International Journal of Architectural Heritage

Conservation, Analysis, and Restoration

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/uarc20

Loss Estimation for Typical Adobe Façades of

Cuenca (Ecuador) Due to Earthquake Scenarios

Xavier Cárdenas-Haro, Nicola Tarque, Leonardo Todisco & Javier León

To cite this article: Xavier Cárdenas-Haro, Nicola Tarque, Leonardo Todisco & Javier León
(2021): Loss Estimation for Typical Adobe Façades of Cuenca (Ecuador) Due to Earthquake
Scenarios, International Journal of Architectural Heritage, DOI: 10.1080/15583058.2021.1977417

To link to this article: https://doi.org/10.1080/15583058.2021.1977417

Published online: 21 Sep 2021.

Submit your article to this journal

View related articles

View Crossmark data

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

https://www.tandfonline.com/action/journalInformation?journalCode=uarc20
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE
https://doi.org/10.1080/15583058.2021.1977417

Loss Estimation for Typical Adobe Façades of Cuenca (Ecuador) Due to

Earthquake Scenarios
a,b c a a
Xavier Cárdenas-Haro , Nicola Tarque , Leonardo Todisco , and Javier León
a
Department of Continuum Mechanics and Structures - E.T.S.I Caminos, Canales y Puertos, Universidad Politécnica de Madrid, Madrid, España;
b
Facultad de Arquitectura y Urbanismo, Universidad de Cuenca, Cuenca, Ecuador; cGerdis Group, Civil Engineering Division, Pontificia
Universidad Católica del Perú, Perú

ABSTRACT ARTICLE HISTORY

A relevant percentage of the world population lives in constructions made of sun-dried earth, which Received 31 March 2021
are built according to several techniques such as rammed earth, bahareque or adobe. This work Accepted 2 September 2021
specifically focuses on the adobe buildings located in the Historic Centre of Cuenca (Ecuador), KEYWORDS
recognized as a World Heritage Site since 1999. Although these buildings are characterized by Adobe; capacity; damage
different geometrical and material configurations, all of them have load-bearing façade walls made factor; demand; fragility
of adobe. This paper illustrates an analysis of the seismic vulnerability of these adobe façade walls curves; seismic vulnerability
according to different geometrical configurations and considering an out-of-plane loading. This
study is based on original in-situ collected data about 45 one and multi-storey buildings. The
seismic behaviour of the case studies is investigated through a demand versus capacity analysis
based on 5 different failure mechanisms and considering a variable intensity of the seismic action.
This procedure has allowed to classify the case studies according to their seismic vulnerability and
to identify the most relevant failure mechanisms in such adobe façades. As a main result, some
continued fragility curves have been produced to allow replicating this procedure to buildings with
similar features. Finally, a repair cost estimation after seismic scenarios is estimated.

1. Introduction following composition of elements: 15–20% of clay,

10–25% of silt, and 55–70% of sand. In addition, it
A relevant percentage of the world population lives in
should be considered that the earth to produce the
constructions made of sun-dried earth (Costa et al.
adobe blocks has to be free of organic parts (an organic
2019), (Angulo-Ibañez et al. 2012), (Laborel-Préneron
content lower than 5% is accepted) (Doat et al. 1991).
et al. 2016). The United Nations Educational, Scientific,
The fabrication process consists in extracting the earth,
and Cultural Organization (UNESCO) considers that
mixing it with water, and adding vegetal fibres until
earth architecture represents around 20% of the build­
achieving an optimal plasticity and workability. In addi­
ings included in the world heritage site list (Gandreau
tion, big aggregates are removed from the earth before
and Delboy 2012). Earth constructions are spread
its use. After that, the mix is moulded and sun-dried for
around the world, although a relevant part is concen­
at least 2–3 weeks. The blocks can have a square or
trated in South American countries, such as Bolivia,
rectangular base (in the latter case with a width-to-
Colombia, Ecuador, and Peru. In the latter, for example,
height ratio of around 2), and a height between 50 and
around 30% of the buildings are made of sun-dried earth
200 mm. The Ecuadorian Inventory of constructions
(INEI 2019). Sun-dried earth is also known as raw or
(INEC) (INEC 2010a) indicates that 346,107 buildings
unbaked earth, and these terms are indifferently used
in Ecuador (9% of the totality) are characterized using
through the text. This material does not require baking
sun-dried earth in the load-bearing walls. The same
treatment before its use as a construction material,
INEC indicates that 83% of these constructions exhibit
resulting in simple and affordable building techniques
a bad or fair state of preservation. Specifically, in the
such as rammed earth, bahareque, or adobe. One of the
Cuenca municipality, there are 21,244 buildings made of
most adopted construction methods for sun-dried earth
unbaked earth: 55% of them reveal a fair state of con­
is adobe: it consists in using blocks produced by a
servation and 23% of them exhibit a bad state of main­
mortar composed of earth, water, and, in some cases,
tenance. In addition to this alarming situation, Ecuador
natural fibres. Generally, the mix is based on the
is a high seismic zone and these vernacular

CONTACT Xavier Cárdenas-Haro [email protected] Department of Continuum Mechanics and Structures - E.T.S.I Caminos, Canales
y Puertos, Universidad Politécnica de Madrid, Madrid, España
© 2021 Taylor & Francis
2 X. CÁRDENAS-HARO ET AL.

constructions are extremely weak against lateral actions, capacity, and the demand corresponding to various
due to the low strength of their building materials. It is ground motion intensities. The results allow to define
worth mentioning that the Adobe compressive strength fragility curves in Section 4 and to estimate the corre­
is around 1 MPa, while its tensile strength is totally sponding damage and repair costs in Section 5. Finally,
negligible (Cárdenas-Haro, Todisco, and León 2021). Section 6 summarizes the main findings of the study and
This is especially warned in cities such as Cuenca, defines future lines of research.
which the historic centre was declared World Heritage
Site in 1999 for preserving its image of a colonial town
2. Construction system of adobe buildings in
and the essential aspects of its original character. The
Cuenca
historic centre of Cuenca gathers 2400 buildings made of
unbaked earth and around 60% of these buildings are Although adobe constructions are spread across the
made of adobe (30% used in façade and 70% in interior world, this research is geographically limited to
elements) (Achig et al. 2013). The vernacular building Ecuador, where examples of adobe buildings exist since
tradition in Cuenca does not correspond to a unique 1470 (Medina and Abad 2002). Specifically, the histor­
constructive style and it is affected by many variables ical centre of Cuenca includes multistorey adobe build­
such as architectural configuration, geometric distribu­ ings built between 1809 and 1830 that represent a
tion of spaces, diversity of materials, and structural relevant showcase of colonial architecture (Salazar
typology (Pesántes and González 2011). Aware of this 2004). A typical construction system of an adobe build­
circumstance, relevant research has been carried out to ing located in Cuenca is illustrated in Figure 1. The
characterize the building materials and define strategies foundation of the building is composed by strip footings
for improving its performance (Donkor and Obonyo located below the walls. The generally employed mate­
2015), (Sharma et al. 2015), (Brown and Clifton 1978), rial for foundation is block masonry or concrete. The
(Silveira et al. 2015), (Lima et al. 2012), (Silveira et al. most common option combines stone block sizing from
2012), to analyse the construction system (Neves and 150 to 250 mm (gravel or rip-rap) with mortar to fill the
Borges Faria 2011), (Minke 2006), (Cárdenas-Haro et al. voids. The depth of the foundation is variable and it
2021), to develop procedures for performing numerical depends on the level of adequate soil to support the
simulations (Lourenço et al. 2019), (Al Aqtash et. al building. The strip footings generally are larger than
2017), (Tarque et al. 2014), (Bui et al. 2020), to evaluate the adobe walls of minimum 100 mm at each side
damages after earthquakes (Gautam et al. 2016), influ­ (Minke 2006), (Minke 2005). Then, a skirting board is
ence of site effects on seismic parameters (Tarque et al. placed on the footing; its role is basically to avoid the
2013), to outline fragility curves (Sumerente et al. 2020), ascending moisture and water splash on the adobe wall,
(Tarque et al. 2012) and to estimate deficiencies after whose base is the top surface of the skirting board (over
seismic events (Achig et al. 2013), (Illampas, Ioannou, foundation). It has generally a height of 300 mm at least
and Charmpis 2013), (Brando et al. 2019). The research and may have or not the same material composition and
presented in this paper takes advantage of the original aspect of the foundation itself (AIS Asociación
on-site surveys on 45 buildings to evaluate their seismic Colombiana de Ingeniería Sísmica 2004). The adobe
vulnerability. This is achieved by analysing different walls are essentially the lateral and gravity load-bearing
collapse mechanisms corresponding to out-of-plane elements of these buildings and, as mentioned above, are
loading of adobe façade walls. The procedure consists built with unbaked blocks made of water, earth, fibres,
in superposing the demand and capacity spectrums for and, in specific cases, additives. The mortar is basically
each case of the study and considering a different an earthen mortar and its thickness is variable between
ground seismic acceleration. This method allows to 10 and 30 mm. Its use is important to guarantee a uni­
define the corresponding fragility curves that can be form contact surface among blocks. These walls include
applied to similar cases. Finally, the damages and corre­ wooden lintels to allocate openings for doors and win­
spondent repair costs are estimated. In summary, this dows. The floors are based on wooden beams (with a
research strongly contributes to mitigate the seismic risk cross-section of around 100 × 100 mm) which cover the
associated with the collapse of the adobe buildings by shortest span between the adobe walls. These beams are
assessing the façade stability. The document is organized placed on wooden ring beams that rely on adobe walls.
as follows: Section 2 presents a description of the typical The wooden joists that compose the main structure of
adobe buildings located in Cuenca (Ecuador) and an the floor support timber planks fixed with nails or
overview of the 45 case studies. Next, Section 3 analyses screws. These planks are characterized by a width
the different collapse mechanisms, the calculation of the between 40 and 100 mm, a length of around 3 m, and
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 3

Figure 1. Typical construction system of an adobe building located in Cuenca (Ecuador).

a thickness variable between 5 and 10 mm. From a Cuenca, part of the Azuay province (Ecuador). The
structural perspective, these floors are considered as Historic Centre of Santa Ana de los Ríos of Cuenca is
flexible systems which are not able to stiff the structure organized in 10 districts, being the Gil Ramírez Dávalos
under seismic actions (Sumerente et al. 2020). district the one selected as the study area for this
Therefore, the floor is considered as a flexible dia­ research. In this district exist 298 buildings classified as
phragm, supported directly on the walls. Finally, the heritage sites and made of unbaked earth: 45 of them
roof is generally composed of a wooden truss that is were selected as case studies for this work. The building
supported by ring wooden beams located on the top of selection was based on identifying buildings in various
the walls. The horizontal top ceiling, below the roof, is states of preservation, to create a representative data­
composed by wood elements. Conversely, canes (reeds) base. The second criterion was operative privileging
intertwined with natural fibres are located on the ridge buildings close to each other, for an easy transportation
beams. Finally, an earthen mortar overlay with a variable of measuring equipment. A showcase of the case studies
thickness of 50–100 mm is spread above the canes to (8 of 45) is show in Figure 2. To each building, an
support the baked tiles placed on an outward slope to alphanumeric code has been assigned. The first letter
facilitate the water drainage. The roof generally has a indicates the city block while the numbers identify the
cantilever of around 0.6 m to protect the adobe façade building register number. The details of the character­
from rain. ization of the buildings can be reviewed in a previous
work by (Cárdenas-Haro et al. 2021).

3. Characterization of the case studies 3.2. Number of floors and the use of case studies
3.1. General overview of case studies Vernacular architecture is often associated with very small
Around 10% of the buildings located in Ecuador are structures. Nevertheless, the analysis of Figure 3a shows
made of unbaked earth, such as, adobe, rammed earth that only 31% of the case studies are composed by one-
and bahareque (INEC 2010b). The buildings employed storey. Most of the constructions (65%) are two-storey
as case studies in this research are located in the city of buildings and even a few of them (4%, 2 out of 45) have
4 X. CÁRDENAS-HARO ET AL.

Figure 2. Examples (8 out of 45) of one and two-storey case studies.

Figure 3. Analysis of the 45 case studies: (a) distribution of storey number; (b) distribution of the use surface area.

three or four storeys. Figure 3b presents the distribution 3.3. Geometrical configuration of adobe façades
of the surface associated with a specific use for the 45 case
Figure 4, from (a) to (f), plots the distributions of different
studies. If a mixed use exists (e.g., housing and shop), the
geometrical parameters (height, thickness, and slender­
surface for each use has been measured. Most of the case
ness) corresponding to the adobe façades collected in the
studies (42%) are used as private houses, while 53% of the
case studies. The analysed buildings are gathered into two
buildings are involved in commercial activities (31% as
groups depending on the number of stories: one-storey
hotel or similar and 22% as shops).
buildings (14 case studies, left column) and multi-storey
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 5

Figure 4. Geometrical configuration of the adobe façades gathered in one and multi-storey buildings: height(a-b), wall thickness (c-d);
wall slenderness (e-f).

buildings (31 case studies, right column). The relevant Next, Figure 4candFigure 4d plot the façade thickness of
statistical data are included at the top-right corner of each the case studies. Results illustrate that, for the 14 one-
plot (n = number of cases; Min. = minimum; storey buildings, 0.40 m and 0.85 m represent, respec­
Mean = mean; Max = maximum; σ = Standard deviation; tively, the minimum and the maximum values. Most of
Coef. Variation = Coefficient of variation). Figure 4a them (43%) are in the range between 0.4 and 0.6 m. In
andFigure 4b illustrate, respectively, the distribution of multi-storey buildings, there is an increase of façade
the height of the first and second storey (the height is thickness, in fact the minimum and maximum values
evaluated from the ground) for the 45 case studies. The are, respectively, 0.43 and 1.13 m, with most of the build­
average height of the first storey, measured from the ings (45%, 14 of 31) in a range between 0.8 and 1.0 m.
ground, is 3.8 m for the 14 one-storey case studies; while Finally, Figure 4e andFigure 4f analyse the façade slender­
the average height of second floor level/roof level from the ness, calculated as the ratio between the height of the wall
ground is 7.3 m for the 31 multi-storey case studies between consecutive storeys and its thickness. Most of the
(buildings L24 and M26 have 4 and 3 floors, respectively, one-storey case studies present a slenderness between 6
for the analysis only the first 2 floors were considered). and 8. By contrast, the slenderness increases with the
6 X. CÁRDENAS-HARO ET AL.

height of the building, in fact 36% of the multi-storey These advances led (Doherty et al. 2002), (Lam et al.
buildings are characterized by slenderness between 8 2003), (Griffith et al. 2003), (Ferreira et al. 2012),
and 10. As pointed out in another research (Cárdenas- (Gabellieri et al. 2012), (Derakhshan et al. 2013),
Haro et al. 2021), these results illustrate a great discre­ (Ferreira et al. 2014), (Landi et al. 2015), (Griffith,
pancy of the collected slenderness ratios with the values Lam, and Wilson 2006) to the use of rigid models for
established by the building codes. Finally, it is relevant to describing the out-of-plane behaviour of walls, moving
recall that façades include wooden lintels to allocate open­ from the simple bilinear model (force vs. displacement)
ings for doors and windows. Around 30% of the façade to a tri-linear semi-rigid model, in which parameters
areas are represented by openings for one-storey build­ such as damping, state of joint degradation, etc., are
ings. This value is almost the same (33%) for multi-storey considered. The analysis of each of these parameters
buildings, where 18% of the openings are located at the led several researchers to propose different failure
ground floor and 15% at the second storey. mechanisms and collapse coefficients, including parti­
cularities to consider the materiality of each building
(Sumerente et al. 2020), (Tarque et al. 2012), (Melis
4. Methodology to evaluate the collapse
2002), (D’Ayala et al. 2003), (Restrepo 2003),
mechanism of façades
(Derakhshan 2011), (Gabellieri 2012), (Milano et al.
4.1. General overview 2009).
The out-of-plane collapse mechanism is dominant
The assessment of the building seismic vulnerability
in adobe buildings (Apaza and Tarque 2020).
can be performed following, for example, a statistical
Therefore, this study analyses the overturning capa­
approach (probability of failure, probability of col­
city of 45 adobe façades based on mechanical models,
lapse), based on past events that occurred on similar
comparing the capacity with the seismic demand
structure typologies (Brando et al. 2019), (Whitman,
according to different seismic scenarios. The NEC
Reed, and Hong 2020), (Braga et al. 1982), (Gavarini
standard (NEC Norma Ecuatoriana de la
et al. 1984), (Corsanego et al. 1986), (Benedetti et al.
Construcción 2015) illustrates the different seismic
1988), (Coburn et al. 1992), (Augusti et al. 2000),
zones of Ecuador (Figure 5), considering a seismic
(Bernardini et al. 2008), (Mosoarca et al. 2019).
scenario of 10% probability to be exceeded in
This methodology, based on a mechanical approach,
50 years (i.e., a 475-year return period). The same
uses survey sheets to carry out visual building inspec­
figure also includes the number of earthen buildings
tions, assigning weights to establish a classification
located in each province. The province of Azuay has
based on the building vulnerability. Then, a damage
42,485 earthen buildings, which represent 12% of the
probability matrix can be created with the gathered
country’s total. The case studies are located in
information. This procedure has been applied to
Cuenca, the capital of that province.
assess the structural vulnerability of reinforced con­
The application of different seismic scenarios enables
crete structures, clay brick masonry, adobe, etc. Its
the assessment of the seismic behaviour of a given build­
application provides a general idea of the behaviour
ing typology in different situations and to evaluate the
of the building before a seismic event. In addition,
construction vulnerability degree with the aim of estab­
due to its simplicity, the procedure allows working
lishing priorities for intervention.
with large sets of buildings. An interesting field for
future research is the development of procedures to
perform numerical simulations. 4.2. Capacity assessment
Some authors (D’Ayala et al. 1997), (D’Ayala 1999),
The definition of the capacity curve of a given struc­
(Doherty et al. 2002) experimentally identified the sus­
ture is based on the application of lateral loads,
ceptibility of load-bearing walls of unreinforced
which gradually increase until causing a predeter­
masonry buildings to overturn due to out-of-plane
mined lateral displacement or building collapse. The
loading and proposed an analytical model to represent
model adopted for the analysis of the out-of-plane
this failure mechanism. Furthermore, (Calvi 1999) stu­
capacity of 45 façades is tri-linear and it relates the
died how to simplify a model for multi-storey buildings
force and the displacement according to the method
without losing accuracy in the results. As a result, Calvi
described by (D’Ayala and Speranza 2003), (D’Ayala
proposed to transform a system with several degrees of
and Speranza 2002). The mechanical model for a one
freedom (MDOF) to an equivalent single degree of
degree of freedom system assumes the simplification
freedom (SDOF), in which the influence of lateral
of the wall given by (Doherty et al. 2002). The con­
load as a function of displacement can be determined.
struction system of each façade allows to identify the
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 7

Figure 5. Identification of seismic zones of Ecuador. The different colours correspond to the maximum expected acceleration in rock
(NEC Norma Ecuatoriana de la Construcción 2015), while, the numbers below the text define the amount of earth buildings in the
different provinces.

boundary conditions adopted for the mechanical displacement at the top and a pivot at the base. The
modelling. Considering that all façades have a collar lateral forces which cause the unbalance are distributed
beam, the façade walls of the case studies are con­ in a triangular way; then, to simplify the model, the
sidered as simply supported, as shown in Figure 6. action of a force F0 is considered at the centre of gravity
The structural system of one-storey buildings is of the triangle (2/3 of the height measured from the
characterized using a collar or ring beam at the top of base). The mechanical model looks for the balance
the walls; this beam supports the roof system. Although between horizontal and vertical forces. The collapse is
the function of the beam is bracing, the assembly does achieved when the vertical force cannot be contained in
not guarantee it. Therefore, when lateral forces act on the thickness t of the wall. (Doherty et al. 2002) men­
the walls, it may behave as a cantilever, with free tion that this unbalance occurs when the horizontal
8 X. CÁRDENAS-HARO ET AL.

Figure 6. Mechanical model to define the capacity curve: (a) one-storey building; (b) multi-storey building. Figure adapted by authors
from (Doherty et al. 2002).

displacement Δe (calculated between pivot and the As described by Doherty et al. (Doherty et al. 2002), the
resultant of horizontal forces) achieves a value of Δe SDOF depends on two parameters: the mass given by the
= 2/3 t (Figure 6a). wall and the roof system, and the spring that models the
In two-storey buildings, two collar or ring beams are ability to bring the wall back after the application of a lateral
placed, one in the upper part of the first floor and the force.
other at the end of the second floor. The most suscep­ Experimental tests proved the differences between
tible point to undergo displacement is the connection modelling a rigid and a semi-rigid wall; in fact, Griffith
between the two floors. The mezzanine system (wooden et al. (Griffith et al. 2003) state that semi-rigid walls
floor) is considered flexible and it does not absorb the decrease their strength on the plateau compared to a
lateral force; therefore, it causes forces on the wall which rigid wall, deviating significantly from the bilinear
result in horizontal displacement. The approach of the model. In this research, a tri-linear model for computing
force system is similar to of the one-storey buildings; in the out-of-plane capacity of masonry walls is adopted.
this case, the wall is divided into 2 elements, in which 3 The construction starts from the bilinear model, as
pivots are marked (at the ends and in the centre). Again, shown in Figure 7, where the idealized force F0 repre­
as indicated in Figure 6b, the unbalance occurs when the sents the initial rocking force of an adobe wall. The wall
horizontal displacement reaches Δe = 2/3 t. reaches its ultimate displacement ∆u, which means
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 9

Figure 7. Force — displacement relationship for bi-linear and tri-linear models (adapted by authors from Griffith et al. (Griffith et al.
2003)).

instability. The bilinear model presents a single slope 2002)) propose the values listed in Table 1. In this
called stiffness K0, being the relationship between the research, assuming that the walls are in good condition,
force F0 and the ultimate displacement ∆u. This char­ these values were adopted: Δ1/Δu = 6% and Δ2/Δu = 28%.
acteristic is typical of the rigid body idealization. As demonstrated by other studies (Doherty et al. 2002),
The experimental tests carried out by Doherty et al. (Vaculik 2012), the effective displacement (Δe) of an
(Doherty et al. 2000), confirmed that unreinforced unreinforced masonry wall under triangular loading is
masonry walls can deform when subjected to compressive equal to 2/3 of the wall thickness. Furthermore, Restrepo
forces, demonstrating that the wall can behave as semi- (Restrepo 2003) indicates that, due to the low ductility of
rigid bodies, and the threshold resistance to rocking the adobe walls and the high degradation of the material,
reduces significantly from the original level associated 0.8 Δe should be taken as a limit value for the ultimate
with a rigid wall. The curve is idealized as a tri-linear displacement (Δu). This work follows this indication.
model, where three displacements Δ1, Δ2, and Δu are The value of the maximum lateral force (F0) causing
defined and the force Fmax is computed by expression 1. the ultimate displacement (Δu) is expressed as F0 = λW,
� � where λ is the collapse factor and W is the total weight of
Δ2
Fmax ¼ F0 1 (1) the wall. The collapse factor (λ) is specific to each failure
Δu
mechanism, and it generally takes into account the total
The initial slope of the tri-linear model is governed by mass of the failed portion, the connections with the
the force at the plateau Fmax and the value of Δ1. For orthogonal walls, the length of the wall overlap, friction
displacements between Δ1 and Δ2, the force Fmax is and the overload factor, defined as the ratio between the
constant. The plateau force ends when the displacement overload and the weight of the mechanism.
reaches Δ2. For displacements greater than Δ2, the tri­ D’Ayala and Speranza (D’Ayala and Speranza 2003)
linear and bilinear models coincide. The tests carried out show 8 diverse possible failure mechanisms. Tarque et
by Doherty et al. (Lam et al. 2003) proved that the al. (Tarque et al. 2012) indicate the A, C, and D mechan­
displacements Δ1 and Δ2 are related to the ultimate isms as the most probable ones for one-storey adobe
displacement Δu and the state of degradation of the buildings. This idea is confirmed by the experience
joint; the same authors (Doherty et al. (Doherty et al. obtained after the Pisco Earthquake in 2007.
Specifically, in-and out-of-plane collapse mechanisms
were responsible for most of the collapses observed
during this seismic event.
Table 1. Displacement ratios for the tri-linear model (Doherty et
For this research, this last criterion was broadened
al. 2002). and 5 different mechanisms were adopted (Figure 8).
Δ1/Δu Δ2/Δu Specifically, A, D, and G collapse mechanisms do not
State of degradation at cracked joint [%] [%] consider the connection with the perpendicular walls,
New 6 28 while B1 and B2 failure mechanisms contemplate differ­
Moderate 13 40
Severe 20 50 ent types of connection with the transverse walls.
10 X. CÁRDENAS-HARO ET AL.

Figure 8. Description of the 5 failure mechanisms adopted for one- (left column) and two-storey adobe building (right column).

Figure 9 shows the capacity curves, displacement (Δ) in Figure 9, while the maximum and minimum capaci­
vs. force (F), corresponding to one and two-storey ties are highlighted in green and red, respectively. The
façades for all considered mechanisms. The out-of- average capacity is coloured in orange. The mechanism
plane capacity curves of 45 façades are drawn in grey
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 11

Figure 9. Capacity curves considering 5 mechanisms for one and two-storeys adobe façades.

that requires the highest force to be activated is B1, 4.3. Damage Limit State (DLS)
while, on the contrary, the one that demands the lowest
Damage limit states describe the damage that a structure
force is the G mechanism.
suffers under a specific loading. For example, the Hazus
For the reader’s convenience, the calculated para­
Earthquake Model (FEMA October 2020) defines 4 DLS:
meters to define the tri-linear capacity curves are also
low, moderate, extensive, and complete. Several authors
listed in Table 2.
12 X. CÁRDENAS-HARO ET AL.

Table 2. Parameters to define the tri-linear capacity curves for the 45 adobe façades.
Δ1 Δ2 ΔU FA λA FB1 λB1 FB2 λB2 FD λD FG λG
ID N° of floors [mm] [mm] [mm] [kN] [-] [kN] [-] [kN] [-] [kN] [-] [kN] [-]
A1 1 19.2 89.6 320.0 4.0 0.2 13.7 0.27 9.1 0.20 2.4 0.39 6.6 0.19
B2 1 16.0 74.7 266.7 14.6 0.1 59.5 0.28 45.1 0.21 20.1 0.52 9.9 0.12
C3 1 19.2 89.6 320.0 11.1 0.2 30.6 0.26 23.0 0.19 9.0 0.45 8.0 0.16
C4 1 22.4 104.5 373.3 33.5 0.2 93.6 0.24 73.1 0.19 43.2 0.65 15.7 0.11
C5 1 27.2 126.9 453.3 52.9 0.3 148.1 0.35 128.8 0.31 65.5 0.84 16.0 0.09
C6 2 19.2 89.6 320.0 9.4 0.1 23.2 0.12 16.2 0.09 7.1 0.26 10.8 0.17
C7 2 22.4 104.5 373.3 13.3 0.1 28.7 0.13 20.1 0.09 9.1 0.30 10.9 0.15
D8 1 16.0 74.7 266.7 23.0 0.1 84.1 0.21 61.5 0.15 29.4 0.49 14.6 0.11
D9 1 16.0 74.7 266.7 29.8 0.2 142.3 0.41 123.8 0.36 41.8 0.71 11.1 0.09
D10 1 12.8 59.7 213.3 4.6 0.2 24.6 0.50 20.4 0.41 6.5 0.56 3.2 0.11
E11 1 12.8 59.7 213.3 4.9 0.2 23.9 0.37 18.2 0.28 4.8 0.38 4.2 0.13
E12 2 12.8 59.7 213.3 4.0 0.1 13.7 0.13 9.1 0.09 2.4 0.18 6.6 0.20
E13 2 13.8 64.2 229.3 7.1 0.1 25.0 0.10 16.7 0.07 7.2 0.19 14.0 0.17
F14 1 20.2 94.1 336.0 17.9 0.2 62.5 0.41 52.1 0.34 22.6 0.73 7.6 0.11
G15 1 14.4 67.2 240.0 9.5 0.2 45.8 0.46 38.6 0.39 12.3 0.65 4.7 0.11
G16 1 19.8 92.6 330.7 17.9 0.2 61.7 0.39 51.3 0.32 20.5 0.69 7.6 0.11
H17 2 32.0 149.3 533.3 18.5 0.2 35.0 0.16 28.6 0.13 13.0 0.40 14.1 0.18
H18 2 21.4 100.1 357.3 10.7 0.1 21.4 0.10 15.0 0.07 7.9 0.28 12.4 0.18
I19 2 19.8 92.6 330.7 15.6 0.1 42.0 0.13 31.9 0.10 10.6 0.27 12.4 0.14
I20 2 19.2 89.6 320.0 18.7 0.1 47.4 0.10 34.5 0.07 15.0 0.25 18.3 0.14
J21 2 32.0 149.3 533.3 22.1 0.1 33.8 0.10 25.7 0.08 18.8 0.39 20.9 0.18
J22 2 25.6 119.5 426.7 26.7 0.1 61.8 0.13 48.8 0.11 21.1 0.33 17.7 0.13
K23 1 25.6 119.5 426.7 24.2 0.2 50.0 0.20 39.1 0.16 22.4 0.59 14.2 0.16
L24 4 28.8 134.4 480.0 22.6 0.1 47.9 0.14 38.8 0.12 19.0 0.36 17.0 0.14
M25 2 28.8 134.4 480.0 34.2 0.1 66.5 0.10 51.1 0.08 34.1 0.36 25.9 0.13
M26 3 19.2 89.6 320.0 9.5 0.1 16.2 0.07 10.9 0.05 6.5 0.24 14.3 0.22
M27 2 29.1 135.9 485.3 29.6 0.1 60.7 0.13 48.4 0.10 26.4 0.37 20.6 0.13
M28 2 27.2 126.9 453.3 23.3 0.1 39.4 0.09 29.2 0.07 17.6 0.34 19.6 0.16
N29 2 23.0 107.5 384.0 15.7 0.1 39.9 0.14 30.6 0.11 13.3 0.30 13.9 0.14
N30 2 35.2 164.3 586.7 46.1 0.2 85.6 0.17 68.3 0.13 27.7 0.44 17.7 0.12
O31 2 28.8 134.4 480.0 19.8 0.1 39.3 0.14 32.6 0.11 14.3 0.36 15.7 0.17
O32 2 25.6 119.5 426.7 17.6 0.1 28.7 0.09 20.0 0.06 13.5 0.32 17.2 0.18
P33 2 36.2 168.7 602.7 61.9 0.2 131.8 0.17 112.6 0.14 47.3 0.45 25.7 0.11
P34 2 30.4 141.9 506.7 31.0 0.2 76.4 0.21 66.7 0.19 21.7 0.39 14.4 0.12
P35 2 32.0 149.3 533.3 33.5 0.1 63.8 0.14 49.8 0.11 25.4 0.40 19.0 0.13
P36 2 31.0 144.9 517.3 28.7 0.1 54.6 0.14 40.0 0.10 19.5 0.39 16.8 0.15
Q37 2 19.2 89.6 320.0 14.1 0.1 36.9 0.12 25.8 0.09 10.4 0.26 12.6 0.14
R38 2 28.8 134.4 480.0 28.8 0.2 60.0 0.16 47.1 0.13 18.0 0.37 14.4 0.13
S39 2 25.6 119.5 426.7 14.0 0.1 27.9 0.10 21.2 0.08 15.4 0.32 18.6 0.17
T40 1 20.2 94.1 336.0 26.4 0.2 90.9 0.31 73.0 0.25 37.5 0.67 12.6 0.10
U41 2 25.6 119.5 426.7 23.6 0.1 50.1 0.14 38.9 0.11 14.7 0.33 14.5 0.14
V42 2 22.4 104.5 373.3 19.6 0.1 48.5 0.16 36.8 0.12 11.8 0.30 11.3 0.13
W43 2 27.2 126.9 453.3 29.9 0.1 66.1 0.15 54.4 0.12 18.1 0.34 15.7 0.13
X44 2 28.8 134.4 480.0 54.9 0.1 120.5 0.13 101.2 0.11 42.5 0.36 27.2 0.11
Y45 2 26.9 125.4 448.0 35.3 0.1 82.6 0.15 68.9 0.12 24.3 0.34 18.2 0.12

(Calvi 1999), (D’Ayala et al. 2015), (Aguiar 2002), indicates severe damage, green states moderate damage
(D’Ayala et al. 2005), (Vargas et. al 2013), (Agudelo and, finally, light blue expresses low damage. The spe­
and Ricardo 2009), (Baker 2015) propose other DLS, cific definition of each DLS is presented in Table 3.
which are related to the material and the conditions to
which the structure is subjected. In this work, the DLS
4.4. Seismic demand assessment
suggested by Sumerente et al. (Sumerente et al. 2020) for
The seismic demand is related to the seismic hazard
one and two-storey adobe buildings are used, as
evaluated in a specific zone, and it should consider
described in Figure 10. The following limits are used:
the seismic wave amplification due to soil conditions.
DLS0-no damage (Δ = 0), DLS1-light damage (0.75× Δ1)
The representation of the demand can be given by
, DLS2-moderate damage (1.9× Δ1), DLS3-extensive
seismic accelerations through accelerograms or by
damage (Δ2) and DLS4-collapse (Δu).
uniform hazard spectra generally defined by stan­
The DLS range should be in agreement with the
dards or by site studies (Tarque et al. 2013). For
general description of the damage that the structure
this research, the spectrum specified in the
experiences. The assignment of colours facilitates the
Ecuadorian Construction Standard (NEC Norma
visual identification of the different states, adopting the
Ecuatoriana de la Construcción 2015) is taken for
following system: red represents collapse, orange
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 13

Figure 10. Damage limit states (DLS) adopted for adobe façades according to Sumerente et al. (Sumerente et al. 2020).

the city of Cuenca. The initial earthquake considered since the rocking mechanisms are in the inelastic range.
has a 10% probability of being exceeded in 50 years For this, the critical damping values suggested by Tarque
with a return period of 475 years. et al. (Tarque et al. 2012) were used: 10% for DLS1 and
The maximum acceleration value at rock level (Z) DLS2, 12% for DLS3, and 16% for DLS4. The modified
expected for the design earthquake in Cuenca is 0.25 g. spectrum where the spectral acceleration (Sa) is plotted
In this research, also lower rock acceleration values are versus the spectral displacement (Sd) is in accordance
considered for the seismic demand to evaluate changes with the HAZUS Earthquake Model (FEMA October
in vulnerability due to variation in seismic demand. The 2020). The spectral displacement (Figure 11b) is a func­
range of rock acceleration covers Z = 0.25 g, 0.20 g, tion of the spectral acceleration Sa and the period T,
0.15 g, 0.10 g, and 0.05 g. The spectral acceleration (Sa) according to expression 2.
expressed as a fraction of gravity versus the period (T) is 1
shown in Figure 11a. To evaluate out-of-plane mechan­ Sd ¼ Sa T 2 (2)
4π2
isms, it is necessary to establish an inelastic spectrum

Table 3. Description of the DLS (Adapted in (Tarque et al. 2012), (Calvi 1999), (Aguiar 2002)).
Damage
Description Range level
Building with almost no damage, minimal superficial cracks. Habitable building. LS0 < Demand Low
< LS1
Diagonal initiation of cracks. Close to doors and windows more visible cracks. Vertical cracks at the junction of orthogonal LS1 < Demand Moderate
walls. Habitable building. < LS2
Building very damaged, but still repairable. Most of the walls have diagonal cracks, some elements can detach and fall. Very LS2 < Demand Severe
pronounced cracks and in some cases part of the roof loses stability. The structure must be repaired to be inhabited again. < LS3
When, LS3 < Demand < LS4: Building not collapsed, but with severe damage. All walls with large cracks, partially or totally Demand > LS3 Collapse
fallen roof. Some walls have collapsed or are about to collapse. Non-habitable building.
When, Demand > LS4: Walls with large cracks and separation of orthogonal elements. The building is expected to collapse or
the building is collapsed.
14 X. CÁRDENAS-HARO ET AL.

Figure 11. (a) Acceleration spectrum (Sa vs. T); (b) Comparison of the elastic acceleration spectra with the inelastic overdamped spectra
(Sa vs. Sd).

4.5. Demand versus capacity demand spectrum intercepts with the effective secant
period of the capacity curve. In this point, the ductility
The superposition of the capacity spectrum of a given
is the same for capacity and demand, which allows to
structure with the seismic demand spectrum allows to
provide conclusions regarding the expected damage of
predict the damage that the structure may suffer under a
the structure.
given seismic event. Although this approach is conser­
The methodology is explained in detail in Figure 12
vative and some authors define it as level 2 or nonlinear
for building A1, which has one-storey. Here only the
kinematics, several forms of application of the method
failure mechanism A is considered. Figure 12a corre­
have been proposed (FEMA October 2020), (Chopra
sponds to an earthquake of Z = 0.15 g or Sa = 0.48 g,
and Goel 1999). However, the method is based on iden­
while Figure 12b corresponds to an earthquake of Z
tifying the performance point (PP) in a graphical way.
= 0.25 g or Sa = 0.81 g.
This PP is defined as the point where the inelastic
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 15

Figure 12. Superposition of the capacity spectrum and demand spectrum for a specific case study, considering different values of
acceleration at rock level: (a) Z = 0.15 g; (b) Z = 0.25 g.

In the graph, the following relationships are The most representative details are the displacements
drawn: the capacity curve of façade (according to Δ1 = 19.2 mm, Δ2 = 89.6 mm, and Δu = 320 mm. With
Section 4.2), the limit states (according to Section the displacements, the damage states were calculated as
4.3), and the elastic and inelastic spectra (according described in section 4.3, obtaining DLS1 = 14 mm,
to Section 4.4). DLS2 = 36 mm, D LS3 = 90 mm, and DLS4 = 320 mm.
16 X. CÁRDENAS-HARO ET AL.

Figure 13. Capacity vs. demand of the building C6: (a) A mechanism, (b) B1 mechanism, (c) B2 mechanism, (d) D mechanism and (e) G
mechanism.

The black dotted line is defined as the secant period The methodology here detailed has been then applied
and it passes through Δ2 since it is the point considered to consider the 5 failure mechanisms (A, B1, B2, D, and
of maximum displacement or the point where the col­ G) and, at the same time, 5 seismic rock accelerations (Z
lapse phase begins. The intersection of the secant period = 0.05 g, 0.10 g, 0.15 g, 0.20 g and 0.25 g). Therefore, for
with the inelastic or reduced demand spectrum is known each building it is possible to show the capacity-demand
as the performance point and represents the building plots associated to each failure mechanism. As an exam­
maximum capacity as a function of a specific seismic ple, the complete analyses of building C6 are show in
intensity. In the event that the performance point Figure 13.
exceeds the capacity of the element, then it would As expected, when the earthquake intensity increases,
move to the collapse phase since the demand is higher the damage in the structure is greater. The total evalua­
than the capacity of façade. tion of the one and two-storey buildings for each
In the case of Z = 0.15 g (or Sa = 0.48 g), the perfor­ mechanism is shown in Figure 14 with a discrete
mance point is between DLS2 and DLS3 (according to approach. For example, if the B1 failure mechanism is
Table 3), therefore its damage can be classified as severe. considered for one-storey façades (left column, second
For Z = 0.25 g (or Sa = 0.81 g), the performance point row), it is observed that, for an earthquake of Z = 0.05 g,
exceeds the capacity of the element, it is between DLS3 14% of façades exhibit slight damage, while 86% are
and DLS4 which means collapse of the structure. damaged in a moderate way. Conversely, for Z
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 17

Figure 14. Damage matrix of one and two-storey façades considering 5 the failure mechanisms and variable rock seismic acceleration.

= 0.25 g, 50% of the case studies show severe damage and Although it is now graphically seen, the importance of
the other 50% exhibit collapse. From all analysed the degradation state at cracked joints has been analysed.
façades, the D mechanism is associated with the lowest In Figure 14, the walls have been considered in good
damage, but in all cases the structures within this condition, but if the state of conservation is moderate,
mechanism were closed to severe or collapse damage. the capacity of the wall decreases 17% and, if low, it
18 X. CÁRDENAS-HARO ET AL.

Figure 15. Construction process of fragility curves: (a) one-storey adobe façade considering B1 mechanism; (b) two-storey adobe
façade considering D mechanism.

drops down to 31%. Therefore, the wall state of degrada­ represents buildings as a given typology must be used.
tion should be considered an important parameter to Aguiar (Aguiar 2002) indicates that there are two meth­
assess the wall capacity. odologies for damage evaluation: the first is a determi­
nistic one that consists in evaluating the damage in
elements or in the entire structure without considering
5. Fragility curves probabilistic events in its process. The second option is
based on a probabilistic approach through fragility
5.1. Probabilistic approach curves relating the damage with cumulative probability.
When there is a relatively large group of buildings, it According to (Aguiar 2002), the probabilistic approach
may be complex (time consuming) to evaluate the seis­ maybe the most appropriate for evaluating several build­
mic damage in each of them. Then, a methodology that ing typologies.
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 19

Table 4. Mean and standard deviation for the construction of B1, B2, D, and G) and for all case studies. Two exam­
continuous fragility curves for one and two-storey façade for each ples of this process can be seen in Figure 15 for a one
damage state. and two-storey building.
One-storey façade
Light Moderate Severe
damage damage damage Collapse 5.3. Adjustment of fragility curves to a lognormal
Failure ø (Sa) β ø (Sa) β ø (Sa) β ø (Sa) β distribution
mechanism [g] [-] [g] [-] [g] [-] [g] [-]
A 0.050 0.450 0.050 0.450 0.284 0.082 0.607 0.208 To facilitate the use of fragility curves on other structures
B1 0.050 0.450 0.133 0.175 0.320 0.075 0.810 0.159 with similar features, the discrete curves can be converted
B2 0.050 0.450 0.119 0.200 0.312 0.068 0.721 0.144
D 0.050 0.450 0.164 0.128 0.455 0.230 0.879 0.056 into continuous ones. The procedure adopted here is the
G 0.050 0.450 0.050 0.450 0.178 0.100 0.496 0.199 one established by Baker (Baker 2015), which applies sta­
Two-storey façade tistical inference to determine the mean and standard
A 0.048 0.450 0.050 0.450 0.226 0.227 0.550 0.278
B1 0.050 0.450 0.050 0.450 0.240 0.219 0.608 0.212 deviation values that better fit the discrete fragility curves.
B2 0.050 0.450 0.050 0.450 0.181 0.110 0.505 0.222 The usual way to graph the fragility curves is by placing the
D 0.050 0.450 0.161 0.119 0.380 0.217 0.945 0.244
G 0.050 0.450 0.050 0.450 0.272 0.089 0.639 0.044
spectral acceleration (Sa) or spectral displacement (Sd) on
the x-axis and the probability of exceeding a certain level of
damage on the y-axis (FEMA October 2020). In general, the
fragility curve has an approximation to the lognormal func­
The fragility curves represent the probability of a tion and, in this work, it is represented according to expres­
given structural typology to exceed a certain damage sion 3:
level during a seismic event (Vargas et al. 2013), " � �#
(Coronel and López 2015). According to D’Ayala 1 Sa
P½Dsi j Sa� ¼ Φ ln (3)
(D’ayala 2005), once the curve has been defined, it βDSi Dsi
represents the structural behaviour of buildings with
the same characteristics. whereΦ represents the standard normal cumulative
Technical reports, such as the one given by the Global distribution function and β the standard deviation of the
Earthquake Model (D’Ayala et al. 2015) or the Hazus natural logarithm of Ds. The values obtained from the
Technical Manual (FEMA October 2020), explain the mean and standard deviation are summarized in Table 4
methodology providing fragility curves for various for buildings with one and two storeys.
materials, construction systems and places. Several As an example of the final result, the fragility curves
authors computed such curves for different structural generated for one-storey façades are shown in Figure 16.
typologies: Agudelo and López (Agudelo and Ricardo In A and G mechanisms, the low damage state is over­
2009) for concrete structures with filled masonry walls, lapped with the moderate state because there are no case
Brando et al. and Villar-Vega et al. (Brando et al.), studies on light damage. For example, if the mechanism
(Villar-Vega et al. 2017) for adobe buildings, Lovon et A (Figure 16a) and an acceleration of 0.7 g are consid­
al. (Chopra and Goel 1999, 80) for confined masonry ered for one-storey façade, it results that the probability
dwellings, etc. of exceeding the collapse is 76% and the probability of
In this research, the fragility curves for the adobe exceeding the severe damage is 100%.
façades are calculated taking into account the The fragility curves generated for two-storey façades and
Ecuadorian seismic demand. for all failure mechanisms are drawn in Figure 17. In most
mechanisms, the low damage state curve is overlapped with
the moderate damage state, except for mechanism D.
5.2. Calculation of discrete fragility curves
Based on the procedure described in Section 4.5, a 6. Loss damage ratio
matrix is constructed, the number of buildings in
each damage state is counted, as well as for the different 6.1. Mean damage ratio
seismic intensities. The values described as a percen­ One of the most relevant applications of the fragility
tage, are placed in the damage probability matrix curves consists in estimating the structural damage.
(DPM) and finally those percentage values are cumu­ These curves establish the thresholds between the con­
lated in the new matrix defined probability of excee­ sidered damage states (no damage, moderate, severe,
dance. The matrix allows the generation of discrete and collapse) for different spectral accelerations (Sa)
fragility curves for each failure mode. This procedure and, therefore, they allow to obtain the damage prob­
is repeated for all considered failure mechanisms (A, ability in a structure with similar features (material,
20 X. CÁRDENAS-HARO ET AL.

Figure 16. Fragility curves for one-story adobe façades (a) A mechanism; (b) B1 mechanism; (c) B2 mechanism; (d) D mechanism; (e) G
mechanism.

construction system, geometrical configuration) and = 50% and collapse Df = 100%. The mean damage
located in the same area. The Damage Factor (Df) is factor (Dfm) is obtained as the summation of the dif­
calculated as the ratio between the repair cost and the ferent damage probabilities associated to a structure,
replace cost. The guidelines published by the Global multiplied by the corresponding Df. Figure 18 illus­
Earthquake Model (D’Ayala et al. 2015) identify the trates the mean damage factor for each failure mechan­
different classifications (author, damage state, material, ism. In all case studies, it can be seen that with a higher
country, etc.) with reference values for the damage seismic action corresponds a higher value of the Dfm.
factor. Considering that there is no explicit classifica­ Specifically, in one-storey buildings (Figure 18a), the G
tion for adobe buildings, for the specific case of this mechanism exhibits the highest Dfm (99.5%), while, as
research, the following Df are adopted: light damage Df opposite, the D mechanism shows the lowest Dfm
= 2%, moderate damage Df = 10%, severe damage Df (54%). The results are similar for the multi-storeys
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 21

Figure 17. Fragility curves for two-storey adobe façades (a) A mechanism; (b) B1 mechanism; (c) B2 mechanism; (d) D mechanism; (e) G
mechanism.

buildings (Figure 18b); in fact, the G and B2 mechan­ (1) To identify the fragility curve associated with the
isms have the highest Dfm, respectively, with 100% and relevant failure mechanism
99.5%, while, the D mechanism shows the lowest Dfm (2) To define the spectral acceleration for the analysis
(64%). (3) To determinate the occurrence probability asso­
ciated with each limit state
(4) To specify the damage factor related to each
6.2. Damage cost calculation
damage limit state
To estimate the damage cost, the following procedure (5) To multiply the damage factor by the probability
has been followed of each damage limit state to obtain the mean
damage factor of the building
22 X. CÁRDENAS-HARO ET AL.

Figure 18. Medium damage factor for each failure mechanism: (a) one-storey buildings and (b) multi-storey buildings.

Table 5. Items and corresponding repair costs. The repair cost is based on the information gathered
Items Unit Amount Unit cost Total cost in Table 5.
Adobe wall, thickness of 0.6 m m2 1 51.00 51.00 EUR The unit cost is the sum of direct (equipment and
EUR
Loam plaster, thickness of 40 m2 2 12.20 24.40 EUR tools, materials, transportation, and labour) and indir­
mm EUR ect costs, corresponding to the market in the city of
Painting of walls m2 2 15.35 30.70 EUR
EUR Cuenca (Ecuador). The final cost of 1 m2 of a 0.60
Total 106.10 m-thick adobe wall is 106.10 EUR. This estimation
EUR
does not consider the costs related to the waste man­
agement related to the collapsed wall and ignores the
restoration costs related to floors, roofs, and more
(6) The restoration cost is obtained by multiplying possibly affected elements.
the area affected by the out-of-plane failure Considering that the areas associated to the façade of
mechanism with the repair cost and dividing the each case study are different, Figure 19 shows the aver­
result by the medium damage factor. age restoration cost per m2, for each failure mechanism
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 23

Figure 19. Repair costs corresponding to different failure mechanisms and intensities of the seismic actions: (a) one-storey buildings
and (b) multi-storeys buildings.

and for different intensities of the seismic action. As a 7. Conclusions

reference, the red line represents the average area
The UNESCO World Heritage Site Historic Centre of
damaged by each failure mechanism.
Cuenca (Ecuador) represents an extraordinary example
It is interesting to evidence that for a Sa of 0.81 g, the
of colonial architecture that preserves essential aspects
G mechanism generates the highest repair cost per m2
of its original character. The load-bearing façades of
(102.80 EUR and 142.30 EUR, respectively, for one-
these buildings are made of adobe and this results in
storey and multi-storey buildings), although the
extremely weak structures with lateral loading. This
mechanism with the highest area to be repaired is B2.
paper aims at preserving these constructions by asses­
Therefore, it is necessary to consider both elements to
sing their seismic vulnerability.
estimate the total repair cost.
The main conceptual remarks can be outlined as follows
24 X. CÁRDENAS-HARO ET AL.

● Mechanisms G and D are associated with the mini­ damage factor can be improved, considering also other
mum required force F0 in the failure process, since failure mechanisms in this kind of buildings.
they entail the lowest values of the mobilized area.
Conversely, mechanisms B1 and B2 require the
highest values of F0.
Disclosure statement
● As expected, to higher values of the seismic action No potential conflict of interest was reported by the author(s).
correspond to a higher building vulnerability.
However, independently of the seismic scenario,
all case studies show an extremely high vulnerabil­
ORCID
ity. As an example, considering only B1 failure Xavier Cárdenas-Haro http://orcid.org/0000-0001-5063-
mechanism of one-storey building, for a very low 7366
earthquake of Z = 0.05 g 86% of façades exhibit Nicola Tarque http://orcid.org/0000-0002-1029-9240
Leonardo Todisco http://orcid.org/0000-0003-1077-4942
moderate damage.
Javier León http://orcid.org/0000-0003-1284-2216
● If the seismic scenario according to the Ecuadorian
code is considered (Z = 0.25 g), it can be stated that
most of the one and two-storey buildings with the References
mechanism G, A, B1 and B2 are responsible for the Achig, M.-C., M. Zúñiga, K. Van-Balen, and L. Abad. 2013.
greatest damages. On the contrary, mechanism D Sistema de registro de daños para determinar el estado
causes the lowest damage in the structure. constructivo en muros de adobe. Maskana 4 (2):71–84.
● With a higher seismic action corresponds a doi:10.18537/mskn.04.02.06.
higher value of the mean damage factor. Agudelo, J., and L. Ricardo. 2009. Curvas de fragilidad para
estructuras de pórticos de hormigón armado con paredes de
Specifically, in one-storey buildings, the G
mamposteria. Caso de estudio: Estructuras de 1 y 2 niveles
mechanism exhibits the highest Dfm (99.5%), en puerto rico. Rev. Int. Desastr. Nat. Accid. E Infraestruct.
while, as opposite, the D mechanism shows the Civil 9 (1–2):163–86. [Online]. Available: http://ezproxy.
lowest Dfm (54%). The results are similar for the uniandes.edu.co .
two-storey buildings. Aguiar, R. 2002. Sistema de computación CEINCI3 para eva­
● A relevant parameter to assess the building vulner­ luar daño sísmico en los países Bolivarianos. Quito,
Ecuador: Centro de Investigación Científica Escuela
ability is the state of degradation at cracked joints. Politécnica del Ejército.
In this work, the walls have been considered in a AIS, A. (2004). Manual para la rehabilitación de viviendas
good condition, but if the state of conservation is construidas en adobe y tapia pisada. www.asosismica.org
moderate, the capacity of the wall decreases 17% Al Aqtash, U., P. Bandini, and S. L. Cooper. August 2017.
and, if low, it drops down to 31%. Therefore, the Numerical approach to model the effect of moisture in
adobe masonry walls subjected to in-plane loading.
wall state of degradation is a key parameter to
International Journal of Architectural Heritage 11
assess the wall capacity. (6):805–15. doi:10.1080/15583058.2017.1298010.
● The reduction of the roof weight is not beneficial Angulo-Ibáñez, Q., Á. Mas-Tomás, V. Galvañ-LLopis, and J.
for the seismic behavior of the wall, in fact accord­ L. Sántolaria-Montesinos. 2012. Traditional braces of earth
ing to the failure mechanisms A, B1, B2, and G, a constructions. Construction and Building Materials 30
reduction of this load corresponds to a decrease of (May):389–99. doi:10.1016/j.conbuildmat.2011.11.024.
Apaza, D., and N. Tarque, “Seismic vulnerability assessment of
the collapse factor λ and, consequently, to a higher peruvian colonial churches using the collapse mechanisms
building vulnerability. methodology, case study: Puno cathedral - Peru,” in 8th
● Continuous fragility curves have been provided Euro-American Congress on Construction Pathology,
through the paper. They can be applied by Rehabilitation Technology and Heritage Management,
other researchers to buildings with similar REHABEND 2020, 2020, pp. 953–61. Granada, Spain
Augusti, G., and M. Ciampoli. April 2000. Heritage buildings
features.
and seismic reliability. Progress in Structural Engineering
● Finally, the restoration costs are estimated for dif­ and Materials 2(2):225–37. doi:10.1002/1528-2716
ferent failure mechanisms and considering a wide (200004/06)2:2<225::AID-PSE28>3.0.CO;2-5.
range of intensities of the seismic action. Baker, J. W. February 2015. Efficient analytical fragility
function fitting using dynamic structural analysis.
This research is based on data collected by the Earthquake Spectra 31(1):579–99. doi:10.1193/
021113EQS025M.
authors on 45 case studies. As future lines of research, Benedetti, D., G. Benzoni, and M. A. Parisi. 1988. Seismic
this database can be extended to consider more build­ vulnerability and risk evaluation for old urban nuclei.
ings and, therefore, to improve the accuracy of the Earthquake Engineering & Structural Dynamics 16
fragility curves. Moreover, the computation of the (2):183–201. doi:10.1002/eqe.4290160203.
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 25

Bernardini, A., and S. Lagomarsino. August 2008. The seismic D’Ayala, D. 1999. Correlation of seismic vulnerability and
vulnerability of architectural heritage. Proceedings of the damages between classes of buildings: Churches and
Institution of Civil Engineers - Structures and Buildings 161 houses. In Seismic damage to masonry buildings, 41–58.
(4):171–81. doi:10.1680/stbu.2008.161.4.171. Routledge.
Braga, F., M. Dolce, and D. Liberatore. 1982. A statistical study D’Ayala, D., and E. Speranza. 2002. An integrated procedure
on damaged buildings and an ensuing review of the M.S. for the assessment of seismic vulnerability of historic build­
K76 scale. in Proceedings of 7th European Conference on ings. in 12th European Conference on Earthquake
Earthquake Engineering, pp. 431–50, Athens, Greece. Engineering, 2002, vol. 1–561, no. January. London, UK.
(Accessed: December. 22, 2020). [Online]. Available: D’Ayala, D., A. Meslem, D. Vamvatsikos, K. Porter, T.
https://scholar.google.es Rossetto, and V. Silva. 2015. Guidelines for analytical vul­
Brando, G., G. Cocco, C. Mazzanti, M. Peruch, E. Spacone, C. nerability assessment of low/mid-rise buildings.
Alfaro, K. Sovero and N. Tarque. November 2019. D’Ayala, D., and E. Speranza. 2003. Definition of collapse
Structural survey and empirical seismic vulnerability assess­ mechanisms and seismic vulnerability of historic masonry
ment of dwellings in the historical centre of Cusco, Peru. buildings. Earthquake Spectra 19 (3):479–509. doi:10.1193/
International Journal of Architectural Heritage 1–29. 1.1599896.
doi:10.1080/15583058.2019.1685022 D’Ayala, D., R. Spence, C. S. Oliveira, and A. Pomonis. 1997.
Brando, G., A. Pagliaroli, G. Cocco, and F. Di Buccio. 2019. Earthquake loss estimation for Europe’s historic town centers.
Site effects and damage scenarios: The case study of two Earthquake Spectra 13 (4):773–93. doi:10.1193/1.1585980.
historic centers following the 2016 Central Italy earthquake. D’ayala, D. F. December 2005. Force and displacement based
Engineering Geology 272 August :105647. doi:10.1016/j. vulnerability assessment for traditional buildings. Bulletin
enggeo.2020.105647. of Earthquake Engineering 3(3):235–65. doi:10.1007/
Brown, P. W., and J. R. Clifton. 1978. Adobe I: The properties s10518-005-1239-x.
of adobe. Studies in Conservation 23 (4):139–46. de la Construcción, N. E. 2015. Peligro Sísmico/Diseño Sismo
doi:10.1179/sic.1978.019. Resistente. Código: NEC-SE-DS, Quito, Ecuador.
Bui, T.-L., -T.-T. Bui, Q.-B. Bui, X.-H. Nguyen, and A. Limam. Derakhshan, H. 2011. Seismic assessment of out-of-plane
2020. Out-of-plane behavior of rammed earth walls under loaded unreinforced masonry walls. University of
seismic loading: Finite element simulation. Structures 24 Auckland.
(January):191–208. doi:10.1016/j.istruc.2020.01.009. Derakhshan, H., M. C. Griffith, and J. M. Ingham. 2013.
Calvi, G. M. 1999. A displacement-based approach for vulner­ Airbag testing of multi-leaf unreinforced masonry walls
ability evaluation of classes of buildings. Journal of subjected to one-way bending. Engineering Structures
Earthquake Engineering 3 (3):411–38. doi:10.1080/ 57:512–22. doi:10.1016/j.engstruct.2013.10.006.
13632469909350353. Doat, P., A. Hays, H. Houben, S. Matuk, and F. Vitoux. 1991.
Cárdenas-Haro, X., L. Todisco, and J. León. 2021. Database Building with earth, First Engl. New Delhi: The Mud Village
with compression and bending tests on unbaked earth Society.
specimens and comparisons with international code Doherty, K., B. Rodolico, N. T. K. Lam, J. Wilson, and M. C.
provisions. Construction and Building Materials 276 Griffith. 2000. The modelling of earthquake induced col­
(March):122232. doi:10.1016/j.conbuildmat.2020.122232. lapse of unreinforced masonry walls combining force and
Cárdenas-Haro, X., L. Todisco, J. León, C. Jurado, and E. displacement principals. 12th World Conf. Earthq. Eng., no.
Vergara. 2021 February. Geometry and proportions of January, pp. 1–8. London, UK.
adobe vernacular buildings in cuenca, Ecuador. Doherty, K., M. C. Griffith, N. Lam, and J. Wilson. 2002.
International Journal of Architectural Heritage 1–19. Displacement-based seismic analysis for out-of-plane bend­
doi:10.1080/15583058.2021.1879312. ing of unreinforced masonry walls. Earthquake Engineering
Chopra, A. K., and R. K. Goel. 1999. Capacity-demand- & Structural Dynamics 31 (4):833–50. doi:10.1002/eqe.126.
diagram methods for estimating seismic deformation of Donkor, P., and E. Obonyo. 2015. Earthen construction mate­
inelastic structures : SDF systems. rials: Assessing the feasibility of improving strength and
Coburn, A., and R. Spence. 1992. Earthquake protection. deformability of compressed earth blocks using polypropy­
Earthquake Protection. doi:10.5459/bnzsee.27.2.163. lene fibers. Materials & Design 83:813–19. doi:10.1016/j.
Coronel, G., and O. López. 2015. Metodología para la matdes.2015.06.017.
estimación de daños por sismo en edificios escolares de FEMA, H. October 2020. HAZUS earthquake model, technical
Venezuela mediante curvas de fragilidad. Revista De La manual. Federal Emergency Management Agency – FEMA.
Facultad De Ingeniería Universidad Central De Venezuela Washington DC, EE. UU.
30 (2):159–80. Ferreira, T., R. Vicente, H. Varum, A. Costa, and A. A. Costa.
Corsanego, A., A. Del Grosso, and D. Stura. 1986. Seismic 2012. Out-of-plane seismic response of stone masonry
vulnerability assessment for buildings: A critical review of walls: Experimental and analytical study of real piers in
current methodologies. in Proceedings of the 8 th European 15th World Conference on Earthquake Engineering, Lisbon,
Conference on Earthquake Engineering. Lisbon, Portugal. Portugal.
Costa, C., Â. Cerqueira, F. Rocha, and A. Velosa. 2019. The Ferreira, T. M., A. A. Costa, V. Romeu, H. Varum, A. Arêde,
sustainability of adobe construction: Past to future. and A. Costa. 2014. Application of a trilinear model for the
International Journal of Architectural Heritage 13 analytical study of the out-of-plane behaviour of unrein­
(5):639–47. doi:10.1080/15583058.2018.1459954. forced stone masonry walls. in 9 th International Masonry
Conference 2014, pp. 1–8. Guimaraes (Portugal)
26 X. CÁRDENAS-HARO ET AL.

Gabellieri, R., P. P. Diotallevi, and L. Landi. 2012. Effect of Lourenço, P. B., F. Greco, A. Barontini, M. P. Ciocci, and G.
diaphragm flexibility on the dynamic behaviour of unrein­ Karanikoloudis. 2019. Seismic retrofitting project:
forced masonry walls in out-of-plane bending in Proc. of Modeling of prototype buildings. Los Angeles, [Online].
the 15th World Conference on Earthquake Engineering, Available: http://www.getty.edu/ .
[Online]. Lisboa, Portugal. Available: http://www.iitk.ac.in Medina, M. I. C., and P. A. E. Abad. 2002. La cité cuencana: el
Gabellieri, R. 2012. Comportamento dinamico fuori del piano afrancesamiento de Cuenca en la época republicana (1860–
di pareti murarie : Influenza della deformabilità degli impal­ 1940) (Vol. 1). Universidad de Cuenca. Cuenca, Ecuador
cati. Universitá di Bologna. Melis, G. 2002. Displacement-Based Seismic Analysis for Out
Gandreau, D., and L. Delboy. 2012. World heritage inventory of Plane Bending of unreinforced masonry walls. Istituto
of earthen architecture. CRATerre-ENSAG. UNESCO. Universitario di Studi Superiori di Pavia.
Grenoble, France. Milano, L., A. Mannella, C. Morisi, and A. Martinelli. 2009.
Gautam, D., H. Rodrigues, K. K. Bhetwal, P. Neupane, and Y. Schede illustrative dei principali meccanismi di collasso
Sanada. 2016. Common structural and construction defi­ locali negli edifici esistenti in muratura e dei relativi modelli
ciencies of Nepalese buildings. Innovative Infrastructure cinematici di analisi.
Solutions 1 (1):1–18. doi:10.1007/s41062-016-0001-3. Minke, G. 2005. Manual de construcción para viviendas
Gavarini, C., and P. Angeletti. 1984. Assessing seismic vulner­ antisísmicas de tierra, Vol. 1, 3ra ed. Kassel:
ability in view of developing cost/benefit ratios for existing Forschungslabor für Experimentelles Bauen - Universidad
r.c buildings in Italy. in Proceedings of the 8th Word de Kassel, Alemania.
Conference on Earthquake Engineering, vol. 1, pp. 445– Minke, G. 2006. Building with earth: Design and technology of
52. San Francisco, EE. UU. a sustainable architecture. Kassel: Birkhäuser.
Griffith, M. C., G. Magenes, G. Melis, and L. Picchi. January Mosoarca, M., I. Onescu, E. Onescu, B. Azap, N. Chieffo, and
2003. Evaluation of out-of-plane stability of unreinforced M. Szitar-Sirbu. July 2019. Seismic vulnerability assessment
masonry walls subjected to seismic excitation. Journal of for the historical areas of the Timisoara city, Romania.
Earthquake Engineering 7(sup001):141–69. doi:10.1080/ Engineering Failure Analysis 101(March):86–112.
13632460309350476. doi:10.1016/j.engfailanal.2019.03.013.
Griffith, M. C., N. Lam, and J. Wilson. January 2006. Neves, C., and O. Borges Faria. 2011. Técnicas de Construcción
Displacement-based assessment of the seismic capacity of con tierra. Bauru: FEB-UNESP/PROTERRA.
unreinforced masonry walls in Bending. Australian Journal Pesántes, M., and I. González. 2011. Arquitectura tradicional
of Structural Engineering 6(2):119–32. doi:10.1080/ en Azuay y Cañar. Ecuador: Cuenca.
13287982.2006.11464949. Restrepo, L. 2003. A simplified mechanics – based procedure
Illampas, R., I. Ioannou, and D. C. Charmpis. 2013. Overview for the seismic risk assessment of unreinforced masonry
of the pathology, repair and strengthening of adobe buildings. Istituto Universitario di Studi Superiori di Pavia.
structures. International Journal of Architectural Heritage Salazar, E. 2004. Cuenca: Santa Ana de las Aguas. Ediciones
7 (2):165–88. doi:10.1080/15583058.2011.624254. Libri Mundi. Cuenca, Ecuador
Sharma, V., H. K. Vinayak, and B. M. Marwaha. 2015.
INEC. 2010a. Base de datos de población y vivienda 2010. Enhancing sustainability of rural adobe houses of hills by
Instituto Nacional de Estadísticas y Censos. https://www. addition of vernacular fiber reinforcement. International
ecuadorencifras.gob.ec/resultados/ . Journal of Sustainable Built Environment 4 (2):348–58.
INEC. 2010b. Censo de Población y Vivienda 2010 Censo de doi:10.1016/j.ijsbe.2015.07.002.
Población y Vivienda 2010. https://www.ecuadorencifras. Silveira, D., H. Varum, A. Costa, and J. Carvalho. 2015.
gob.ec/informacion-censal-cantonal/ . Mechanical properties and behavior of traditional adobe
INEI. 2019. Infraestructura física de las viviendas particulares wall panels of the Aveiro District. Journal of Materials in
INEI. https://www.inei.gob.pe (accessed January 26, 2021). Civil Engineering 27 (9):04014253. doi:10.1061/(ASCE)
Laborel-Préneron, A., J. E. Aubert, C. Magniont, C. Tribout, MT.1943-5533.0001194.
and A. Bertron. 2016. Plant aggregates and fibers in earth Silveira, D., H. Varum, A. Costa, T. Martins, H. Pereira, and J.
construction materials: A review. Construction and Building Almeida. 2012. Mechanical properties of adobe bricks in
Materials 111 (May):719–34. doi:10.1016/j. ancient constructions. Construction and Building Materials
conbuildmat.2016.02.119. 28 (1):36–44. doi:10.1016/j.conbuildmat.2011.08.046.
Lam, N. T. K., M. Griffith, J. Wilson, and K. Doherty. May Sumerente, G., H. Lovon, N. Tarque, and C. Chácara. May
2003. Time–history analysis of URM walls in out-of-plane 2020. Assessment of combined in-plane and out-of-plane
flexure. Engineering Structures 25(6):743–54. doi:10.1016/ fragility functions for adobe masonry buildings in the per­
S0141-0296(02)00218-3. uvian andes. Frontiers in Built Environment 6(May):1–10.
Landi, L., R. Gabellieri, and P. P. Diotallevi. 2015. A model for doi:10.3389/fbuil.2020.00052.
the out-of-plane dynamic analysis of unreinforced masonry
walls in buildings with flexible diaphragms. Soil Dynamics Tarque, N., C.G. Lai, F. Bozzoni, E. Miccadei, T. Piacentini, G.
and Earthquake Engineering 79:211–22. doi:10.1016/j. Camata and E. Spacone. November 2013. Expected ground
soildyn.2015.09.013. motion at the historical site of Poggio Picenze, Central Italy,
Lima, S. A., H. Varum, A. Sales, and V. F. Neto. 2012. Analysis with reference to current Italian building code. Engineering
of the mechanical properties of compressed earth block Geology 166:100–15. doi:10.1016/j.enggeo.2013.09.003
masonry using the sugarcane bagasse ash. Construction Tarque, N., G. Camata, H. Varum, E. Spacone, and M.
and Building Materials 35:829–37. doi:10.1016/j. Blondet. June 2014. Numerical simulation of an adobe
conbuildmat.2012.04.127. wall under in-plane loading. Earthquakes and Structures 6
(6):627–46. doi:10.12989/eas.2014.6.6.627.
 INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 27

Tarque, N., H. Crowley, R. Pinho, and H. Varum. May 2012. Villar-Vega, M., V. Silva, H. Crowley, C. Yepes, N. Tarque, A.
Displacement-based fragility curves for seismic assessment B. Acevedo, M. A. Hube, G. Coronel and M. H. Santa. May
of adobe buildings in Cusco, Peru. Earthquake Spectra 28 2017. Development of a fragility model for the residential
(2):759–94. doi:10.1193/1.4000001. building stock in South America. Earthquake Spectra 33
Vaculik, J. 2012. Unreinforced masonry walls subjected to (2):581–604. doi:10.1193/010716EQS005M
out-of-plane seismic actions. University of Adelaide. Whitman, R. V., J. W. Reed, and S. T. Hong.1973.
Vargas, Y. F., L. G. Pujades, A. H. Barbat, and J. E. Hurtado. April Earthquake damage probability matrices. In
2013. Evaluación probabilista de la capacidad, fragilidad y daño Proceedings of 5th European Conference on
sísmico de edificios de hormigón armado. Revista Earthquake Engineering, p. 2531, Accessed: December.
Internacional De Métodos Numéricos Para Cálculo Y Diseño 22, 2020. [Online]. Rome, Italy. Available: https://scho­
En Ingeniería 29(2):63–78. doi:10.1016/j.rimni.2013.04.003. lar.google.es