced in Vamvatsikos
(2013, 2014) for fitting the hazard curve. According to this general framework, the
vulnerability curve associated to the PLs can be obtained passing, for example, through
the incremental dynamic analysis (IDA) curves (Vamvatsikos and Cornell 2002), introdu-
cing the IM-EDP relation aimed at providing a description of the structural seismic
response or “demand” versus the IM. IDA obtained by a set of ground motion records
may be used to get a probabilistic representation of the seismic demand conditioned to
IM, with the definition of a median, 16% and 84% quantiles. From the IDA, the capacity of
the non-structural elements is evaluated. From the fraction of records causing the occur-
rence of LS at each IM level, the fragility curves P(DL|IM) can be evaluated. From the
median and quantile values, the IDA curves are obtained by a linear regression defined by
the parameters a and b (Equation 14) in log-log coordinates:

EDP50%;D ðimÞ  aðimÞb (14)


Thus, the mean annual frequency of exceedance of each DL is computed through the SAC/
FEMA equation that becomes (15):
" 1 #  2
EDP50%;D b k1  2 
λDL ¼ H exp 2 βD þ βC 2
(15)
a 2b

where the EDP demand is also taken as lognormal with dispersion βD, which is assumed
independent on the level of IM, and the EDP50%,D is the conditional median that depends
on intensity via a power-law relationship. Considering, for the hazard curve the second-
order polynomial in log-space and the probabilistic EDP capacity, Equation (15) becomes:
 2
pffiffiffi 1ϕ   ϕ k1
λDL ¼ ϕk0 H imEDP50%;C exp ð1  ϕÞ (16)
4k2

1
ϕ¼   (17)
1þ 2k2 β2D þ β2C =b2

 b1
EDP50%;C
imEDP50%;C ¼ (18)
a
Indeed, in the application proposed in the paper the NLDA is executed following the MSA
approach based on suite of ground motion records selected for a given IM and not on
JOURNAL OF EARTHQUAKE ENGINEERING 37

Figure 9. (a) Limit state loss curve and (b) consequence functions (Cardone and Perrone 2015).

a fixed set of time histories scaled to several levels of seismic intensity as per IDA.
However, in both cases, the results of NLDA can be interpreted to define the relationship
EDP-IM and thus proceeding to the linear regression required by Equation (14). Once
defined the exceedance of each λDL, the loss assessment of non-structural elements is
determined by the introduction of the consequence functions (DLi-LR,NON-STRUCTURAL),
from which the values of losses associated to the different limit states are extrapolated, as
in Fig. 9b. From the values of λDL and loss ratio (LR,NON_STRUCTURAL), the loss curve for
the non-structural elements can be traced (Fig. 9a) and, finally, by taking the area under
the loss curve, the
EALNON-STRUCTURAL is computed. If a discrete form is considered (as a function of the
selected DLi), the EAL can be defined as:
X LR;DLi þ LR;DLiþ1 
EALNONSTRUCTURAL ¼ ðλDLi  λDLiþ1 Þ (19)
all
2
LR;DLi

2.2.3. Loss Assessment Procedure for Structural Elements Due to the Out-of-plane
Response
The methodology proposed in this section allows the computation of losses due to the
occurrence of the out-of-plane mechanisms. This is particularly important for the existing
buildings prone to local mechanisms, such as structures without tie-rods or buildings in
aggregate typical of historical centers.
The loss evaluation is carried out through the following steps:

(i) identification of all the possible local mechanisms in each macroelement (m = 1 …


M), following a critical analysis of the structural and technological characteristics
of the building (e.g., the presence of tie-rods or ring beams, a good connection
between the walls and walls and floors, the presence of pushing elements like roof
or vaults, …);
(ii) attribution of a weight for the macroelement interested by the mechanism (βm)
with respect of the overall building, mainly based on geometric characteristics;
(iii) computation of the capacity curve for each possible mechanism identified in step
(i) (e.g., through the application of the nonlinear kinematic analysis as depicted in
Fig. 10a);
38 D. OTTONELLI ET AL.

Figure 10. (a) Example of capacity curve and damage levels (Lagomarsino 2015); (b) The collapse of
a wall due to its out-of-plane response.

(iv) identification of the damage levels on the capacity curves assessed in step (iii) (Fig. 10a);
(v) computation of the mean annual frequency of exceedance of each damage level for
all possible mechanisms;
(vi) selection of the mechanism most likely to occur.

The macro-blocks model represents the most common choice to analyze the local
mechanisms according to the kinematic approach of the limit analysis or to execute
dynamic analyses of simple single or multi-body systems. In fact, the observation of
damage after many post-earthquake surveys highlighted the propensity of historical
masonry structures to develop damage modes involving the formation of macro-blocks
as well as recurring typologies of mechanism (D’Ayala and Speranza 2003; Lagomarsino
2015). The latter ones mainly support the analyst in step i) to identify a priori the
structural masonry portions (blocks) involved in the mechanism.
Then, the most common solution is to carry out the seismic assessment thought the
limit analysis following a kinematic approach with a macro-block model, as also recom-
mended by the Italian Structural Code (MIT 2009). To this aim it is necessary to define
the: the blocks of the kinematic chain considering the presence or absence of good
connections between the walls, the direction of the floors, the presence or the absence
of tie-rods, the distribution of the openings in the wall, …; the constraints between the
blocks; the forces involved in the mechanisms.
For the aim of losses computation, two damage levels are particularly significant:

● the activation of the mechanism assumed to correspond to DL2, representative of the point
where the losses begin to accumulate (LR,DL2 = 0);
● the collapse (Fig. 10b) assumed to correspond to DL4, representative of the point
which the reconstruction of the wall corresponds to (LR,DL4 = Cm).

The LR,DL4 is quantified as a fraction (ζm) of the global loss value (LR,GLOBAL).
LR;DL4 ¼ Cm ¼ ζ m LR;GLOBAL (20)
JOURNAL OF EARTHQUAKE ENGINEERING 39

Once the capacity curve together with the DLs position has been defined, it is compared
with the spectrum in order to complete the seismic assessment; in particular, the over-
damped spectra approach has been adopted (Freeman 1998). Moreover, in the case of
local mechanisms that involve macroelements placed in the upper part of the building,
a proper modified response spectra aimed to take into account the filtering effect provided
by the main structure ought to be considered (e.g., as proposed in Calvi and Sullivan
(2014) and Degli Abbati, Cattari, and Lagomarsino (2018)). So, the values of the max-
imum peak ground acceleration (selected as the IM aimed to describe the seismic demand)
compatible with the fulfillment of a given DL have to be computed. After that, the fragility
curves for each DL of each mechanism are determined and then the mean annual
frequency of exceedance are computed (Equation 15) to finalize the assessment of the
loss curve (LR,DL – λDL), as outlined in Fig. 11. The area under the curve is the EALm
relative to a single mechanism.
Once evaluated the EALm for all the possible out-of-plane mechanisms expected in the
building, it is necessary to add them to that associated to the in-plane response to define
the overall EAL. The latter can be alternatively computed through the following relations:
!
XM XM
EAL ¼ 1  βm Cm EALGLOBAL þ EALm > EALGLOBAL (21)
m¼1 m¼1

X
M
EAL ¼ EALGLOBAL þ EALm > EALGLOBAL (22)
m¼1

Where: the first (Equation 21) represents the case in which the occurrence of the in-plane
and the out-of-plane response regards the same piers; while the second (Equation 22) the
case in which the damage occurs in the different piers. In the application illustrated in
Section 3, the second relation has been adopted.

3. Case-study
In order to illustrate the effectiveness of the analytical procedure proposed in Section 2,
the three storey ordinary building shown in Fig. 12 has been selected as case study. It has
a rectangular plan of 14.5 × 10 m2 and a total height of 10.8 m. The structure is built in

Figure 11. Example of loss curve for a local mechanism.


40 D. OTTONELLI ET AL.

Figure 12. Plan with the orientation of the floors and frontal view of the case study.

solid bricks and lime mortar. The thickness of the exterior walls varies from 48 cm in the
first two stories to 36 cm in the last one; instead the interior walls have a constant
thickness along the height equal to 24 cm. The structure is isolated, and the plan
configuration is simple and regular; the fronts have a regular windows disposition in the
x-direction and irregular in the y-direction.
The case study examined has been analyzed according to two configurations of
structural details aimed to simulate different global in-plane collapse mechanisms, namely:

● Type A configuration (Fig. 13a), with wooden floors and roof, associated to spandrels
coupled with tie-rods, located in each floor, in correspondence of the four exterior
walls. Such configuration is representative of a very common type of existing
masonry buildings and often produces a uniform collapse mechanism (where the
adjective “uniform” refers to a damage localized both in spandrels and piers);
● Type B configuration (Fig. 13b), with rigid floors and roof, and spandrels coupled
with r.c. ring-beams. This model is expected to produce a soft-storey collapse
mechanism.

The presence of tie-rods or r.c. ring-beams is considered sufficient to inhibit the activation
of local mechanisms, thus for these two first configurations, the computation of losses is

Figure 13. Sketches of the different collapse mechanisms analysed (Tomaževic 1999).
JOURNAL OF EARTHQUAKE ENGINEERING 41

limited only to the in-plane seismic response and to the contribution of non-structural
components. In order to exemplify the procedure for estimate the losses due to out-of-
plane response, a third configuration has been examined consistent to the Type
A configuration but assuming the tie-rods not effective to completely inhibit the local
mechanisms.
For the purpose of the evaluation, the building has been located in L’Aquila (AQ) that
represents a high seismicity site. The seismic action is then characterized in terms of: the
discrete hazard curve aimed to define the annual frequency of exceedance of ground
motions for nine different return periods; a set of 10 time-histories (considering x, y and
z components) for each point of the hazard curve useful to perform the NLDA. The
selection criteria of the time-histories are described in O’Reilly et al. (2019). Then starting
from the selected records, the acceleration and displacement response spectra have been
computed to perform the nonlinear static and kinematic analyses.

3.1. The Equivalent Frame Model Adopted for Analyzing the In-plane Response
For assessing the global in-plane seismic response and the EDPs necessary for the non-
structural components, the equivalent frame (EF) modeling approach has been adopted.
In particular, the analyses discussed in the following sections have been performed by
Tremuri software that allows executing both non-linear static and dynamic analyses
(Lagomarsino et al. 2013). Figure 14 shows the equivalent frame idealization of some
walls of the building. The diaphragms are modeled as orthotropic membrane finite
elements characterized by a shear modulus (Gd) and the Young moduli in two main
orientations (E1d and E2d).
Masonry panels are modeled as non-linear beam elements with lumped inelasticity
idealization and a piecewise linear behavior according to the formulation proposed in
(Cattari et al. 2018). The constitutive law allows to describe the non-linear response until
very severe damage levels (from 1 to 5) through progressing strength decay in correspon-
dence of assigned values of drift (Fig. 15). Moreover, also a hysteretic response is
formulated based on a phenomenological approach able to capture the differences in the
various failure modes that may occur (if flexural or shear prevailing in addition to the

Figure 14. Equivalent Frame model of three walls (where the piers are the red elements, the spandrels
the greens and the nodes the greys): (a) x-direction, parallel to the façade and (b) y-direction,
orthogonal to the façade.
42 D. OTTONELLI ET AL.

θ1 θ2 θ3 θ4 θ5 θ
Damage Levels (DLEs):
DLE0 DLE3
DLE1 DLE4
DLE2 DLE5

Figure 15. Piecewise linear constitutive laws implemented in TREMURI for nonlinear beams targeted for
masonry panels from Cattari et al. (2018

hybrid one) and in the response of piers and spandrels. Such a formulation revealed quite
efficient in performing non-linear analyses as recently shown in Cattari et al. (2018).

3.2. The Uncertainties Involved in the Analysis


The loss assessment procedure proposed is based on a fully probabilistic approach, that
requires understanding the inherent uncertainties, both of aleatory and epistemic nature,
involved in risk analysis. In the application illustrated herein, only the aleatory uncertain-
ties have been considered which include: the seismic intensity at the site, governed by the
hazard function; the record-to-record variability, described by the set of 10 records; the
material properties and the parameters that affect the non-linear response of masonry
panels. Moreover, the input components have been applied with 10 random directions:
thus the variability of results from the MSA takes into account the randomness in the
input motion. The effect of the aleatory uncertainties is quantified by associating one-to-
one a distinct realization of the random variables to each of the ground motions (10)
selected for each return period.

3.2.1. Range of Variability of the Aleatory Uncertainties


More specifically, the aleatory uncertainties considered are:

● the mechanical parameters of the solid bricks and lime mortar masonry. Within this
class, a group of correlated parameters has been considered: the Young’s modulus E,
the shear modulus G, the compressive strength fm, the mean shear strength fvm,0 and
the equivalent friction coefficient μ. For these variables, a lognormal probability
distribution is assumed (Table 4) while the range of variation is consistent with
that proposed in the Instruction of Italian Structural Code (MIT 2009) for the same
masonry typology;
● The parameters of the constitutive laws of piers and spandrels, which are: the drift,
the residual strength and the parameters aimed to describe the hysteretic response
(producing responses more or less dissipative). For the first, a lognormal probability
JOURNAL OF EARTHQUAKE ENGINEERING 43

Table 4. Probabilistic characterization of the random variables (parameters are mean μ and standard
deviation β for lognormal variables, and the shape parameters a and b for beta variables).
Set Parameter Variable Distribution μ/a μ/b Mean value
1 Masonry E (MPa] Lognormal 6.60 0.3 750
G (MPa] 5.50 0.3 250
^μ −1.06 0.2 0.35
fvm0 (MPa] −2.32 0.3 0.1
fm (MPa] 0.95 0.4 2.8
2 Constitutive Law Piers θT3 Lognormal −5.84 0.24 0.003
θT4 −5.32 0.22 0.005
θT5 −4.98 0.2 0.007
θPF3 −5.14 0.24 0.006
θPF4 −4.63 0.22 0.01
θPF5 −4.22 0.2 0.015
θT 3 Beta 0.77 0.33 0.7
θ T4 0.24 0.36 0.4
θ PF 1.317 0.233 0.85
3 Constitutive Law Spandrels θT3 = θPF3 Lognormal −6.24 0.24 0.002
θT4 = θPF4 −5.15 0.25 0.006
θT5 = θPF5 −3.94 0.25 0.02
θT3 = θT4 = θPF Beta 0.12 0.08 0.6
4 Damping Piers θmT Beta 0.480 0.12 0.8
θmT deterministic 0.8
θmT deterministic 0
θmT deterministic 0
θmPF −0.09 −0.01 0.9
θmPF deterministic 0.8
θmPF 0.840 0.56 0.6
θmPF 0.750 0.75 0.5
5 Damping Spandrels θmT=θmPF Beta −0.04 −0.16 0.2
θmTθθmPF 0.330 0.77 0.3

distribution is assumed, instead for the latter two (being described by factor varying
between 0 and 1) a beta probability distribution is considered (Table 4). The disper-
sion values of drift derive from experimental data, as proposed in Petry and Beyer
(2014), Morandi et al. (2018b), Kržan et al. (2015). The results of experimental tests
carried out on solid brick masonry piers helped us to calibrate the parameters
associated to the hysteretic response (Anthoine et al. 1995); for the spandrels
reference was made to the works of Beyer and Dazio (2012).

Naturally, other random variables may also play a role in a masonry building such as: the
stiffness and masses of diaphragms and roof or the properties of r.c. for the ring beam
(present in the case study Type B). For the former, the present work refers to a specific
type of floor that entails a reduced variability of the parameters allowing these values to be
deterministic and corresponding:

● for Type A, to timber floors and roof assumed to be representative of a semi-flexible


condition with Gd = 100 MPa, E1d = 12000 MPa and E2d = 1000 MPa;
● for Type B, to rigid floors and roof with Gd = 12500 MPa, E1d = 58800 MPa and E2d
= 30000 MPa.

The properties of reinforced concrete (r.c.) and reinforcement were instead considered
deterministic (Table 5) since the strength and ductility (associated to the flexural
44 D. OTTONELLI ET AL.

Table 5. Properties and geometrical characteristics of the RC ring beams.


Transversal section Longitudinal reinforcement Transversal reinforcement
Concrete C20/25 Steel B450 C Steel B450 C
B (cm] 40 Top 4 Φ 16 Asw Φ8
H (cm] 25 Bottom 4 Φ 16 Spacing (cm] 15
Concrete cover (cm] 3

response) of these elements is not so important as the ring beam mostly remains in elastic
phase. Indeed, the stirrups spacing value has been calibrated in order to avoid the shear
failure in r.c. ring beams; this is motivated by the fact this type of collapse was not
observed after earthquakes in URM buildings and it appears justified by the different
confinement effect which a r.c. ring beam is subjected to with respect to a r.c. beam in
a frame.
For the structural analyses, 10 different structural models were established, character-
ized by parameters randomly generated in accordance with such probability distributions
as illustrated in the following Fig. 16 (solid lines) and Table 4. The choice of adopting only
10 samples derived from the need to limit the computational effort but however provide
a general overview of all issues involved in the computation of fragility and loss functions.
In fact, despite the limit number of the sampling, it allows to include also the effect of the
aleatory uncertainty in the case study; to his aim, it is useful to recall that this uncertainty
would be representative of that of a class of buildings characterized by a homogeneous
behavior and not that could intrinsically be present in a single building. Figure 16 shows
the sampling that has been selected to be consistent with the median value and standard
deviation of the target distribution.

3.3. The Macro-block Model for Analyzing the Out-of-plane Response


In the examined case, the walls most vulnerable to the out-of-plane response are identified
in the P2 and P4 (as indicated in the plan view of the Fig. 12), since these walls are not
loaded by floors and thus they are subjected to low stabilizing actions. In this third case
study, the tie-rods have been considered ineffective. Two different possible mechanisms

4.5 4.5
Sampling Sampling
4 Target distribution 4 Target distribution
Median value Tar. Dist. =1.17 Median value Tar. Dist. =230
Median value Sampling =0.81 Median value Sampling =180
3.5 3.5

3 3

2.5 2.5
p(fm)

p(E)

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 1 2 3 4 5 6 200 300 400 500 600 700 800 900 1000 1100 1200
fm - MPa E - MPa

Figure 16. Histograms and new distributions resulting from the sampling compared with the original
distributions.
JOURNAL OF EARTHQUAKE ENGINEERING 45

Figure 17. Mechanisms and macroelements considered in the analyses.

have been analyzed for each wall that imply the partial overturning of the façade (Figure
17). In particular, the cylindrical hinge is expected to be activated at the basis of third
(following named “Mechanism 1”) and second floor (namely “Mechanism 2”), respec-
tively. The considered mechanisms in the case study are those assumed as the ones
potentially feasible considering the structural details that characterize it. In fact, according
to the assumption of a good connection between the walls (that is reduced in the upper
floor) and its effectiveness greater at the ground floor (due to a higher friction force), the
global overturning of the façade is not realistic.

3.4. Structural and Damage Analysis


In order to apply the loss assessment procedure proposed in Section 2, the structural
analyses have to be carried out, in particular performing for the in-plane response:

● Nonlinear dynamic analysis, by the MSA approach: 10 different records for each
return period were used (according to the selection criteria illustrated in O’Reilly
et al. (2019)). The reference IM is the spectral acceleration at the first mode period of
the structure, Sa(T1);
● Nonlinear static analyses with two different loading distributions: proportional to
mass (uniform) and inverted-triangular. These analyses were carried out in order to
apply the multi-scale approach (Lagomarsino and Cattari 2015a, 2015b; Ottonelli
2016) for the computation of the DLs.

For the out-of-plane response, instead, kinematic non-linear analyses have been executed.
Figure 18 shows, for one of the ten models, a comparison between the pushover curves
in two directions (± x and ±y) varying the two configurations (Type A and B – black curve
and grey curve, respectively, in Fig. 18) and the two load patterns (inverted-triangular with
thicker line and uniform with thinner line in Fig. 18, respectively).
Pushover curves confirm what it was expected for the two configurations in terms of
different seismic response: higher strength and less ductility for Type B (characterized by
the presence of r.c. ring beams) than Type A. Figure 19 illustrates, for two different
intensities, the comparison between the results of nonlinear dynamic and static analyses in
terms of base shear – top storey displacement. While in the elastic phase the two load
46 D. OTTONELLI ET AL.

2000 2000

1500 1500

V [KN]
V [KN]
1000 1000
500 500
0 0
-500 -500 Type A | Uniform
Type A | Uniform
Tye A | Inverted-Triangular Type A | Inverted-Triangular
-1000 -1000
Type B | Uniform Type B | Uniform
-1500 Type B | Inverted-Triangular -1500 Type B | Inverted-Triangular

-2000 -2000
-0.10 -0.05 0.00 0.05 0.10 -0.10 -0.05 0.00 0.05 0.10
d [m] d [m]
a. b.

Figure 18. Pushover curves for the different configurations of the case study: (a) x-direction; (b)
y-direction.

2000 2000
1500 1500
V [KN]

V [KN]
1000 1000
500 500
0 0
-500 -500
Uniform
Uniform
-1000 Inverted-Triangular -1000 Inverted-Triangular
Dynamic Dinamic
-1500 -1500
-2000 -2000
-0.10 -0.05 0.00 0.05 0.10 -0.10 -0.05 0.00 0.05 0.10
d [m] d [m]
a. b.
2000
2000
1500
V [KN]

1500
V [KN]

1000
1000
500
500
0 0
-500 -500
Uniform Uniform
-1000 Inverted-triangular -1000 Inverted-Triangular
Dinamic Dinamic
-1500 -1500
-2000 -2000
-0.10 -0.05 0.00 0.05 0.10 -0.10 -0.05 0.00 0.05 0.10
d [m] d [m]

c. d.

Figure 19. Comparison between NLSA and NLDA for one model in x-direction: (a and b) TR = 975 years;
c and d) TR = 2475 years.

patterns do not show significant differences, in the field of large displacement (in proxi-
mity of the collapse) the uniform distribution is more consistent with the result of NLDA;
moreover, increasing the intensity measure, a pronounced stiffness degradation and
increase of hysteresis cycles can be seen.
Passing to the out-of-plane response, the pushover curve (collapse multiplier α –
displacement d) and the capacity curve (spectral acceleration Sa – spectral displacement
Sd) of the equivalent SDOF (Fig. 20) have been computed. According to the proposal of
Lagomarsino (2015), in the capacity curve the DLs are identified in correspondence of the
JOURNAL OF EARTHQUAKE ENGINEERING 47

Sa [m/s2]
Mechanism 1 - P4
1.00 Mechanism 2 - P4

0.50

0.00
0.00 0.05 0.10 0.15 0.20 0.25

Sd [m]

Figure 20. Comparison of the capacity curves resulting for mechanism 1 and 2.

displacement at yielding in case of dDL2; the displacement equal to 0.4d0* (where d0* is the
displacement value where the acceleration of the capacity curve become 0) in case of dDL4.
For the aim of seismic assessment, since the hypothesized mechanisms involve upper
parts of the walls, a proper modified response spectrum has been evaluated according to
the formulation proposed in Degli Abbati, Cattari, and Lagomarsino (2018) aimed to take
into account the filtering effect provided by the structure.

4. Loss Estimation
Once defined the structural models of reference (Section 3.1), in order to perform the loss
assessment, an inventory of the structural and non-structural members was compiled
(Table 6). In this case, the contents of the building were not considered. The values of
costs C for each structural element are described in Section 2.2.1.1 and correspond to the
maximum repair cost, equivalent to the reconstruction. Instead for the non-structural
elements, C derives from the prices present in the following Table 7. The values proposed
in this table reflect the specific choices made for the examined case study. Of course,
although adopting different materials and technologies the costs can vary in absolute
terms, it is worth noting that, in case of URM buildings, it is expected they have
a marginal impact than the structural components; this is because, rather than r.c.
structures, in URM buildings not structural partitions are less present being in most
cases internal thin load-bearing masonry walls (Fig. 21). Once defined C and the quantity
of each component, the maximum economic loss of each one can be evaluated (Table 6).

Table 6. Inventory of the components for the case study.


N° ID Components EDPs Unit of measure Costs – C Maximum loss
1. Structural Piers θp ΣAp,front = 628 m2 350 €/m2 220000 €
ΣAp,trasv = 70 m2
h1 = 3.80 m
h2 = h3 = 3.50 m
Spandrels θs Ns = 83 200 €/m2 29000 €
Diaphgrams γd ΣAd = 420 m2 70 €/m2 29400 €
2. Non-structural Partitions θl ΣVpa = 11.25 m3 805 €/m3 9050 €
Ceilings PFA ΣAc = 420 m2 32 €/m2 13440 €
48 D. OTTONELLI ET AL.

Table 7. Interventions and costs for the non-structural elements.


Components Description of the strengthening techniques Costs – C
Partition Brick wall reconstruction 200 €/m3 805 €/m3
Prepare masonry surface (brushing, low-pressure water cleaning, etc.) 1.29 €/m2 *
Patch new plaster 25.50 €/m2 *
Paint 20 €/m2 *
Ceiling Provision and installation of new ceiling, considering all costs necessary to give 32 €/m2
the work done in a professional manner.
*Assuming a thickness of 0.15 m.

30%

5% 65%

Structural Elements Non-structural Elements (Partitions and Ceiling) Others

Figure 21. Graph of the economic weight of the different components in the masonry building
examined.

The latter ones also allow to compute the ratio between the different components of the
building and the economic weight of each one (Table 8).
The maximum repair cost of the structural elements and non-structural elements are
278400 € and 22515 €, respectively. In the case of a strengthening design, the following
items have to be added to this cost: different types of systems, finishing, flooring, and
windows (with frames). In particular, the following costs are assumed: electrical system –
19500 €; hydraulic system – 32400 €; heating system – 17760 €; finishing – 13000 €;
flooring −10000 €; windows – 34000 €. The following graph (Fig. 21) shows the economic
weight of each component, grouping in “others” the different types of systems, finishing,
flooring, and windows.

Table 8. Economic weight of the components for the case study.


N° ID Components Ratio piers and spandrel Maximum loss Ratio walls and floors Ratio str and nonstr*
1. Piers 0.880 249000 € 0.894 0.651* 0.968**
Spandrels 0.120
Diaphgrams 29400 € 0.106
2. Partitions 0.021* 0.032**
Ceilings 0.031*
*The sum is lower than 1, because the rest of the count is due to systems, finishing, flooring, and window.
**Ratio without consider ceilings, systems, finishing, flooring, and window, since they are not considered in the assessment
application.
JOURNAL OF EARTHQUAKE ENGINEERING 49

Finally, the total repair cost per square meter results in 985 €/m2. This is coherent with
recent values coming from the analysis of reconstruction costs after L’Aquila 2009 earth-
quake (Di Ludovico et al. 2017a, 2017b; Dolce and Manfedi 2015) with particular
reference to those including also the demolition of the heavily damaged building. With
regard to these latter, a reference range of prices is 900–1100 €/m2 (finishing included).
The cost related to the demolition of the entire building is considered equal to 18 €/m3. In
the case study under examination, having a volume of 1260 m3, the demolition cost results
in 22680 € while the demolition and construction amount to 954 €/m2. Based on the
assumptions made on the inventory of the components of the case study (that does not
consider, for example, the foundation system or different typologies of non-structural
elements and contents), it is emerged as the total reparation of the building is in practice
equivalent to the demolition and reconstruction, strictly from an economic point of view.
Thus, from the NLDA with MSA approach, at each level of IM the following values are
computed and analyzed statistically:

(1) for the determination of the losses of nonstructural elements: the EDPs for the different
components (Fig. 22), in terms of the mean inter-storey drift and PFA. The set of points
(IM = Sa(T1), EDP) is then fitted with a lognormal distribution (through the maximum
likelihood approach) in order to compute the median, the 16% and 84% quantiles and
the related dispersions;

Type A Type B
25 25
MSA results MSA results
Lognormal distribution Lognormal distribution
20 20
( ) ]
[m/s2 ]

Sa(T1) [m/s 2
IM - Sa(T1)

15 15
Sa(T1)

10 10

5 5

0 0
0 1 2 3 4 5 0 1 2 3 4 5
θ1 # 10
-3 EDP -θ1
IDR- Y1 # 10-3
Interstorey drift of the 1ST level
25 25
MSA results MSA results
Lognormal distribution Lognormal distribution
20 20
[m/s2]
[m/s 2]

IM - Sa(T1)
IM - Sa(T1)

15 15
Sa(T1)
Sa(T1)

10 10

5 5

0 0
0 5 10 15 0 5 10 15
Peak Floor Acceleration of the 3RD level

Figure 22. Result 1) the set of 10 points (IM, EDP) fitted with a lognormal distribution.
50 D. OTTONELLI ET AL.

(2) for the determination of the losses of structural elements: the set of 10 (IM = Sa(T1),
LR) points is fitted with a beta distribution, as in Fig. 23, computing the maximum
likelihood estimates of the beta distribution parameters a and b from the data, in
order to compute the median (Fig. 24) and the quantile values of 16% and 84%
points is fitted with a beta distribution (computing the maximum likelihood
estimates of the beta distribution parameters a and b from the data) in order to
compute the median (Fig. 23).

Finally, the multiscale approach (Lagomarsino and Cattari 2015a, 2015b; Ottonelli 2016)
has been implemented for the identification of the PLs (assumed to corresponds to DLs).
Thus the probabilities of occurrence of each PL have been determined and the numerical
fragility curves (Fig. 25) have been drawn. The points have been fitted with the lognormal
distribution in order to compute the IM median value and the dispersion θ in order to
apply both: the SAC/FEMA approach (Cornell et al. 2002; Vamvatsikos 2013) and the
rigorous calculation of the integral at the base of the PEER-PBEE methodology with the
final aim of determining the mean annual frequency of each PL. Once the median values
of IM that cause the occurrence of each PL are determined, it is possible to identify them
in the numerical vulnerability curve (Fig. 26), that associates to each PL a value of loss
(LR,PL) due to the structural components. The losses associated to the near-collapse PL are
assumed equal to the replacement cost, that is 1 in the curve.

Type A Type B
1.00 1.00
LR

LR

0.80 0.80

0.60 0.60

0.40 0.40

0.20 0.20

0.00 0.00
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Sa(T1) [m/s2] Sa(T1) [m/s2]

Figure 23. Result 2) the set of 10 points (IM, LW) fitted with a beta distribution.

Type A Type B
MSA results Beta distribution MSA results Beta distribution
1 1

0.8 0.8

0.6 0.6
PLv

PLv

0.4 0.4

0.2 0.2

0 0
0 5 10 15 20 0 5 10 15 20
Sa(T1) [m/s2] Sa(T1) [m/s2]

Figure 24. Result 2) the median vulnerability curves.


JOURNAL OF EARTHQUAKE ENGINEERING 51

Type A Type B
ZL ZL
1 SLO 1 SLO
SLD SLD
SLV SLV
0.8 SLC 0.8 SLC
P DM

P DM
0.6 0.6

0.4 0.4

Fitted curve Fitted curve


0.2 Numerical piecewise 0.2 Numerical piecewise
linear curve linear curve
0 0
0 5 10 15 20 0 5 10 15 20
IM - Sa(T1)
Sa(T1) [m/s2] Sa(T1) [m/s2]

Figure 25. The fragility curves for each PL: the numerical points and the fitted distributions.

Type A Type B
1.00 1.00
LR

LR

0.80 0.80

0.60 0.60

0.40 LS 0.40 LS

0.20
DL 0.20 DL
O O
0.00 0.00
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Sa(T1) [m/s2] Sa(T1) [m/s2]

Figure 26. The median vulnerability curves and LSs.

Type A Type B
0.12 0.12

0.10 0.10

0.08 0.08

0.06 0.06

0.04 0.04

0.02 0.02

0.00 0.00
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Loss Ratio Loss Ratio

Figure 27. Median and quantiles (16% and 84%) loss curves for the two configurations of the case
study.

Computed the abovementioned variables, the estimation of the whole loss curve (Fig. 27)
and the EAL of the structural elements can be completed (Table 10).
According to approach (b) mentioned in Section 2.2, as an alternative to the computa-
tion of such rigorous loss curve (Fig. 27), it can be discretized with a piecewise linear curve
identified by θPLs and LR,PLs (Fig. 29). To this aim, three intermediate PLs (operational,
52 D. OTTONELLI ET AL.

damage limitation and life safety) are considered in the transition region among frequent,
low-intensity, and high-intensity events (Welch, Sullivan, and Calvi 2014). The necessary
data θPLs and LR,PL for the construction of the PLs loss curve are listed in the following
table in terms of median values. With these values the EALPLs of the structural elements
can be computed (Table 10).
From the data of Table 9, it is possible to create the consequence functions for the two
case studies analyzed. Figure 28 shows as the obtained data are consistent with some
empirical values of literature (Di Pasquale and Goretti 2001; Ministry Decree n.65 7/3/
2017 2017).
In the case studies, the difference in the EAL is essentially given by the occurrence of
the initial PLs that revealed to be very significant in the loss evaluation also in the case of
URM buildings. Naturally, in the case of Type B configuration with r.c. ring beams, the
θZL is lower than Type A, because it is stronger (although less ductile). The final result of
the two case studies is however comparable despite the initial difference: this means that
Type A with the severe performance levels recovers the initial losses. The data of damage
from past earthquake observations and also results from numerical seismic analysis would
have suggested a more important difference between the damage and consequently losses
of the case studies. This is not occurred because of the specific characteristics of this

Table 9. λLSs and LR,LS for the construction of the loss curve with the PLs.
LSs
ZL O DL LS NC
λLS LR,LS,50% λLS LR,LS,50% λLS LR,LS,50% λLS LR,LS,50% λLS LR,LS,50%
Type A 0.153 0 0.026 0.085 0.010 0.175 0.003 0.363 0.001 1
Type B 0.101 0 0.033 0.052 0.009 0.153 0.002 0.326 0.001 1

Table 10. Values of EALSTRUCTURAL considering the entire hazard curve and the curve with PLs.
Configurations of case study EAL EALPLs
Type A 1.218% 1.243%
Type B 0.728% 0.755%

1.00
Di Pasquale and Goretti, 2001
Cr

M.D. n.65 7/3/2017


0.80 Type A
Type B
0.60

0.40

0.20

0.00
Zero Loss Operational Damage Life safety Near
Limitation Collapse
LS

Figure 28. Comparison between the median data from the two case studies and the empirical data
from Di Pasquale and Goretti (2001; Ministry Decree n.65 7/3/2017 2017).
JOURNAL OF EARTHQUAKE ENGINEERING 53

Type A Type B
0.16
0.12

0.14 0.10
0.12
0.08
0.10
0.08 0.06

0.06 0.04
0.04
0.02
0.02
0.00 0.00
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Loss Ratio Loss Ratio

Figure 29. Median loss curves with the identification of the LSs for the two case studies (the black line
is the rigorous curve instead the red line is the one with the PLs).

building and the properties assigned to the tie-rods and ring beams, that influence these
different seismic responses.
As regards the loss assessment procedure for structural elements due to the out-of-
plane, the values of the maximum peak ground acceleration compatible with the fulfilment
of a given DL have been computed. From these values, the mean annual frequency of
exceedance (Equation 15) of each DL was evaluated in order to construe the loss curves
(LR,DL – λDL) (see Fig. 30 and Table 11) where the losses are LR,DL2 = 0 and LR,DL4 = Cm,
quantified as a fraction ζm of the global loss value LR,GLOBAL.
The values of Cm were determined assuming: the reconstruction cost of the wall equal
to 350 €/m3 and the cost of an opening equal to 300 €. The global loss value was defined as
278400 €, only for the structural elements. So for the different macroelements and

Figure 30. The loss curves of the local mechanisms considered.

Table 11. Input for the loss curves of the local mechanisms.
Mechanism Macroelement Cm ζm λDL,2 λDL,4
1 P2 4500 € 0.016 0.1417 0.0013
P4 4230 € 0.015 0.1670 0.0013
2 P2 10365 € 0.037 0.1153 0.0012
P4 9535 € 0.034 0.1153 0.0012
54 D. OTTONELLI ET AL.

Table 12. EALs of the local mechanisms.


Mechanism Macroelement EALm
1 P2 0.116%
P4 0.128%

mechanisms analyzed, the values of Cm vary as a function of the volume and the number
of openings (Table 11).
From the values of Table 11, the EAL of the local mechanism more vulnerable
(mechanism 1) is computed (Table 12); it is assumed as the most likely to occur.
Unlike the structural elements, for which the economic losses have been obtained
directly from an analytical procedure function, for the non-structural ones the conse-
quence functions were taken from literature. In the inventory of the case studies, the
partitions are considered as non-structural element. As above-mentioned, they are con-
sidered as drift sensitive elements (FEMA P58-1 2012; Taghavi and Miranda 2003) and the
reference work herein adopted for the definition of the EDP capacity and for the
consequence functions for these elements is that of Cardone and Perrone (2015). In this
research work, the authors analyzed some results from previous experimental tests on
laboratory specimens of r.c./steel frames with masonry infills, in order to develop: i)
fragility functions as a function of attained peak interstory drift addressed to the estima-
tion of damage in typical non-structural component; ii) loss functions for such non-
structural components. It is highlighted that the behavior of the partitions can be in some
extent comparable to that of infills as assumed in Cardone and Perrone (2015); other
works (Chiozzi and Miranda 2017), thanks to the analysis of a large dataset comprising
152 masonry infill, highlighted a negligible influence of the brick type while conversely
a significant influence of the compressive strength of mortar on the infill behavior, with
overall values comparable to those proposed in Cardone and Perrone (2015). In the future,
targeted experimental tests on URM partition walls could corroborate and improve the
computation presented in the case study examined in this paper. The total repair cost is
below 1/2 the construction costs at DL1 and DL2, while it results larger than the
construction cost of a new panel with same characteristics at DL3. From the NLDA for
each time-history an EDP demand value is recorded. These values form a set of points
(IM = Sa(T1), EDP) that are fitted with a lognormal distribution in order to compute the
median, the 16% and 84% quantiles and the dispersion θD. In particular, considering the
median values for each IM and level, the following numerical curves have been defined
(solid lines in Fig. 31). Starting from the latter ones, the IM-EDP curve (dashed lines in
Fig. 31) can be constructed by a linear regression in log-log coordinates with Equation 14
and the following values of a and b (Table 13).
After the definition of the IM-EDP curves, the capacity of the non-structural elements is
evaluated, introducing the reference values of Cardone and Perrone (2015). Finally, the
mean annual frequency of exceedance of each DL is computed through the SAC/FEMA
equation (Eq 16) and shown in Table 14.
By associating to each DL (identified through the corresponding θDL) a relative loss
ratio (as defined in Table 14) the loss curve of the partitions can be drawn (Fig. 32). From
these curves, by taking the area under them, the EALNON-STRUCTURAL of the different
configurations of partitions is evaluated (Table 15).
JOURNAL OF EARTHQUAKE ENGINEERING 55

Figure 31. IDA curves (dashed lines) that represent IM-EDP demand relations.

Table 13. Values of a and b of Equation (14) for each floor of the two case studies.
Type A Type B
1 Level 2 Level 3 Level 1 Level 2 Level 3 Level
a 2.27E-05 2.20E-04 3.42E-04 2.46E-05 2.38E-04 2.67E-04
b 2.0275 1.0678 0.9645 2.05 1 1.009

Table 14. Values of λDL of the partitions.


λDL
Configurations of case study DS1 DS2 DS3
Type A 1 – without openings 1.39E-02 4.08E-04 9.77E-05
2 – with openings 5.23E-02 7.54E-03 2.73E-04
Type B 1 – without openings 8.77E-03 4.86E-04 1.20E-04
2 – with openings 3.37E-02 4.75E-03 1.18E-03

Figure 32. Loss curve with structural elements and partitions.

In order to determine the global EAL of the building (Table 16) the EALSTRUCTURAL
(Table 10), EALNON-STRUCTURAL (Table 15) and EALMECHANISMS (Table 12) should be
properly weighted through the values summarized in Table 8. The resulted values are
consistent with those reported in literature. For example, in Bothara et al. (2007), the EAL
56 D. OTTONELLI ET AL.

Table 15. Values of EAL for the non-structural elements.


Configurations of case study Configurations of partitions EALNON-STRUCTURAL
Type A 1 0.800%
2 2.582%
Type B 1 0.615%
2 1.962%

Table 16. Values of the final EAL.


Configurations of case study Configurations of partitions EALSTRUCTURAL EALMECHANISMS EALNON-STRUCTURAL EAL
Type A with local mechanism 1 1.179% 0.243% 0.025% 1.448%
2 0.082% 1.504%
Type A 1 1.179% - 0.025% 1.205%
2 0.082% 1.261%
Type B 1 0.705% - 0.019% 0.725%
2 0.064% 0.769%

of typical New Zealand URM houses is evaluated equal to 0.88% while in O’Reilly et al.
(2018) for an Italian school building located in Cassino the EAL resulted equal to 0.48%.

5. Concluding Remarks and Further Developments


Within the context of seismic risk analyses, loss assessment still represents the weakest point
because in the past the attention was substantially focused to the hazard and vulnerability
models. With the attempt of covering this gap, the paper faces the issue of loss computation in
masonry buildings. To this aim, a building specific analytical procedure has been developed,
following the FEMA P-58 methodology and differentiating it for each individual component, in
particular structural and non-structural parts. Although the procedure is based on the execution
of nonlinear dynamic analysis, it could constitute the basis for the proposal and calibration of
a simplified practice-oriented method, based on the direct use of the annual probability of
occurrence of all PLs and the related loss ratios. This is relevant at the regional scale.
Even if, much progress has been made in the loss evaluation of URM buildings with this
procedure, there are many issues that require future research activities. In particular, it would be
interesting to perform an extended parametric analysis on other prototype buildings, represen-
tative of further classes with homogeneous seismic behavior in order to establish other PLs and
repair cost relationships, on which the consequence functions are based. In that way, results
could support also assessment at the regional scale. In fact, the availability of cost functions for
buildings characterized by different collapse mechanisms and damage distribution at the
element scale would be a useful reference for the literature. Furthermore, the inventory and
loss procedure could be enriched in the future with different typologies of costs and sources of
consequences. As far the costs concern, possible developments involve, for example the
technical costs that have been revealed so significant after the reconstruction of L’Aquila (Del
Vecchio et al. 2018); or the costs of different type of materials to be adopted in modern
reconstruction of partitions including, for example, the solutions addressed to improve the
acoustic insulation. As far as the consequences concern, these additional contributions could be
surely deepened: the losses deriving from different types of equipment, as the electric and
hydraulic ones due to the damaged partitions; the indirect economic losses; and finally, the
downtime.
JOURNAL OF EARTHQUAKE ENGINEERING 57

Acknowledgments
The results were achieved in the national research project ReLUIS III 2014-2018 (www.reluis.it),
supported by the Italian Civil Protection Agency within the Line 7 – Displacement Based Assessment
of Existing Buildings (Coordinators: G.M. Calvi, T. Sullivan, and R. Monteiro).

ORCID
Daria Ottonelli http://orcid.org/0000-0001-7220-2013
Serena Cattari http://orcid.org/0000-0001-9459-5989
Sergio Lagomarsino http://orcid.org/0000-0002-6597-3474

References
Abruzzo. 2001. Servizio Tecnico Regionale dei LL.PP. Prezzario Regione Abruzzo - “Edizione 2011”.
(in Italian).
Akhoundi, F., G. Vasconcelos, and P. Lourenço. 2018. In-plane behaviour of infills using glass fiber
shear connectors in textile reinforced mortar (TRM) technique. International Journal of
Structural Glass and Advanced Materials Research 2: 1–14. doi: 10.3844/sgamrsp.2018.1.14.
Anthoine, A., G. Magonette, and G. Magenes. 1995. Shear compression testing and analysis of brick
masonry walls. 10th European Conference on Earthquake Engineering, Vienna, Austria.
ASCE/SEI 41-13. 2014. Seismic evaluation and retrofit of existing buildings. Reston, VA:American
Society of Civil Engineers. ISBN 978-0-7844-7791-5.
Beyer, K., and A. Dazio. 2012. Quasi static cyclic tests on masonry spandrels. Earthquake Spectra 28
(3): 907–29. doi: 10.1193/1.4000063.
Bothara, J. K., J. B. Mander, R. P. Dhakal, R. K. Khare, and M. M. Maniyar. 2007. Seismic
performance and financial risk of masonry house. ISET Journal of Earthquake Technology,
Paper No. 493 44 (3–4): 421–444.
Brignola, A., S. Pampanin, and S. Podestà. 2012. Experimental evaluation of the in-plane stiffness of
timber diaphragm. Earthquake Spectra 28 (4): 1687–709. doi: 10.1193/1.4000088.
Calvi, G. M., T. J. Sullivan, and D. P. Welch. 2014. A seismic performance classification framework
to provide increased seismic resilience. In Perspectives on European earthquake engineering and
seismology, ed. A. Ansal, vol. 34, 361–400. Cham, Switzerland: Springer International Publishing.
Calvi, G. M., and T. J. Sullivan. 2014. Estimating floor spectra in multiple degree of freedom
systems. Earthquakes and Structures 7 (1): 17–38. doi: 10.12989/eas.2014.7.1.017.
Cardone, D., and G. Perrone. 2015. Developing fragility curves and loss functions for masonry infill
walls. Earthquakes and Structures 9 (1): 257–79. doi: 10.12989/eas.2015.9.1.257.
Cattari, S., D. Camilletti, S. Lagomarsino, S. Bracchi, M. Rota, and A. Penna. 2018. Masonry Italian
code-conforming buildings. Part 2: Nonlinear modelling and time-history analysis. Journal of
Earthquake Engineering 22 (sup2): 2010–40. doi: 10.1080/13632469.2018.1541030.
Chiozzi, A., and E. Miranda. 2017. Fragility functions for masonry infill walls with in-plane loading.
Earthquake Engineering and Structural Dynamics 46 (15): 2831–50. doi: 10.1002/eqe.2934.
CNR-DT 212/2013. 2014. Istruzioni per la valutazione affidabilistica della sicurezza sismica di edifici
esistenti. Rome, Italy:Consiglio Nazionale delle Ricerche. (in italian).
Cornell, C. A., F. Jaylayer, R. O. Hamburger, and D. A. Foutch. 2002. Probabilistic basis for 2000
SAC federal emergency management agency steel moment frame guidelines. Journal of Structural
Engineering 128 (4): 526–53. doi: 10.1061/(ASCE)0733-9445(2002)128:4(526).
D’Ayala, D., and E. Speranza. 2003. Definition of collapse mechanisms and seismic vulnerability of
historic masonry buildings. Earthquake Spectra 19 (3): 479–509. doi: 10.1193/1.1599896.
Degli Abbati, S., S. Cattari, and S. Lagomarsino. 2018. Theoretically-based and practice-oriented
formulations for the floor spectra evaluation. Earthquake and Structures 5: 565–81. doi: 10.12989/
eas.2018.15.5.565.
58 D. OTTONELLI ET AL.

Del Vecchio, C., M. Di Ludovico, S. Pampanin, and A. Prota. 2018. Repair costs of existing RC
buildings damaged by the L'Aquila earthquake and comparison with FEMA P-58 predictions.
Earthquake Spectra 34 (1): 237–63. doi: 10.1193/122916EQS257M.
Di Ludovico, M., A. Prota, C. Moroni, G. Manfredi, and M. Dolce. 2017a. Reconstruction process of
damaged residential buildings outside historical centres after the L’Aquila earthquake: Part I
—”light damage” reconstruction. Bulletin of Earthquake Engineering 15 (2): 667–92. doi: 10.1007/
s10518-016-9877-8.
Di Ludovico, M., A. Prota, C. Moroni, G. Manfredi, and M. Dolce. 2017b. Reconstruction process of
damaged residential buildings outside historical centres after the L’Aquila earthquake: Part II
—“heavy damage” reconstruction. Bulletin of Earthquake Engineering 15 (2): 693–729. doi:
10.1007/s10518-016-9979-3.
Di Pasquale, G., and A. Goretti (2001] “Vulnerabilità Funzionale ed Economica negli Edifici
Residenziali Colpiti da Recenti Eventi Sismici Nazionali,” X Convegno Nazionale L’Ingegneria
Sismica in Italia, Potenza-Matera, Italy (in Italian).
Dolce, M., and G. Manfedi. 2015. Libro bianco sulla ricostruzione privata fuori dai centri storici nei
comuni colpiti dal sisma dell’Abruzzo del 6 aprile 2009. ISBN 978-88-89972-50-2, ReLUIS,
FINTECNA, CINEAS (in Italian).
EC8-3. 2005. Eurocode 8: Design of structures for earthquake resistance – Part 3:Assessment and
retrofitting of buildings. Brussels, Belgium: Comité Européen de Normalisation.
Elenco Regionale dei Prezzi delle Opere Pubbliche – Regione Emilia-Romagna – “Edizione 2015. (in
italian).
Fajfar, P. 2000. A non linear analysis method for performance-based seismic design. Earthquake
Spectra 16 (3): 573–91. doi: 10.1193/1.1586128.
FEMA P58-1. 2012. Seismic performance assessment of buildings: Volume 1 – Methodology, FEMA
P-58-1. Prepared by the Applied Technology Council for the Federal Emergency Management
Agency, Washington, DC.
Freeman, S. A. 1998. The capacity spectrum method as a tool for seismic design. 11th European
Conference of Earthquake Engineering, Paris, France.
Giongo, I., A. Wilson, D. Y. Dizhur, H. Derakhshan, R. Tomasi, M. C. Griffith, P. Quenneville, and
J. M. Ingham. 2014. Detailed seismic assessment and improvement procedure for vintage flexible
timber diaphragms. Bulletin of the New Zealand Society for Earthquake Engineering 47 (2):
97–118. doi: 10.5459/bnzsee.47.2.97-118.
Graziotti, F., G. Magenes, and A. Penna (2012] “Experimental cyclic behaviour of stone masonry
spandrels,” 15th World Conference on Earthquake Engineering, Lisboa, Portugal.
Kržan, M., S. Gostič, S. Cattari, and V. Bosiljkov. 2015. Acquiring reference parameters of masonry
for the structural performance analysis of historical buildings. Bulletin of Earthquake Engineering
13 (1): 203–36. doi: 10.1007/s10518-014-9686-x.
Lagomarsino, S. 2015. Seismic assessment of rocking masonry structures. Bulletin of Earthquake
Engineering 13 (1): 97–128. doi: 10.1007/s10518-014-9609-x.
Lagomarsino, S., and S. Cattari. 2015b. Seismic performance of historical masonry structures
through pushover and nonlinear dynamic analyses. In Perspectives on European earthquake
engineering and seismology, ed. A. Ansal, Vol. 39, 265–99. doi:10.1007/978-3-319-16964-4_11.
Lagomarsino, S., A. Penna, A. Galasco, and S. Cattari. 2013. TREMURI program: An equivalent
frame model for the nonlinear seismic analysis of masonry buildings. Engineering Structures 56:
1787–99. doi: 10.1016/J.ENGSTRUCT.2013.08.002.
Lagomarsino, S., and S. Cattari. 2015a. PERPETUATE guidelines for seismic performance-based
assessment of cultural heritage masonry structures. Bulletin of Earthquake Engineering 13 (1):
13–47. doi: 10.1007/s10518-014-9674-1.
Ministry Decree n.65 7/3/2017. 2017. Ministerial Decree n.65 of 27 February 2017, integrated with
the changes of n.58 of 7 March 2017. Guidelines for the seismic risk classification of constructions.
Rome, Italy: Italian Ministry of Infrastructures and Trasport. (In Italian).
MIT. 2009. Circolare 02/02/2009 n. 617 (2009). “Istruzioni per l’applicazione delle “Nuove norme
tecniche per le costruzioni” di cui al D.M. 14 gennaio 2008.” M. d. I. e. d. Trasporti, ed.Gazzetta
Ufficiale n. 47 del 26 febbraio 2009.
JOURNAL OF EARTHQUAKE ENGINEERING 59

Mitrani-Reiser, J., and J. Beck (2007] “An ounce of prevention: Probabilistic loss estimation for
performance-based earthquake engineering,” PhD Thesis, CalTech, USA.
Moehle, J. P., and G. G. Deirelein. 2004. A framework methodology for performance-based earth-
quake engineering. 13th World Conference on Earthquake Engineering, Vancouver, BC, Canada.
August 1–6, Paper No. 679.
Morandi, P., L. Albanesi, F. Graziotti, T. Li Piani, A. Penna, and G. Magenes. 2018b. Development
of a dataset on the in-plane experimental response of URM piers with bricks and blocks.
Construction & Building Materials 190: 593–611. doi: 10.1016/j.conbuildmat.2018.09.070.
Morandi, P., S. Hak, and G. Magenes. 2018a. Performance-based interpretation of in-plane cyclic
tests on RC frames with strong masonry infills. Engineering Structures 156: 503–21. doi: 10.1016/j.
engstruct.2017.11.058.
O’Reilly, G. J., D. Perrone, M. Fox, R. Monteiro, and A. Filiatrault. 2018. Seismic assessment and
loss estimation of existing school buildings in Italy. Engineering Structures 168: 142–62. doi:
10.1016/j.engstruct.2018.04.056.
O’Reilly, G. J., and G. M. Calvi. 2019. Conceptual seismic design in performance-based earthquake
engineering. Earthquake Engineering & Structural Dynamics 48 (4): 389–411. doi: 10.1002/
eqe.3141.
O’Reilly, G. J., R. Monteiro, A. M. Bellah Nafeh, T. J. Sullivan, and G. M. Calvi. 2019. Displacement-
based framework for simplified seismic loss assessment. Journal of Earthquake Engineering 1–22.
doi: 10.1080/13632469.2020.1730272
Ottonelli, D. 2016. Seismic performance assessment of unreinforced masonry buildings: The loss
calculation. Ph.D. Thesis, University of Genova, Genova, Italy.
Petry, S., and K. Beyer. 2014. Influence of boundary conditions and size effect on the drift capacity
of URM walls. Engineering Structures 65: 76–88. doi: 10.1016/J.ENGSTRUCT.2014.01.048.
Porter, K. A. 2003. An overview of PEER’s performance-based earthquake engineering methodol-
ogy. Ninth International Conference on Applications of Statistics and Probability in Engineering,
San Francisco, USA.
Porter, K. A., J. L. Beck, and R. Shaikhutdinov. 2004. Simplified estimation of economic seismic risk
for buildings. Earthq Spectra 20 (4): 1239–63. doi: 10.1193/1.1809129.
Sassun, K., T. J. Sullivan, P. Morandi, and D. Cardone. 2016. Characterising the in-plane seismic
performance of infill masonry. Bulletin of the New Zealand Society for Earthquake Engineering 49
(1): 98–115. doi: 10.5459/bnzsee.49.1.98-115.
Solberg, K. M., R. P. Dhakal, J. B. Mander, and B. A. Bradley. 2008. Computational and rapid
expected annual loss estimation methodologies for structures. Earthquake Engineering and
Structural Dynamics 37 (1): 81–101. doi: 10.1002/EQE.746.
Taghavi, S., and E. Miranda 2003. Response assessment of nonstructural building elements. PEER
Report 2003/05, Pacific Earthquake Engineering Research Center, College of Engineering
University of California Berkeley.
Tomaževic, M. 1999. Earthquake -resistant design of masonry buildings. In Series on innovation in
structures and construction, vol. 1, 268. London, UK: Imperial College Press.
Vamvatsikos, D. 2013. Derivation of new SAC/FEMA performance evaluation solutions with
second-order hazard approximation. Earthquake Engineering and Structural Dynamics 42 (8):
1171–88. doi: 10.1002/eqe.2265.
Vamvatsikos, D. 2014. Accurate application and Second-order improvement of SAC/FEMA prob-
abilistic formats for seismic performance assessment. ASCE Journal of Structural Engineering 140
(2): 04013058. doi: 10.1061/(ASCE)ST.1943-541X.0000774.
Vamvatsikos, D., and C. A. Cornell. 2002. Incremental dynamic analysis. Earthquake Engineering &
Structural Dynamics 31 (3): 491–514. doi: 10.1002/eqe.141.
Vanin, F., D. Zaganelli, A. Penna, and K. Beyer. 2017. Estimates for the stiffness, strength and drift
capacity of stone masonry walls based on 123 quasi-static cyclic tests reported in the literature.
Bulletin of Earthquake Engineering 15 (12): 5435–79. doi: 10.1007/s10518-017-0188-5.
Welch, D. P., T. J. Sullivan, and G. M. Calvi. 2014. Developing direct displacement-based proce-
dures for simplified loss assessment in performance-based earthquake engineering. Journal of
Earthquake Engineering 18 (2): 290–322. doi: 10.1080/13632469.2013.851046.

You might also like