Bulletin of Earthquake Engineering (2020) 18:3433–3457

https://doi.org/10.1007/s10518-020-00831-1

ORIGINAL RESEARCH

Seismic damage and fragility assessment of ancient masonry

churches located in central Chile

Nuria Chiara Palazzi1,2,3 · Philomene Favier4 · Luisa Rovero1 · Cristián Sandoval2 ·

Juan Carlos de la Llera2,3

Received: 26 July 2019 / Accepted: 21 March 2020 / Published online: 31 March 2020
© Springer Nature B.V. 2020

Abstract
An assessment of damage and seismic fragility of historical unreinforced masonry churches
located in Chile’s central valley was carried out employing the results of a field survey of
106 ancient churches, after the 2010 Maule earthquake. The observed damage was cor-
related with recurrent failure mechanisms of unreinforced masonry structures, by taking
into account 22 local mechanisms involving macro-elements on these churches. The aver-
age damage level suffered by each church was computed considering the global and local
behaviors of the structures through a damage index computed as a weighted mean of the
levels of damage observed for each mechanism. The results of this damage index method
are used to obtain Probability Mass Functions and suitable probabilistic tools are used to
propose Empirical Fragility Curves (EFCs) for these structures. The EFCs are directly
usable by stakeholders involved in risk assessment aimed to the prioritization of possible
future damage mitigation strategies and other decision making processes relative to this
historical heritage.

Keywords  Fragility curves · Masonry churches · Heritage built · Seismic fragility

assessment

1 Introduction

Past earthquake surveying activities after the 2010 Maule earthquake (Mw 8.8), demon-
strated that Chilean Built Heritage underwent extensive structural damage in particular, in
unreinforced masonry (URM) churches (D’Ayala and Benzoni 2012; Palazzi et al. 2019a,

* Nuria Chiara Palazzi

nuriachiara.palazzi@cigiden.cl
1
Department of Architecture, University of Florence, Piazza Brunelleschi, 50121 Florence, 6, Italy
2
Department of Structural and Geotechnical Engineering, Pontificia Universidad Católica de Chile,
Santiago, Chile
3
National Research Center for Integrated Natural Disaster Management CIGIDEN,
CONICYT/FONDAP/15110017, Santiago, Chile
4
University Grenoble Alpes, INRAE, 38402 Saint‑Martin‑d’Hères, LESSEM, France

13
Vol.:(0123456789)
 3434 Bulletin of Earthquake Engineering (2020) 18:3433–3457

b). Protection strategies for this heritage not only need studies using detailed analyses of
individual churches (Indirli et al. 2011; Rendel et al. 2014; Jorquera et al. 2016; Sandoval
et al. 2017; Torres et al. 2018; Palazzi et al. 2018), but also studies at urban and territorial
level. In fact, urban and/or territorial analyses of the seismic vulnerability and fragility of
the Built Heritage lead to action plans for risk mitigation by identifying a list of priorities
and procedures suitable for seismic emergency management (Braga et al. 2015; Bergami
and Nuti 2013; Staniscia et al. 2017).
For the last twenty years, significant research has contributed to the assessment of the
seismic vulnerability at a large geographical scale, as summarized elsewhere (Calvi et al.
2006). The first predictions in this field were based on the observation of post-earthquake
scenarios through the derivation of damage Probability Mass Functions (PMFs), usually
called Damage Probability Matrices (DPMs, Whitman et al. 1973; Braga et al. 1982) which
express in a discrete format the probability for the k-th element of reaching
( a given
) dam-
age level Dk = j , conditioned to a ground motion intensity IM, i.e. P Dk = j|IM  . By using
this information, continuous vulnerability and fragility functions, are proposed to describe
the probability of being in a given damage state conditioned to a specific intensity of the
seismic hazard (Orsini 1999; Singhal and Kiremidjian 2004; Rota et al. 2006; Martinelli
et  al. 2008; Rossetto et  al. 2013). A large number of procedures for seismic vulnerabil-
ity and fragility function construction can be found in the literature. Three big classes of
methods are usually employed: (a) empirical (e.g. Colombi et al. 2008; Rota et al. 2006;
Cara et al. 2018); (b) expert judgment based (e.g. Lagomarsino and Giovinazzi 2006); and
(c) analytically based on mechanical models (e.g. Kircher et al. 1997; Restrepo-Vélez and
Magenes 2004; Kalkbrenner et al. 2019). A fourth class, denoted as hybrid, is sometimes
considered when a combination of the three previous methods is used, say analytical and
empirical (Singhal and Kiremidjian 1998; Basaglia et al. 2018), or expert judgment based
and empirical (Kishor et al. 2011).
The Chilean territory is characterized by an Architectural Heritage with unique con-
structive and typological features, which has undergone significant ground motions in time.
Despite this high seismic hazard, no seismic fragility assessments of monumental build-
ings at the territorial level have been reported in the literature. The MARVASTO project
(Indirli et  al. 2006; MARVASTO 2007) and Jiménez et  al. (2018) are the only research
pieces focused on the evaluation of the seismic vulnerability of the urban area of Val-
paraiso, focused on a sample of monuments identified by UNESCO (i.e., Cerro Cordillera
and historical center of Valparaiso, respectively).
Consequently, this paper focuses on the seismic damage during the 2010 Maule earth-
quake of 106 URM churches located in the central Valley of Chile. A throrough post-seis-
mic structural survey using an abacus developed by Italian practice (DPCM 2011) was
used for each church typology. It should be noted that in this study, the stock of churches
was divided into three categories with homogeneous features: (1) Colonial (CL); (2) Neo-
classical style & Variants (NC&V); and (3) Neo-gothic (NG).
A macro-seismic method proposed earlier (Lagomarsino and Podestà 2004c) was
adapted and applied in order to estimate the damage index, id , for each church considered.
In Lagomarsino and Podestà (2004c), index id considers the possibility of the generation of
28 possible failure mechanisms, which are computed based on a weighted mean of the lev-
els of damage detected for each mechanism. Since values of the damage index are defined
during field inspection using real numbers, a transformation is needed into discrete vari-
ables, representative of the level of damage consistent with the European Macro-seismic
Scale (Grünthal 1998). Given the significant data collected for the 106 URM churches,
an empirical approach was carried out based on a probabilistic analysis of the observed

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3435

earthquake failures. Thus, Probability Mass Functions (PMFs) for global and local behav-
iors (De Matteis et al. 2016; Marotta et al. 2016), and Empirical Fragility Curves (EFCs)
are proposed herein for the Chilean URM churches.

2 Ancient unreinforced churches of Chile

Chilean Architectural Heritage exhibits distinctive architectural features due to the merge
of the European Architecture characteristics with the Chilean local constructive culture
generated during Spanish domination period (1536–1818). The local indigenous construc-
tive culture, well before the Spanish domination was aware of the seismic risk and many
structural concepts came from the period of Incan domination (1470–1530). Conversely,
Spanish constructive culture was unaware of seismic risk and was characterized mostly by
the European architectural revivalisms (i.e., Neo-Baroque, Neo-Classical, Neo-Renaissance
and Neo-Gothic). Consequently, this architectural heritage comprises a considerable vari-
ety of buildings with different characteristics, as well as buildings with different materials
and construction techniques. In particular, the use of adobe (i.e. earthen blocks) came from
the Incan culture, and the use of cyclope stone masonry, cob techniques (i.e. mix of earth,
straw and water), and the quincha technique (i.e. timber structure with earth and straw)
were inherited from indigenous architecture (Cancino et al. 2009; Cancino 2010; Fonseca
and D’Ayala 2012; Torrealva and Vicente 2013; Varum et al. 2014). Furthermore, the larg-
est number of historic buildings declared National Monuments in Chile is located mainly
in the Central part of the country, which comprises the regions of Valparaíso (V), Met-
ropolitan (Santiago), Libertador General Bernardo O’Higgins (VI), the Maule (VII), and
Biobío Region (VIII). A significant portion of this Heritage consists of URM structures.

2.1 Database of the 106 URM churches

Geometrical, constructive, and structural features were surveyed for a representative stock
of 106 churches of central Chile (Palazzi et al. 2019a, b), following the Global Earthquake
Model (GEM) Guidelines for empirical vulnerability assessment (Rossetto et  al. 2014),
which provides the basic characteristics of a high-quality empirical database.
The essential data categories included the: (1) number of investigated buildings rep-
resentative of the total population (almost the whole population was analyzed); (2) range
of peak ground acceleration (PGA) seismic intensity (0.15 g < PGA < 0.55 g); (3) location
(latitude and longitude); (4) height; (5) foot-print area; (6) construction material; and (7)
structural system of the church. Several direct and indirect sources have been used for the
compilation of the damage database following the Maule earthquake; reports gathered by
the authors using in-situ survey activities, and reports provided by architects, and engineers
of the Consejo de Monumentos Nacionales (CMN 2010) and the Episcopal Conference of
Chile (Episcopal Conference of Chile 2010).
Figure 1a shows schematically the location of the 106 surveyed URM churches in the
Metropolitan (RM) and Libertador General Bernardo O’Higgins (VI) regions. Figure  1b
the corresponding seismic zoning: zone1 (Z1), with maximum PGA, ­A0 = 0.2  g; zone2
(Z2) with maximum PGA, A ­ 0 = 0.3  g; and zone3 (Z3), with maximum PGA, A ­ 0 = 0.4  g
(DE 2010, MINVU 2011). All 2010 Maule earthquake PGAs were obtained from the
USGS Shake-Maps (USGS).

13
 3436 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Fig. 1  a Location of the 106 surveyed URM churches in the metropolitan region (RM) and Libertador Gen-
eral Bernardo O’Higgins (VI) region, classified into three categories with homogeneous characteristics:
Colonial (CL), Neo-classic & Variants (NC&V), and neo-gothic (NG); and b seismic zoning of the RM and
region VI (according to the DE 2010, MINVU 2011): zone 1 (Z1), effective acceleration ­A0 = 0.2 g; zone 2
(Z2) ­A0 = 0.3 g; and zone 3 (Z3), ­A0 = 0.4 g. The 2010 Maule earthquake PGAs are taken from the USGS
Shake Map (USGS; United States Geological Survey)

2.2 Classification of URM churches

A detailed analysis of the database was developed using the identification of parameters
that characterize the structures: masonry type, architectural layout, architectural style, and
foot-print area. The stock of churches was divided into three categories with homogeneous
characters: (1) Colonial (CL), (2) Neo-classical & Variants (NC&V); and (3) Neo-gothic
(NG).
In the analyzed area, 47 CL churches were identified (Fig.  1). These religious build-
ings represent a constructive variant of the original Colonial typology, i.e., the Northern
Andean typology, determined by different climatic conditions. The Colonial churches of
the Chilean Central Valley are characterized by a simple and austere design characterized
by a single nave with an elongated plan aspect ratio, a sloping timber roof with a beam and
wooden-collar-tie (“par and nudillo” in Spanish) traditional trusses (“tijeral” in Spanish), a
flat ceiling, buttresses in some cases, and an adobe or wooden bell-tower. An emblematic
example of this typology is the Loica parish (Fig. 2a) located in the San Pedro municipal-
ity, built in 1856 and declared a Historical Monument in 2009 (Recabarren 2009; Pinto
2011). The church was built with adobe masonry and has a wall thickness of 1.0 m with a
typical wooden bell-tower, which stands on a prothyrum of eight timber columns.
The churches of the NC&V style were also identified (Fig. 1a). Most of these monu-
ments, built in the transition period between the Colonial and Modern Age, are localized
in the City of Santiago (i.e. 75%). As opposed to the CL religious structures, churches
of the Neo-Classical& Variants are designed with greater freedom and resourceful-
ness (Bahamondez et  al. 2012), which is typical of this epochal change. As an exam-
ple, a Basilica layout (rarely Latin-cross, only in 8% of the cases) with a more complex
morphology than the original single nave was introduced. The new structural system
consists of a central nave higher than the lateral ones, two side aisles (in some cases
crossed by a transept), an apse, two bell towers, a sloping roof, and false vaults. This
is the case of the Santo Domingo church (Fig. 2b) located in the historic city centre of

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3437

Fig. 2  Church examples of Central Chile: a Loica church in San Pedro commune (Colonial); b Santo
Domingo church in downtown Santiago (Neo-Classic); and c Santa Filomena parish in Recoleta commune,
Santiago (Neo-Gothic)

Santiago, whose façade in neoclassical style was the work of Roman architect Joaquin
Toesca. The church was built of stone masonry with a wall thickness of 1.5  m and a
roof structure, which was originally made of timber king-post trusses with a collar tie.
The roof was replaced by steel elements in 1963. As observed earlier (Jorquera et  al.

13
 3438 Bulletin of Earthquake Engineering (2020) 18:3433–3457

2016, 2017a, b) some typological features are apparent, which are common with several
NC&V churches, such as the plan arrangement, the massive masonry walls, the two
bell-towers, and the false wooden vaults.
Additionally, Neo-Gothic (NG) churches of clay masonry have been identified in the
studied area (Fig. 1a). These large brick buildings are the transposition of the Backstein-
gotik (German Brick-Gothic) in a highly seismic framework. The Santa Filomena church
is a significant example of NG architectural style (Fig. 2c), whose complexity is due to the
presence of the following macro-elements: a façade bell-tower setting on a gable, a narthex,
three naves in which the lateral aisles are shorter than the central nave, generally crossed by
a transept, a semicircular apse, an ambulatory, and false rib vaults. The absence of effective
seismic resistant devices able to guarantee an integrated box-behavior of the structure, the
great conventional slenderness of the walls, and the lack of rigid roof diaphragms are the
seismically weakest components that characterize these structures.

3 Damage scenarios following the 2010 Maule earthquake

3.1 The 2010 Maule earthquake

On February 27, 2010 at 3.34 am, the Maule earthquake ­(Mw = 8.8) struck the central
region of Chile. This seismic event generated one of the most intense ground shak-
ing ever measured with peak horizontal and vertical ground accelerations greater than
0.6 g. The Maule megathrust earthquake with epicenter located at 35.846°S 72.719°W
(USGS) was produced by the interplate subductions phenomena at the convergence
zone between the Nazca and South American plates. The tsunami triggered by the
earthquake, left several coastal towns either devastated or heavily damage in the south
and central area of the country. An average of 12 Mio. people was impacted by the
earthquake and tsunami, leading to 524 deaths (Jünemann et al. 2015). The built herit-
age suffered significant losses and substantial damage, comprising an estimated of 290
Mio. dollars in repair costs (Conferencia Episcopal de Chile 2010). The adobe and brick
unreinforced masonry buildings built before 1940 were the most affected. In particular,
heavy structural damage was observed as shaking occurred in alluvial and fluvial soils.
For the V, VII, VIII, IX, and RM regions, researchers (Astroza et al. 2010) developed
the Macroseismic Medvedev-Sponheuer-Karnik Scale Intensities map for the Maule earth-
quake. In D’Ayala and Benzoni (2012), the European Macroseismic Scale (EMS-98) inten-
sity map was proposed for the cities of Santiago and Valparaíso, the O’Higgins region,
and the cities of Curicó and Talca in the Maule region, according to the seismic damage
classification used by the Chilean authority. The PGAs were recorded by the stations of the
National Seismological Network and by the RENADIC of the Faculty of Civil Engineer-
ing at the University of Chile (Boroschek et al. 2010a, b, 2012; Decanini et al. 2012). Due
to the low number of records in the vicinity of the investigated churches (i.e., distance less
than 2  km), and because of their high concentration in the city of Santiago, which does
not allow a sufficiently large intensity range of measurement levels, the PGAs were taken
from the USGS Shake Maps for 0.3  s, 1.0  s and 3.0  s in this research, as suggested by
GEM guidelines (Rossetto et al. 2013). A summary of the values of MSK, EMS’98, PGA,
PGV and A0 (maximum ground acceleration at T = 0 s for different seismic zones accord-
ing NCh433, INN 1996) values are presented in Table 1 for the different sites of interest.

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3439

Table 1  Value of MSK intensities (Astroza et al. 2010), EMS’98 intensities (D’Ayala and Benzoni 2012),
horizontal and vertical PGAs (Boroschek et al. 2010a, b and USGS), and A­ 0 (INN 1996) for different sites
of interest

City/town Lat°.mil; Lon°.mil Epicenter MSK PGA (g) PGA (g) vertical Seismic
distance horizon- zone ­A0
(km) tal (g)

Melipilla  − 33.687; − 71.214 284 6.5 0.343 0.279 III-0.4

Talagante  − 33.77622; − 70.9886 297 6.5 0.28 0.251 II-0.3
Santiago Centro  − 33.3404; − 70.6428 330 6.5 0.309 0.182 II-0.3
Santiago, Provi-  − 33.4314; − 70.6093 330 6.5 0.139 0.08 II-0.3
dencia
Codegua  − 34.0370; − 70.6729 280 6.0 0.274 0.249 II-0.3
Rancagua  − 34.166667; − 70.75 264 6.5 0.257 0.249 II-0.3
Rengo  − 34.4024; − 70.8674 238 7 0.248 0.225 II-0.3
Requínoa  − 34.2884; − 70.8158 250 6 0.302 0.166 II-0.3
Doñihue  − 34.2290; − 70.9579 247 7.5 0.354 0.343 II-0.3
San Vicente Tagua  − 34.3917; − 71.0848 222 7 0.326 0.384 II-0.3
Tagua
Peralillo  − 34.4773; − 71.4873 196 8 0.304 0.079 III-0.4
Pumanque  − 34.6027; − 71.6553 175 8 0.44 0.37 III-0.4
Lolol  − 34.7284; − 71.6458 165 7 0.382 0.223 III-0.4

3.2 Damage survey

The seismic performance of the URM churches depends on several parameters such as
the quality of masonry and the connection between orthogonal walls, the irregularity
of the in-plane and elevation arrangement, the presence of discontinuities in the struc-
tural system, the implementation of inadequate interventions, and the absence of a rigid
diaphragm. Nevertheless, in post seismic scenarios, the damage assessments presented
by numerous authors (Augusti et al. 2002; Lagomarsino et al. 2004; Lagomarsino and
Podestà 2004a, b, c; Lagomarsino 2012) have highlighted that the structural response
of churches exhibit recurrent patterns. In fact, the seismic action selects the most vul-
nerable building portions, called macro-elements, which present a rather independent
behavior relative to the global response of the structure (Giuffré 1991; Da Porto et al.
2010; Lagomarsino and Podestà 2004b).
Hence, the post-seismic damage scenario of Chilean churches after the 2010 Maule
earthquake was analyzed according to the feasible dominant behaviors of macro-ele-
ments of the church, i.e. architectural typology, façade, narthex, bell-tower, lateral
walls, transversal walls, colonnade, transept, apse, and chapels, using the catalogue of
mechanisms developed elsewhere (Doglioni 2000).
Following Giuffrè, 1989, the crack pattern was analyzed using the three fundamental
damage phenomena associated with well-known mechanisms: the zero mode (disaggre-
gation of masonry wall); the first mode (the out-of-plane behavior, OOP); and the sec-
ond mode (the in-plane behavior, IP) mechanisms. Shown in Fig. 3 are percentages of
the possible and activated mechanisms in the sample of churches considered.

13
 3440 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Fig. 3  Percentage of possible collapse mechanisms relative to the total sample—a possible collapse mecha-
nism is one not activated on a specific church by the earthquake, which is likely to occur in case of a higher
intensity motion—and the actual mechanisms activated during the 2010 earthquake (compared to the pos-
sible sample space).Chilean churches of the central region cannot develop 6 out of the 28 collapse mecha-
nisms (i.e. crack and/collapse vaults of central nave vaults, M8;side naves vaults, M9;transept, M12; apses,
M18; chapels, M24; and roof lantern mechanisms, M15)

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3441

Possible mechanisms are defined as the potential collapse modes that could be acti-
vated in the churches due to the presence of a macro-element associated with that type of
mechanism.
In particular, due to specific structural features of the Chilean churches, the 28 analyzed
collapse mechanisms were assessed using the Italian post-earthquake survey form. Results
led to 22 mechanisms (Fig.  3) present in the survey. In the proposed methodology, the
excluded damage mechanisms are consistent with the presence, or absence, of a specific
macro-element in the analyzed structure. In particular, Chilean churches of the central
region cannot develop some of the typical collapse mechanisms of URM vaults and roof
lantern, i.e., crack and/collapse vaults of central nave vaults, M8; side naves vaults, M9;
transept, M12; apses, M18; chapels, M24; and roof lantern mechanisms, M15.Such is the
case because these elements are built with light wooden elements. This analysis justifies
the elimination of these mechanisms in the analysis.
In the aftermath of the Maule earthquake, the most common mechanisms observed
in the churches are the activation of a simple overturning of the façade (M1, 83%), the
transversal vibration of the nave (M5, 83%), the shear mechanism on the nave lateral walls
(M6, 83%) and the bell tower mechanism (M27, 84%). The façade simple overturning
was typically evidenced by vertical cracks in the wall corners and the haunch of the trans-
verse arches of the narthex. The failure pattern, which is usually due to poor connections
between horizontal diaphragms and walls, varies according to the geometry of the façade.
A typical out-of-plane mechanism of the central part of façade was observed in the
San Francisco de Mostazal’s church (PGA = 0.287 g) built in 1858 in the municipality of
Rancagua. Deep vertical cracks were observed due to the lack of connection between the
façade, the longitudinal nave walls and the bell-tower (Fig. 4a).
In the San Agustín’s church (PGA = 0.343 g), located in the town of Melipilla and built
in 1744, the first mode mechanism of the façade was triggered. Different materials in the
façade (i.e., brick masonry) and the side walls (i.e., adobe) were a critical link between
structural elements. Maule’s shaking produced detachment of the entire façade with 7 m
wide and 1.5 m height (Fig. 4a) and its stucco (Lira and Arévalo 2010).
The presence of good quality connections between the façade and lateral walls of the
side aisles limited the collapse of the façade macro-element at the upper part (i.e. gable).
In several Colonial churches, the collapse of tympanums was observed. Its reconstruction
with wooden elements and brick (“tabiqueria” in Spanish), and the lack of a good con-
nection between the roof and façade represent an important discontinuity and weakness to
present an overturning of the wall.
That is the case for Rancagua’s Cathedral (PGA = 0.257 g), Nuestra Señora de la Merced
parish in Doñihue (PGA = 0.354 g), and Nuestra Señora de la Merced church in Codegua
(PGA = 0.274 g), which presented collapse of the façade’s gable (Fig. 3b). Moreover, the
San Saturnino church (PGA = 0.378 g) and the Salvador basilica (PGA = 0.337 g) are both
Neo-gothic structures, which were designed by the architect Teodoro Burchard at the end
of nineteenth century. Both structures have shown dangerous cracking with partial and total
collapse of the transverse arches and lateral walls at the base of columns as a consequence
of the in-plane response of the central nave and the side aisle arcades (Fig. 4c).
The activation of apse overturning was also frequently observed in Neo-gothic and
Revivalist churches (Fig.  5a), such as the San Francisco’s church (PGA = 0.276  g) in
the city of San Fernando built by Jesuit missionaries in 1744 and declared a National
Monument in 1984, as well as the Salvador basilica. Walls of the side aisles show wide
openings in the proximity of the wall ends, the hammering roof, and the lack of linkages
among wooden trusses and masonry walls triggered first mode mechanisms with deep

13
 3442 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Fig. 4  a Out-of-plane mechanisms of the façade due to poor connections at the corners: in the San Fran-
cisco de Mostazal the simple overturning involved the central part of façade, and in the San Agustín
church the complete façade; b gable overturning effect of inadequate connection between roof structure
and masonry wall of the upper part of façade on Cathedral of Rancagua, Doñihue’s parish, and Codegua’s
church; and c observed collapses on the haunch of the transverse arches of side aisles

horizontal cracks and partial collapse in parishes Nuestra Señora del Carmen in Oli-
var, in San Juan Tadeo in Malloa (PGA = 0.264 g) and in San Francisco Javier of Per-
alillo (PGA = 0.419  g), which west nave completely collapsed (Fig.  5b). Other typical
localized failures were diagonal cracks and total collapse of the bell-gables. Diagonal
cracks were produced as a result of shear failure on walls, usually at the bell-tower of
Neo-gothic churches such as the Santísimo Sacramento (PGA = 0.378 g) and San José
churches (PGA = 0.341 g) in Santiago, (Fig. 5c). Total collapse of bell-gables was due
to out-of-plane behavior, which is recurrent in Colonial churches.
Parishes San Andrés in Ciruelos, San Nicodemo in Coinco (PGA = 0.352 g), Nuestra
señora del Rosario in Guacarhue, San Andrés in Pichilemu (PGA = 0.531 g), are some
examples of CL churches that underwent total bell-tower collapse.

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3443

Fig. 5  a Vertical cracks in windows due to the hammering roof covering; b deep horizontal cracks windows
and buttresses, and total collapse of the lateral wall due to the hammering roof and connection between the
wooden trusses and the masonry walls; and c diagonal cracks on bell-tower walls

3.3 Damage indices

Herein, the macro-seismic method (Lagomarsino and Podestà 2004b; Lagomarsino et al.
2004) for estimating a damage index, id , was implemented in each church. This damage
index considers the possibility of generating various possible failure mechanisms (m = 28
mechanisms) by using a standardized mean of the weighted level of damage detected for
each possible mechanism computed by the following equation:
mt ( ) mt
1 ∑ 𝜌k 1 ∑
id = ⋅ ⋅ dk = ⋅ 𝜌̂ ⋅ d (1)
5 k=1 𝜌 5 k=1 k k

here 𝜌̂k is the normalized score of the participation of each mechanism on the global behav-
∑mt
ior of the structure, and its value ranges between 0 and 1, while the sum k=1 𝜌̂k = 1 ; and
dk is the damage score that considers the k-th mechanism, which value ranges from 0 to
5. Only mt = 22 of the 28 possible damage mechanisms defined for the Italian church
case were considered in Chilean URM churches. Such is the case because mechanisms of

13
 3444 Bulletin of Earthquake Engineering (2020) 18:3433–3457

macro-elements that were not activated in the buildings were excluded from the analysis
(Fig. 6).
Since the values of the damage index are defined during the field inspection using real
numbers, a transformation of the indices into discrete variables is carried out to obtain a meas-
urable level of damage comparable to the European Macro-seismic Scale (Grünthal 1998).

Fig. 6  Damage levels according to Table 2 (Grünthal 1998) and global damage index (Lagomarsino et al.
2004; Lagomarsino and Podestà 2004b) for the 106 URM churches of Central Chile affected by the 2010
Maule earthquake

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3445

Thus, each damage index associated with a ­Dk damage level was considered as suggested else-
where (Lagomarsino and Podestà 2004b; De Matteis et al. 2016), by classifying the damage in
six levels according to the EMS-1998 scale (Table 2).
In Fig.  6, the results obtained by the damage index method, show better correlation in
stone masonry churches (7 in totals) rather than brick and adobe structures. Indeed, during the
Maule earthquake, CL and NG churches were the most damaged structures.
Several of CL churches, of which 57.4% were located in region VI (Libertador General
Bernardo O’Higgins), underwent heavy structural damage. Extensive and deep cracks were
observed activating local failure modes of different macro-elements, and in some cases, even
total or near total collapse mechanisms (65.9% of global damage levels in Colonial churches
reached ­D4 − D5, id > 0.6). The NG buildings located in the city of Santiago 325 km from the
epicenter exhibited activation of failures of numerous macro-elements and a prevalent damage
level Dk in 66.6% of the cases. Compared to the previously analyzed buildings, the NC&V
churches showed better structural performance, with moderate structural and non-structural
damage in 75% of the cases. On the other hand, stone churches underwenton a more limited
range of PGA (range between 0.25 and 0.3 g).

3.4 Probability Mass Functions (PMF)

Probability Mass Functions (PMFs) of seismic damage level (Whitman et  al. 1973; Dolce
et al. 2006; Di Pasquale et al. 2005; De Matteis et al. 2016; De Matteis and Zizi 2019) were
derived from the empirical data of 106 URM churches described above for seismic intensities
ranging from 0.16 g < PGA ≤ 0.28 g, 0.28 g < PGA ≤ 0.41 g and 0.41 g < PGA ≤ 0.53 g. The
limits on these brackets are based only on the clustering of the information and the existence
of significant differences on the corresponding states of damage. The PMF expresses the prob-
ability of getting a damage level ( k due to a) ground motion with intensity measure IM, and is
expressed mathematically as P Dk = j|IM  . These PMFs are shown in Fig. 7 and consider the
damage levels derived from Eq. (1).
In the literature, the Binomial distribution is widely used to fit data to PFMs (e.g., Lago-
marsino and Podestà 2004b; Di Pasquale et  al. 2005; De Matteis et  al. 2016, Brando et  al.
2017; De Matteis and Zizi 2019). Theoretically, the Binomial distribution is a discrete prob-
ability distribution simulating binomial outputs, usually expressed as “success” or “failure”.
Indeed, it models the number of “successes”—e.g., no damage when an earthquake occurs—
and its complementary “failures” (damage), in a sequence of n independent trials, say in
this case, for n different churches struck by the same earthquake intensity. As a generaliza-
tion of the Binomial distribution, the Multinomial one is capable of modeling the number of
outcomes among n trials that fall into the i-th class among k predetermined classes. In this
research, the Multinomial distribution models the number of churches struck by the same
earthquake intensity that fall in either of 6 damage levels. Therefore, as long as the needed
outcomes may fall into multiple categories greater than two, this distribution is the most
appropriate for statistical purposes.
Therefore, among n churches, the number that falls in each of the 6 damage levels, denoted
as x0 , x1 , x2 , x3 , x4 , x5 , follows a Multinomial distribution, which is defined as:
( ) n! x x x
f x0 , … , x5 |n, p0 , … , p5 = p 0 p 1 … p55 ,
x0 ! … x5 ! 0 1 (2)

where pi is the probability to lie in the i-th damage level. The calculated mean ratio 𝜇D
(Eq.  3), computed as the ratio between Dk,i (the number of churches that reached the

13
 3446 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Table 1  Damage classification proposed by (Lagomarsino and Podestà 2004b; De Matteis et  al. 2016),
according to EMS-1998 scale, and description of damages

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3447

Table 1  (continued)
Reaction condition: Hematite@NH2/Pd (100 mg, 0.01 mol% of Pd, 6 ppm in 1.65 mL solvent), nitroarene
(0.4 mmol), 2 wt% TPGS-750-M/H2O (1.5 mL) and THF (0.15 mL), N ­ aBH4 (1.6 mmol) at room tempera-
ture
Yields are isolated
a
 Yields determined GC yields, octadecane was used as an internal standard
b
 NaBH4 (3.2 mmol)
c
 Nitroarene (0.2 mmol), ­NaBH4 (0.8 mmol), catalyst (100 mg, 0.038 mol% of Pd, 25 ppm in 1.65 mL), 2
wt% TPGS-750-M/H2O (1.2 mL) and THF (0.5 mL)
d
 [TON] values [(mol product/mol catalyst)]
e
 [TOF] values [(mol product/mol catalyst)/time of reaction (h)

Fig. 7  Probability Mass Functions (PMFs) for the whole sample using the observed data and the Multino-
mial distribution (MPDF) for ground motion intensities in the range from 0.16 g < PGA ≤ 0.53 g

denoted as damage level for intensity, IM = IMj ), and NJ the total number of observed
churches that underwent a given intensity, IM = IMj.
∑n
Dk,i
𝜇D = i=1 (3)
NJ

The only presence of a macro-element (façade, triumphal arch, bell tower, etc.) is
enough to make the activation of an associated collapse mechanism (event) possible. Thus,
each potential failure mode that could be activated in a church, i.e. a possible mechanism
[PM] even if it is not present, has a ´probability that is computed as the ratio between the
number of churches for which the mechanism was activated during the 2010 earthquake,
and the total number of possible mechanisms. PMFs for the out-of-plane and in-plane

13
 3448 Bulletin of Earthquake Engineering (2020) 18:3433–3457

behaviors of the façade (M1, M2, and M3), and of the nave (M19, M5, M6, and M13),
were evaluated (Fig. 8a-b, and Fig. 9a-b). As suggested elsewhere (Matteis et al. 2016) and
applied by previous authors in the study of New Zealand URM churches (Marotta et  al.
2016, 2017; De Matteis and Zizi 2019), the mean damage is computed according to:
∑n
dk,i
𝜇d = i=1 (4)
NJ

where 𝜇d is the average of the observed damage level for mechanisms, which is obtained
as the ratio between dk,I the number of macro-elements (i.e. façade and lateral walls) of
the buildings that reached the k-th damage level at a given intensity, IM = IMj , and Nj is
the total number of observed macro-elements at a given intensity, IMj . In general, a good
agreement is observed between the PMFs and the empirical distributions.

3.5 Empirical Fragility Curves (EFCs)

In this research, empirical fragility curves based on the observed database of URM
churches were derived as a function of PGA using different statistical models proposed
elsewhere (e.g., Lallemant et  al. 2015). First, we use a lognormal cumulative distribu-
tion function (CDF) fit that minimizes the weighted sum of the squared errors (SSE), the
expression of the lognormal CDF for each of the damage state is:
( ( ) )
( ) ln IMi − 𝜇j
P DS ≥ dsj |IM = IMi = 𝜑 (5)
𝛽j

where φ is the standard cumulative normal distribution function; 𝜇j and 𝛽j are the sample
mean and the standard deviation, computed suchthat the weighted sum of squared errors
between the probabilities predicted by the fragility function and the ones observed from the
data is minimum. Parameters 𝜇j and 𝛽j are estimated as follows:
L
( ( ( ) ))2
∑ ni,j ln IMi − 𝜇j
𝜇j , 𝛽j = arg min Ni −𝜑 (6)
𝜇j ,𝛽j i=1
Ni 𝛽j
ni,j
where Ni
is the ratio between a numerator that represents the number of buildings that
reached a damage level j at a given intensity IM = IMi , and Ni which is the total number of
observed buildings that experienced intensity IM = IMi ; and where L is the number of
considered intensity classes. The given fragility curves for the SSE-based lognormal model
are shown in Fig. 10a.
Another common model used for regression is a generalized linear model (GLM), which
is defined by three components: (1) a conditional probability distribution of the exponential
family; (2) a linear predictor; and (3) a link function that relates the linear predictor with
the response (Rossetto et al. 2013; Lallemant et al. 2015; Fig. 10b). These models are writ-
ten by the linear form (Lallemant et al. 2015):
g(𝜇) = 𝛼 + 𝛽1 X1 + 𝛽2 X2 + ⋯ + 𝛽n Xn = 𝜂 (7)
where μ is the expected response of given predictor variables X1, X2, …, Xn, and η is the
linear predictor related to the expected response through function g.

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3449

(a) (b)

(c)
Fig. 8  Failure mechanisms and Probability Mass Functions (PMFs) following a Multinomial distribution
(MPDF): a out-of-plane behavior of the façade (M1, M2); b in-plane behavior of the façade (M3); and c
MPDFs for intensities ranging from 0.16 g < PGA ≤ 0.53 g

13
 3450 Bulletin of Earthquake Engineering (2020) 18:3433–3457

(a) (b)

(c)
Fig. 9  Failure mechanisms and PMFs following an MPDF: a out-of-plane behavior of the lateral walls
(M19); (b) in-plane behavior of the lateral walls (M5, M6, M7, and M13); and c MPDFs for intensities in
the range 0.16 g < PGA ≤ 0.53 g

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3451

Fig. 10  Fragility curves of global behavior of Chilean churches using: a a lognormal distribution fit by least
squares; and b a GLM distribution fit by MLE

In developing fragility curves, Eq.  (7) is reduced to a single explicative variable say the
logarithm of IM, and two linear coefficients, the intercept α and the slope β. The value of μ is
the probability of exceedance of a particular damage state threshold (DS > dsj), conditioned to
the value of IM (i.e. Lallemant et al. 2015.). When adding the j subscript associated with to
damage state j, it yields:
( ) ( )
𝜇j = P DS ≥ dsj |IM = IMi = g−1 (𝛼j + 𝛽j log IMi ) (8)

The process of fitting a GLM then involves finding the coefficients that maximize the like-
lihood function (MLE) based on the assumption of a conditional distribution of the exponen-
tial family. The Fragility curves obtained from a lognormal distribution fit using least squares
and from a GLM distribution by a MLE are presented in Fig. 10b. The shapes and values of
the resulting curves are very similar at low and medium PGAs, though the GLM distribution
fit has a sharper increase at low IMs and a sharper asymptotic behaviour at high IMs.

13
 3452 Bulletin of Earthquake Engineering (2020) 18:3433–3457

4 Discussion and conclusion

This research presented a macro-scale statistical analysis based on detailed in-situ

inspections of the structural damage undergone by 106 URM Chilean churches during
the 2010 Maule earthquake.
The macro-seismic method presented earlier (Lagomarsino et al. 2004) for estimat-
ing the global damage index, id  , was implemented for each church subject to the 2010
Maule earthquake. Six of the twenty-eight mechanisms usually employed in the litera-
ture for these type of buildings were not activated in any structure due mainly to the
absence of vaults and isolated elements standing for the structural masonry. Indeed,
these elements are commonly built in timber or other similar light materials, such as a
mixture of wood and bamboo canes. Some mechanisms were activated in about 80% of
the cases and involved in-plane and out-of-plane mechanisms of the façade, side aisle
walls, and bell tower.
Qualitative and quantitative post-seismic damage scenarios have been reported by
using a macro-element approach, which was related to PGA, Probability Mass Func-
tions and Fragility Curves. A statistical analysis, to correlate the global damage levels to
the recorded PGA, was carried out. Also, empirical PMFs were computed, and a Multi-
nomial distribution was fit to the data. PMFs were defined for both, global damage indi-
ces and damage mechanisms of single macro-elements that most frequently activated
during the Maule earthquake.
Furthermore, Fragility functions were determined by fitting a lognormal distribution
and a generalized linear model by means of least squares error minimization and maxi-
mum likelihood estimation, respectively.
Results of PMFs show good agreement between predicted damage and observed data.
This is consistent with what it was noted for the Umbro-Marches churches (Lagomars-
ino and Podestà 2004a, b, c) and for the L’Aquila basin (De Matteis et al 2016). Moreo-
ver, for mechanisms that activated more frequently, such as in-plane and out-of-plane
behaviors involving the façade (M1 and M2 activated in 4/5 of the total sample) and the
sidewalls (M19 activated in 3.5 / 5), specific PMFs were quantified.
For the PMFs of the out-of-plane behavior (M19) of the nave macro-element (Fig. 8a)
the following conclusions can be derived:

• at high intensity range 0.41 g < PGA ≤ 0.53 g, the probability of total or partial col-

lapse ­(Dk = D5) of at least 2/3 of the mechanisms is ­P[Dk = D5] = 50%;
• at an intermediate intensity range 0.28 g < PGA ≤ 0.41 g, the probability of activa-
tion of mechanisms with severe structural damage is ­P[Dk = D3] = 45%;
• at a low intensity range 0.16  g < PGA ≤ 0.28  g, the probability of light damage in
several cases with the activation of one or two mechanisms is P ­[Dk = D2] = 22%.

For the PMFs of the in-plane behavior of the same macro-element (M5, M6, M7, and
M13) (Fig. 8b) the following conclusions can be obtained:

• at high intensity, the probability of reaching a damage level ­D5 is P ­[Dk = D5] = 38%;

• at an intermediate intensity, the probability of activation of several mechanisms that
include the collapse of some macro-elements is P [­ Dk = D4] = 33%;
• at a low intensity, the probability of severe damage is P
­ [Dk = D3] = 44%.

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3453

Unlike the side walls, the façade macro-element shows greater vulnerability in the in-
plane rather than in the out-of-plane behavior (Fig. 7b):

• at high intensity, the probability of reaching a level of severe damage with collapse
of 2 or more macro-elements is P­ [Dk ≥ D4] = 88%;
• at an intermediate intensity, the probability of having substantial to heavy damage is
­P[Dk = D3] = 28%;
• at a low intensity, the probability of having light damage in several mechanisms is
­P[Dk = D2] = 38%.

As stated above, Fragility functions were derived using two methods, a generalized
linear model using maximum likelihood estimation, and a lognormal distribution using
least squares error minimization. The results are very similar at low and medium PGAs,
though the GLM curves have a shaper increase at low PGAs and a shaper asymptotic
behavior at high PGAs. Future post-seismic damage scenarios for URM churches should
be contrasted with the proposed PMFs and FCs.
Despite the heterogeneity of churches presented elsewhere (Palazzi et  al. 2019a,
2019b), the whole sample considered (106 churches) was used to obtain a first attempt
of PMFs and FCs. A basic assumption of this research lies in the use of this heteroge-
neous database, which comes from specific architectural, constructive, and geometric
features. Based on such assumption, it was possible to define empirical seismic fragili-
ties for different historical “church” architectural typologies. Further investigations are
necessary to obtain specific fragility functions for each architectural style. Currently,
this is not possible because data limitations impede such goal. In the future, empirical
data of damage caused by low, medium, and high earthquake intensities must be sys-
tematically recorded and survey activities with the aim to collect a sufficient number of
observations may be realized using the methodology proposed herein. In fact, as shown
elsewhere (Rossetto et al. 2014) the quality of PMFs and FCs strongly depends on the
quality and quantity of observations within the empirical database. A high quality data-
base should derive from a large sample say larger than 100 buildings (representative of
total population) for each building class.
Regardless of these limitations, this research presents a first important attempt toward
defining systematically the seismic fragility of these historical buildings. Such large-
scale empirical seismic fragility assessments are useful to calibrate proper risk evalu-
ation and management tools. Indeed, structural fragilities are one of the most relevant
components of any risk analysis, especially in such complex historical systems, which
may activate a large number of different partial structural mechanisms that are diffi-
cult to predict analytically. This empirical analysis requires enough data to classify the
observed performance in the different damage states, which in turn need to be carefully
defined in terms of the seismic performance of the structure. These fragilities together
with a detailed exposure model and a spatially and spectrally correlated seismic haz-
ard, enables the development of reliable quantitative risk assessment studies that may
be used in the definition of proper mitigation strategies and critical local interventions.
Reliable quantification of seismic risk enables decision makers to establish—given
physical, social, patrimonial and economical constraints- priorities for the interventions
along the territory and optimize the use of resources in seismic protection measures that
minimize the equity at risk. Consequently, credible empirical information behind fragil-
ity functions are a crucial step toward reliable quantitative risk analysis.

13
 3454 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Acknowledgements  This work was supported by the National Research Center for Integrated Natural Dis-
aster Management CONICYT/FONDAP/15110017, and by the SIBER-RISK Regular Fondecyt Project
CONICYT/FONDECYT/1170836. Authors also thank the ECOS-CONICYT Scientific cooperation pro-
gram project “Multi-risk assessment in Chile and France: Application to seismic engineering and moun-
tain hazards”, ECOS170044 and ECOS action C17U02. Authors also thank the ECOS-CONICYT Scientific
cooperation program project “Multi-risk assessment in Chile and France: Application to seismic engineer-
ing and mountain hazards”, ECOS170044 and ECOS action C17U02, and the project Cheers. Cultural HEr-
itagE. Risks and Securing activities (ASP693) co-financed by the Interreg VB Alpine Space Programme
through the European Regional Development Fund (ERDF).

References
Astroza M, Cabezas F, Moroni M, Massone L, Ruiz S, Parra E, Cordero F, Mottadelli A (2010) Intensidades
sísmicas en el área de Daños del Terremoto del 27 de Febrero de 2010. Departamento de Ingeniería
Civil, Universidad de Chile, Chile, Santiago
Augusti G, Ciampoli M, Zanobi S (2002) Bounds to the probability of collapse of monumental buildings.
Struct Saf 24(2–4):89–105
Bahamondez P, Muñoz GE, Morales AM (2012) Patrimonio religioso en Chile: su valoración un proceso en
desarrollo. Conserva: Revista del Centro Nacional de Conservación y Restauración 17:13–24.
Basaglia A, Aprile A, Spacone E, Pilla F (2018) Performance-based seismic risk assessment of urban sys-
tems. Int J Archit Herit 12(7–8):1131–1149
Bergami AV, Nuti C (2013) A design procedure of dissipative braces for seismic upgrading structures.
Earthq Struct 4:85–108. https​://doi.org/10.12989​/eas.2013.4.1.085
Boroschek R, Soto P, Leon R (2010a) Red Local de Registros Edificio Camara Chilena de la Construccion.
Universidad de Chile, Facultad de Ciencias Fisicas y Matermaticas, Departamento de Ingegneria Civil,
Informe Renadic 10/03 Rev. 2.
Boroschek R, Soto P, Leon R (2010b) Registros del Terremoto del Maule Mw = 8.8 27 de Febrero de 2010,
Universidad de Chile, Facultad de Ciencias Fisicas y Matermaticas, Departamento de Ingegneria Civil,
Informe Renadic 10/05 Rev. 2 (in Spanish).
Boroschek R, Contreras V, Kwak DY, Stewart JP (2012) Strong Ground Motion Attributes of the 2010 M w
8.8 Maule, Chile. Earthq Earthq Spectra 28(S1):S19–S38.
Braga F, Dolce M, Liberatore D (1982). A statistical study on damaged buildings and ensuing review of
the MSK-76 scale. In: Proceedings of 7th European conference on earthquake engineering, Athens,
September
Braga F, Gigliotti R, Monti G, Morelli F, Nuti C, Salvatore W, Vanzi I (2015) Post-seismic assessment of
existing constructions: Evaluation of the shake maps for identifying exclusion zones in Emilia. Earthq
Struct 8(1):37–56. https​://doi.org/10.12989​/eas.2015.8.1.037
Brando G, De Matteis G, Spacone E (2017) Predictive model for the seismic vulnerability assessment of
small historic centres: application to the inner Abruzzi Region in Italy. Eng Struct 153:81–96
Calvi GM, Pinho R, Magenes G, Bommer JJ, Restrepo-Veléz LF, Crowley H (2006). The development of
seismic vulnerability assessment methodologies for variable geographical scales over the past 30 years.
ISET J Earthq Technol 43(3):75–104; Paper No. 472.
Cancino C (2010) Damage assessment of historic earthen sites after the 2007 earthquake in Peru. Adv
Mater Res 133–134:665–670
Cancino C, Farneth S, Garnier P, Neumann JV, Webster FD (2009) Damage Assessment of historic earthen
buildings after the August 15, 2007 Pisco Earthquake. Peru, The Getty Conservation Institute, LA, Los
Angeles
Cara S, Aprile A, Pelà L, Roca P (2018) Seismic Risk Assessment and Mitigation at Emergency Limit Con-
dition of Historical Buildings along Strategic Urban Roadways. Application to the “Antiga Esquerra de
L’Eixample” Neighborhood of Barcelona. Int J Archit Herit 12(7–8):1055–1075
Colombi M, Borzi B, Crowley H, Onida M, Meroni F, Pinho R (2008) Deriving vulnerability curves using
Italian earthquake damage data. Bull Earthq Eng 6(3):485–504
Conferencia Episcopal de Chile (2010) Magnitud de los Daños en Templos y Otros Recintos Católicos, No.
11535. https​://www.igles​ia.cl/breve​s_new/archi​vos/20100​325_catas​tro.pdf (in Spanish).
Da Porto F, Quelhas Da Silva B, Lorenzoni F, Girardella P, Valluzzi MR (2010) New integrated knowledge
based approaches to the protection of cultural heritage from earthquake-induced risk. In 4th structural
engineers world congress. At Villa Erba, Como.

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3455

D’Ayala D, Benzoni G (2012) Historic and traditional structures during the 2010 Chile Earthquake:
Observations, codes, and conservation strategies. Earthq Spectra 28–1:425–S451
De Matteis G, Criber E, Brando G (2016) Damage probability matrices for three-nave Masonry Churches
in Abruzzi after the 2009 L’Aquila Earthquake. Int J Archit Herit 10(2–3):120–145
De Matteis G, Zizi M (2019) Seismic damage prediction of Masonry Churches by a PGA-based
approach. Int J Archit Herit 13(7):1165–1179
Decanini L, Liberatore D, Liberatore L, Magenes G, Penna A, Sorrentino L (2012) Report on the Maule
(Chile) Febrary 27th, 2010 eartquake. IUSS Press, Pavia
Di Pasquale G, Orsini G, Romeo RW (2005) New Developments in Seismic Risk Assessment in Italy.
Bull Earthq Eng 3:101–128
Doglioni F (2000) Codice di Pratica (Linee Guida) per la progettazione degli interventi di riparazione,
miglioramento sismico e restauro dei beni architettonici danneggiati dal terremoto umbro marchi-
giano del 1997. Bollettino Ufficiale Regione Marche (15).
Dolce M, Kappos A, Masi A, Penelis G, Vona M (2006) Vulnerability assessment and earthquake dam-
age scenarios of the building stock of Potenza (Southern Italy) using Italian and Greek methodolo-
gies. Eng Struct 28(3):357–371
Fonseca FC, D’Ayala D (2012) Seismic assessment and retrofitting of Peruvian earthen churches by
means of numerical modelling. In: Proceedings of 15th world conference on earthquake engineer-
ing-15WCEE, Lisbon
GG.U. no. 47. (2011), February 26. (suppl. ord. no. 54). Directive of the Prime Minister dated on
9/02/2011, Assessment and mitigation of seismic risk of cultural heritage with reference to the
Technical Code for the design of constructions, issued by D.M. 14/1/2008 (in Italian). Ministry of
Public Building and Works, Rome, Italy
Giuffré A (1989) La meccanica nell’architettura: la statica. La Nuova Italia Scientifica, Rome
Giuffré A (1991) Lettura sulla meccanica delle murature storiche. Kappa, Rome
Grünthal G (1998) European macroseismic scale 1998. (EMS-98) European Seismological Commission,
subcommission on Engineering Seismology, Working Group Macroseismic Scales, vol 15. Conseil
de l’Europe, Cahiers du Centre Européen de Géodynamique et de Séismologie, Luxembourg, 99 pp
Indirli M, Razafindrakoto H, Romanelli F, Puglisi C, Lanzoni L, Milani E, Munari M, Apablaza S (2011)
Hazard evaluation in Valparaiso: the MAR VASTO project. Pure Appl Geophys 168(3–4):543–582
Indirli M, Valpreda E, Panza G, Romanelli F, Lanzoni L, Teston S, Rossi G (2006) Natural multi-hazard
and building vulnerability assessment in the historical centers: the examples of San Giuliano di
Puglia (Italy) and Valparaiso (Chile). In: 7th European Commission conference “SAUVEUR”.
Instituto Nacional de Normalización (INN) (1996) Earthquake resistant design of buildings. Official
Chilean Code NCh433.Of96.
Jiménez B, Pelà L, Hurtado M (2018) Building survey forms for heterogeneous urban areas in seismi-
cally hazardous zones. Application to the historical center of Valparaíso. Chile Int J Archit Herit
12(7–8):1076–1111
Jorquera N, Ruiz J, Torres N (2017) Analysis of seismic design criteria of Santo Domingo Church, a
Colonial Heritage of Santiago, Chile. Revista de la Construcción J Construct 16(3):388–402
Jorquera N., Palazzi NC, Rovero L, Tonietti U (2017). The church of San Francisco in Santiago, Chile.
Analysis of 400 years of earthquake-resistance behaviour. In: 16th world conference on earthquake,
16WCEE. Santiago, Chile.
Jorquera N, Vargas J, Lobos ML, Cortez D (2016) Revealing Earthquake-Resistant Geometrical Features
in Heritage Masonry Architecture in Santiago. Chile Int J Archit Herit 11(4):519–538
Jünemann R, de La Llera JC, Hube MA, Cifuentes LA, Kausel E (2015) A statistical analysis of rein-
forced concrete wall buildings damaged during the 2010, Chile earthquake. Eng Struct 82:168–185
Kalkbrenner P, Pelà L, Sandoval C (2019) Multi directional pushover analysis of irregular masonry
buildings without box behavior. Eng Struct 201:109534
Kircher CA, Nassar AA, Kustu O, Holmes WT (1997) Development of building damage functions for
earthquake loss estimation. Earthq Spectra 13(4):663–682
Kishor J, David Wald D, D’Ayala D (2011) Developing empirical collapse fragility functions for global
building types. Earthq Spectra 27(3):775–795
Lagomarsino S, Giovinazzi S (2006) Macroseismic and mechanical models for the vulnerability and
damage assessment of current buildings. Bull Earthq Eng 4:415–443
Lagomarsino S (2012) Damage assessment of churches after L’Aquila earthquake. Bull Earthq Eng
10:73–92. https​://doi.org/10.1007/s1051​8-011-9307-x
Lagomarsino S, Podestà S (2004a) Seismic vulnerability of ancient churches: I. Damage assessment and
emergency planning. Earthq Spectra 20(2):377–394. https​://doi.org/10.1193/1.17377​35

13
 3456 Bulletin of Earthquake Engineering (2020) 18:3433–3457

Lagomarsino S, Podestà S (2004b) Seismic vulnerability of ancient churches: II. Statistical analy-
sis of surveyed data and methods for risk analysis. Earthq Spectra 20(2):395–412. https​://doi.
org/10.1193/1.17377​36
Lagomarsino S, Podestà S (2004c) Damage and vulnerability assessment of churches after the 2002 Molise,
Italy, earthquake. Earthq Spectra 20:S271–S283. https​://doi.org/10.1193/1.17671​61
Lagomarsino S, Podestà S, Cifani G, Lemme A (2004) The 31st October 2002 Earthquake in Molise (Italy):
a new methodology for the damage and seismic vulnerability survey of churches. In: 13th world con-
ference on earthquake engineering. 1–6 Aug 2004, Vancouver, Canada. Paper 1366.
Lallemant D, Kiremidjian A, Burton H (2015) Statistical procedures for developing earthquake damage fra-
gility curves. Earthq Eng Struct Dyn.
Lira C, Arévalo C (2010) Informe de la Evaluación Preliminar de Daños Provocados por el sismo de 2010,
Chile: Iglesia de San Agustín de Melipilla. Dirección de Arquitectura del Ministerio de Obras Publi-
cas, Santiago de Chile, Consejo de Monumentos Nacionales
Marotta A, Sorrentino L, Liberatore D, Ingham J (2016) Statistical seismic vulnerability of New Zealand
unreinforced masonry churches. In: 10th international conference on structural analysis of historical
constructions. CRC Press, Boca Raton
Marotta A, Sorrentino L, Liberatore D, Ingham JM (2017) Vulnerability assessment of unreinforced
masonry churches following the 2010–2011 Canterbury earthquake sequence. J Earthq Eng
21(6):912–934
Martinelli A, Cifani G, Cialone G, Corazza L, Petracca A, Petrucci G (2008) Building vulnerability assess-
ment and damage scenarios in Celano (Italy) using a quick survey data-based methodology. Soil Dyn
Earthq Eng 28:875–889
MARVASTO (2007) Risk Management in Valparaíso/Manejo de Riesgos en Valparaíso, Servicios Técnicos
acronym MAR VASTO, funded by BID/IDB (Banco Inter Americano de Desarrollo / Inter-American
Development Bank). Project ATN/II-9816-CH, BID/IDB-ENEA Contract PRM.7.035.00-C, March
2007–June 2008
Orsini G (1999) A model for buildings’ vulnerability assessment using the parameterless scale of seismic
intensity (PSI). Earthq Spectra 15(3):463–483. https​://doi.org/10.1193/1.15860​53
Palazzi, NC, Rovero L, Tonietti U, de la Llera JC, Sandoval C (2018) Kinematic limit analysis of Basilica
del Salvador, a significant example of the neo-gothic architecture in Santiago, Chile. In: 16th European
conference on earthquake engineering, Thessalonica, Greece.
Palazzi NC, Rovero L, Tonietti U, de la Llera JC, Sandoval C (2019) Seismic vulnerability assessment
of unreinforced masonry churches in central Chile. Structural analysis of historical constructions.
Springer, Cham, pp 1172–1181
Palazzi NC, Rovero L, De La Llera JC, Sandoval C (2019b) Preliminary assessment on seismic vulnerability
of masonry churches in central Chile. Int J Archit Herit 1–20.
Pinto JI (2011) Informe de la Evaluación Preliminar de Daños Provocados por el sismo de 2010, Iglesia de
Loica, Chile. Ilustre Municipalidad de San Pedro, Departamento de Proyecto, Santiago de Chile
Recabarren F (2009) Datos Generals del Monuemto Nacional Iglesia de Loica. Dirección de Arquitectura
del Ministerio de Obras Publicas, Consejo de Monumentos Nacionales, Santiago de Chile
Rendel M, Lüders C, Greer M, Vial I, Westenenk B, de la Llera JC, Perez F, Bozzi D, Prado F (2014) Ret-
rofit, using seismic isolation, of the heavily damaged Basílica del Salvador in Santiago, Chile. In: New
Zealand society for earthquake engineering, 14NZEE. Aotea, New Zealand.
Restrepo-Vélez LF, Magenes G (2004) A mechanics-based procedure for the seismic risk assessment of
masonry buildings at urban scale. In: Proceedings of the XI Convegno Nazionale ANIDIS, Genova
Rossetto T, Ioannou I, Grant D (2013) Existing empirical vulnerability and fragility functions: compendium
and guide for selection. GEM Foundation, Pavia
Rossetto T, Ioannou I, Grant DN, Maqsood T (2014) Guidelines for empirical vulnerability assessment.
Pavia, Italy
Rota M, Penna A, Strobbia C (2006) Typological fragility curves from Italian earthquake damage data. In:
Proceedings 1st European conference on earthquake engineering and seismology, Geneva, paper no.
386
Sandoval C, Valledor R, Lopez-Garcia D (2017) Numerical assessment of accumulated seismic damage in a
historic masonry building. A case study. Int J Archit Herit 11(8):1177–1194
Singhal A, Kiremidjian AS (2004) Empirical fragility functions from recent earthquakes, In: Proceedings of
the 13th world conference on earthquake engineering, Vancouver, Canada, Paper No. 1211.
Singhal A, Kiremidjian AS (1998) Bayesian updating of fragilities with application to RC frames. J Struct
Eng 124(8):922–929
Staniscia S, Spacone E, Fabietti V (2017) Performance-based urban planning: framework and L’Aquila His-
toric City Center Case Study. Int J Archit Herit 11(5):656–669

13
Bulletin of Earthquake Engineering (2020) 18:3433–3457 3457

Torrealva D, Vicente E (2013) Testing results, SRP-Seismic Retrofitting Project in Perú Internal Report.
Pontificia Universidad Católica del Perú. Editorial Universitaria, Lima
Torres W, Almazán JL, Sandoval C, Peña F (2018) Fragility analysis of the nave macro-element of the
Cathedral of Santiago. Chile Bull Earthq Eng 16(7):3031–3056
United States Geological Survey’s (USGS) Magnitude 8.8-Offshore Maule, Chile. https​://stron​gmoti​oncen​
ter.org/NCESM​D/data/chile​_27feb​2010/eqinf​o.htm. Last Accessed 15 Jan 2020
Varum H, Tarque N, Silveira D, Camata G, Lobo B, Blondet M, Figueiredo A, Masood Rafi M, Oliveira C,
Costa A (2014) Structural behaviour and retrofitting of adobe masonry buildings, Structural Rehabili-
tation of Old Buildings, Building Pathology and Rehabilitation 2. Springer, Berlin
Whitman RV, Reed JW, Hong ST (1973) Earthquake damage probability matrices. In: Proceedings of the
5th World conference on earthquake engineering

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

13