Engineering Structures 173 (2018) 1–27

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Assessment of the efficiency of traditional earthquake resistant techniques T

for vernacular architecture
⁎
Javier Ortegaa, , Graça Vasconcelosa, Hugo Rodriguesb, Mariana Correiac
a
ISISE, Department of Civil Engineering, University of Minho, Guimarães, Campus de Azurém, 4800-058 Guimarães, Portugal
b
RISCO, School of Technology and Management, Polytechnic Institute of Leiria, Campus 2, 2411-901 Leiria, Portugal
c
CI-ESG Research Centre, Escola Superior Gallaecia, Vila Nova de Cerveira, Portugal

A R T I C LE I N FO A B S T R A C T

Keywords: Specific architectural elements can be identified in constructions located in regions frequently exposed to
Vernacular architecture earthquakes. These earthquake resistant features were developed empirically by local communities to protect
Earthquakes their built-up environment. Research in these traditional practices, resulting from a local seismic culture, is a
Numerical analysis relevant and positive approach, since it focuses on the strength of the system rather than on its weaknesses. Its
Pushover parametric study
integration into current vernacular building practices can help to preserve and retrofit surviving in-use examples
Seismic performance
while respecting their authenticity. The main goal of the present work is to assess numerically the efficiency of
Stone masonry
Rammed earth traditional earthquake resistant solutions to mitigate the seismic vulnerability of vernacular architecture. The
Local seismic culture paper thus presents the results of a detailed numerical study based on finite element modeling and nonlinear
Seismic strengthening static (pushover) analysis intended to quantitatively evaluate the influence of each technique on the seismic
behavior of vernacular constructions and to better understand their structural role under seismic loading.

1. Introduction just in the immediate aftermath of the event, with no future orientation,
developing a reactive response behavior [6]. In any case, traditional
Earthquakes are naturally occurring events that affect negatively earthquake resistant construction techniques arise from this need to
people and their environment by causing loss of life, injury, property repair earthquake damage to both personal and public buildings. These
damage, social and economic disruptions or environmental damage. efforts made by local populations as a reaction to earthquakes, giving
Since they seriously disrupt the functioning of a community, local rise to the development of a LSC, can become a key element for the
builders have thus often integrated seismic risk as a determinant for preservation of cultural identity and vernacular construction practices.
construction, adopting different strategies to protect the population They can eventually help to preserve existing vernacular buildings
from these natural disasters. They have developed rich and varied while respecting their authenticity. However, since these traditional
knowledge resulting in singular construction techniques, building de- techniques are typically the result of empirical knowledge transmitted
tails and temporary technical devices aimed at reducing the vulner- along generations, a numerical evaluation and analytical comprehen-
ability of structures. The existence of a local seismic culture (LSC) was sion of their possible beneficial effect in the seismic resistance of ver-
recognized and firstly investigated by Ferrigni [12]. From that moment nacular buildings is necessary.
on, a LSC has been identified in many countries throughout the world Therefore, the present paper addresses the gap in knowledge about
frequently exposed to earthquakes, such as Italy [43], Greece [49], the real efficiency of the most common measures adopted by local
Turkey [18], Algeria [13], Iran [36], India [24], Nepal [15], Japan communities to repair and restore their dwellings. A better awareness
[40], Haiti [2], Colombia [34]and Portugal [41]. of traditional earthquake resistant measures and a better understanding
LSC thus emerges from the need of local population to react to of their structural role when subjected to earthquake loading is con-
earthquakes and from the efforts made for the physical community to sidered important to protect and reduce the seismic vulnerability of the
survive. People can either undertake preventive measures, repairing built vernacular heritage by encouraging local communities to re-
and refurbishing their personal properties in order to minimize future cognize and readopt techniques emerging from a local seismic culture.
losses in the following earthquakes, or they can respond to earthquakes For that matter, the paper firstly outlines briefly some of the most

⁎
Corresponding author.
E-mail addresses: [email protected] (J. Ortega), [email protected] (G. Vasconcelos), [email protected] (H. Rodrigues),
[email protected] (M. Correia).

https://doi.org/10.1016/j.engstruct.2018.06.101
Received 6 February 2018; Received in revised form 8 June 2018; Accepted 26 June 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
J. Ortega et al. Engineering Structures 173 (2018) 1–27

common techniques traditionally used throughout the world, focusing forces also ensures the ‘box-behavior’ of the building by enhancing the
on those that will be later further evaluated numerically. Secondly, the development of in-plane resisting mechanisms in the walls, which are
paper introduces the numerical strategy adopted for the evaluation of typically the main structural elements in vernacular buildings. The first
the efficiency of the different techniques in improving the seismic be- group thus includes techniques that are mainly aimed at improving the
havior of vernacular buildings. Thereafter, the paper presents and dis- connections between the different structural elements, such as ring
cusses the numerical assessment of each technique independently, fol- beams, corner braces, quoins and ties.
lowing the same organization: (a) description of the numerical models The second set of techniques includes those aimed at stabilizing
prepared for the simulation of the different strengthening solutions; and structural elements and buildings by imparting resistance and de-
(b) discussion of the results showing a comparative analysis of the formation capacity to the typically brittle stone masonry or earthen
seismic performance of the buildings with and without the different walls. This added strength to the main resisting structural elements can
techniques in terms of damage patterns, failure mechanisms and result in preventing common out-of-plane failure patterns associated to
strength capacity. Finally, the paper will present a summary of the the bending of the masonry walls or in-plane failure associated to the
different traditional earthquake resistant techniques particularly ad- typically low shear strength of earthen and stone masonry materials.
dressing their effect on the seismic behavior of vernacular buildings in Additionally, by enhancing the quality of the masonry with the use of
terms of the particular construction and structural aspects that are new elements that ensure an adequate bracing between the wall leaves,
improved with the use of the different techniques. This summary is these techniques can also prevent the common delamination or bulging
provided as a conclusion and can be helpful as a criterion for decision- of the external wall leaf commonly observed in stone masonry walls.
making on the seismic strengthening of vernacular constructions. The use of timber elements within the masonry and the wall subdivision
by means of brick horizontal courses are common techniques that
follow this principle.
2. Traditional earthquake resistant techniques for vernacular
Finally, the last group of traditional earthquake resistant techniques
architecture
corresponds to those that are meant to counteract horizontal loads
exerted by the buildings during the shaking by adding extra resistance
The numerical study presented is based on previous research on tra-
to the lateral thrust with the addition of new structural elements. The
ditional earthquake resistant solutions resulting from a local seismic
most common and widespread technique belonging to this group is the
culture carried out by Ortega et al. [41], who presented a comprehensive
construction of buttresses or counterforts. They are typically placed at
overview of the use of the different techniques around the world and
critical locations, such as the mid-span of long walls, which are the
particularly elaborated on their structural role and earthquake resisting
most vulnerable elements to the effects of out-of-plane earthquake vi-
concept. As a continuation of the abovementioned work, the present
brations, and at the corners, in order to avoid the separation of the
paper will delve further into this specific subject with the help of finite
walls. Another efficient technique of this kind is thickening the walls,
element (FE) modeling and nonlinear analysis. The traditional earthquake
which adds extra in-plane resistance because of increasing the resisting
resistant techniques previously identified that will be now further eval-
area. Moreover, it can also help to prevent common out-of-plane me-
uated using numerical analysis are summarized in Table 1.
chanisms and reduces the possibility of the walls to overturn because it
Since traditional earthquake resistant techniques mainly follow si-
reduces their height-to-thickness ratio.
milar earthquake resisting principles, the table is arranged according to
them and makes reference to the type of damage that can be avoided with
3. Numerical strategy
the application of each technique. One of the most common earthquake
damage patterns observed is the separation between structural elements,
The numerical study proposed to evaluate the influence of the tradi-
which is typically led by a lack of structural integrity, i.e. deficient wall-
tional seismic strengthening techniques shown in Table 1 was performed
to-wall connections or inadequate wall-to-roof and wall-to-floor connec-
using detailed FE modeling following a common macro-model approach
tions. That is why many traditional techniques are intended to improve
together with nonlinear static (pushover) analysis. The strategy consists of
these connections, enhancing the global behavior of the structure by
first constructing a reference FE model based on representative verna-
forming closed contours in vertical and horizontal planes so that stress
cular rammed earth constructions commonly found in the South Portu-
concentrations are avoided and forces are transmitted from one compo-
guese region of Alentejo. The different strengthening solutions considered
nent to another even through large deformations. A proper transfer of
will be then simulated and applied to the reference model. Several nu-
merical models will be prepared for each technique, reproducing different
Table 1
Summary of traditional earthquake resistant techniques numerically evaluated,
construction characteristics that can typically vary. Therefore, the nu-
classified according to the main earthquake resisting principle and highlighting merical study also comprises individual parametric analyses for the dif-
the type of damage avoided. ferent solutions in order to assess how the variations influence the seismic
response of the structure by means of comparison with the response of a
Technique Damage prevention Section
reference numerical model. This will provide a better understanding of
Separation Out-of- Delamination In-plane the seismic behavior of these traditional earthquake resistant solutions
plane and their particular structural role showing different characteristics in
terms of construction, geometry and materials.
1. Techniques improving the connection between structural elements
Despite being a simplified procedure because of simulating the
Ring beams X X X X 4.1
Corner braces X 4.2 earthquake loading as a set of equivalent static forces, pushover ana-
Quoins X 4.3 lysis is a widespread and generally accepted tool for the seismic as-
Ties X X 4.4 sessment of existing buildings, due to the relatively low computation
2. Techniques stabilizing structural elements and buildings demand in comparison with other methods, such as nonlinear dynamic
Timber elements X X X X 4.5 analysis. Similarly, the macro-modeling strategy also simplifies the
within the heterogeneity of the masonry into a single material with average ma-
masonry
Wall subdivision X X X 4.6
terial properties. However, their combined use represents a good
compromise between efficiency and accuracy, and has provided suc-
3. Techniques counteracting horizontal loads
cessful results for the analysis of the seismic behavior of complex his-
Buttresses X X 4.7
Walls thickening X X 4.8 torical masonry and rammed earth structures [25,30,3,27,45,29]. Since
the planned numerical campaign comprised a significant number of

2
J. Ortega et al. Engineering Structures 173 (2018) 1–27

analyses, the use of both tools was deemed the most suitable. LS follows recommendations available in the literature [44,11,35].
The distribution of lateral loads used for the pushover analyses is
directly proportional to the mass, uniform along the buildings height. 3.2. Reference numerical model
Pushover consists of increasing the loads monotonically until reaching
the collapse of the structure, which allows determining the ability of the The reference FE models used for the numerical study will vary
building to resist the characteristic horizontal loading caused by the
slightly depending on the traditional earthquake resistant technique
seismic actions taking into account the material nonlinear behavior. under evaluation. Nevertheless, they are all initially based on typical
Thus, the analysis provides an estimation of the seismic capacity of the
Portuguese vernacular rammed earth constructions. Rammed earth
structure, while the structural response of the building can be con- construction, known as ‘taipa’ in Portugal, and adobe masonry have
trolled in terms of damage patterns and failure mechanisms. This re-
been extensively used in the southern regions of Portugal. Detailed
sponse is described by the capacity or pushover curve, which represents information about this vernacular typology can be found in Correia [7]
the base shear or load factor (i.e. the ratio between the horizontal forces
and Correia [8]. From a general point of view, these traditional
at the base and the self-weight of the structure) versus the displacement dwellings have reduced dimensions, simple rectangular shape and one
at the control point, which is usually the point where the highest dis-
to two floors. The maximum wall span is quite variable, but rarely
placements take place. exceeds 10 m. Vernacular constructions are usually simple regarding
their plan configuration, little compartmentalized and present massive
3.1. Definition of limit states walls with few or no openings, other than a single door, as a measure of
thermal protection for the hot summers. Structural load bearing walls
In order to have a common basis for comparison of the seismic ca- typically range between 0.4 and 0.6 m thick and are the main vertical
pacity of the buildings for the different earthquake resistant techniques, resisting elements of the buildings. The walls inter-story height is also
different structural limit states were defined based on the obtained more variable, but is usually low, typically ranging between 2 and 3 m.
capacity curve. Four structural limit states (LS) were quantitatively A stone masonry base course or soco is also constructed, aimed at
defined for each building, according to the capacity curve obtained protecting the rammed earth from the water and rising damp. The
from the pushover analysis. Since the pushover curves depict the global height of this protection element is typically around 0.4–0.5 m, but can
response of the building and the formation of global collapse mechan- typically reach values of 1 and even 1.5 m. Timber elements are used as
isms, the defined LS are associated to different overall damage states of lintels placed over windows and doors. Rammed earth walls are typi-
the building. Descriptions of the LS, the basis for their definition, as cally constructed using corner framework to create a continuous
well as typical damage levels to which they are associated, are shown in structure and proper wall-to-wall connections. Timber is also present in
Table 2. An example of their identification of the different LS on a most of the buildings as the horizontal structural elements (floors and
pushover curve is shown in Fig. 1. It is noted that the definition of the roofs). The timber floors observed in this vernacular architecture

Table 2
Definition of the considered limit states (LS).
LS Definition

LS1 It can be defined as the immediate occupancy limit state because, before this limit, the structural behavior of the building remains in the elastic
part and the structure can be considered as fully operational. LS1 thus corresponds to the formation of the first cracks in the structure,
characterizing the end of the elastic behavior. The first cracks observed in the numerical models appeared for a relatively low degradation of
the initial stiffness, which is related to the low tensile strength of the materials considered for this study. Finally, the beginning of cracking is
established at a degradation of the initial stiffness in the pushover curve up to 2%

LS2 It can be defined as the damage limitation limit state and depicts the transition between a point where the structure is still functional, retaining
most of its original stiffness and strength, showing minor structural damage and cracks, and a state where significant damage is visible so that
the building could not be used after without significant repair. LS2 is calculated based on energy criteria so that the area below the three-linear
curve formed by LS1, LS2 and LS3 coincides with the area below the pushover curve from LS1 to LS3. The criterion also involves that the point
that defines LS2 is on the slope associated to the secant stiffness corresponding to 70% of the maximum resistance

LS3 It can be defined as the life safety limit state because is related to the attainment of the building maximum resistance. As a result, the building
has lost a significant amount of its original stiffness, but is supposed to retain some lateral strength and margin against collapse even if it
cannot be used after the earthquake. LS3 is defined by the base shear coefficient (or load factor) and displacement corresponding to the
attainment of the building maximum resistance

LS4 It can be defined as the near collapse limit state because is related to the collapse of the building and, after reaching this LS, repairing the
building is neither possible nor economically reasonable. LS4 corresponds to the point where the building resistance deteriorates below an
acceptable limit, which is established at the 80% of the maximum resistance

3
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 1. Identification and definition of the considered limit states on the pushover curve of the structure.

Fig. 2. Typical rammed earth constructions in Portugal [7].

typology are typically very simple, being merely composed of wooden and the in-plan dimensions of the models are 8 m × 5.5 m. In all
beams covered with cross boards directly nailed to the beams com- models, the roof and floor loads are simply modeled as distributed load
posing the sheathing. When larger spans are required, two-way floors along the walls. The reference models are thus considered to present
are commonly used, with the addition of a secondary set of timber joist flexible diaphragms of negligible stiffness with beams poorly connected
perpendicular to the main beams [42]. Roofs are commonly mono- to the walls. Even though this is not always the case, it represents a very
pitched roofs or gable roofs with low slope and made with a simple typical condition for vernacular buildings. Moreover, considering no
framework or timber beams. Fig. 2 presents some examples of this beneficial diaphragmatic action of floors and roofs is the most un-
vernacular typology. favorable scenario. The displacements of the elements at the base are
Four reference models were initially prepared based on the pre- fully restrained.
viously described geometric and constructive characteristics, but With respect to the materials adopted for the reference models,
adopting different materials for the walls and varying the number of rammed earth or stone masonry is used for the walls, while timber is
floors in order to better evaluate the influence of the techniques for used for the lintels. Rammed earth buildings are built with the stone
different conditions: (RmdEarth1) one-floor rammed earth building; masonry base course, whose height is finally set at 0.4 m. The stone
(RmdEarth2) two-floor rammed earth building; (StoneMas2) two-floor masonry used for the walls and the base course of the earthen buildings
stone masonry building; and (StoneMas3) three-floor stone masonry is considered as an irregular schist or granite masonry type. The ma-
building. It is noted that the geometry of the models is slightly sim- terial model adopted to represent the nonlinear behavior of the rammed
plified in order to represent generic vernacular buildings. Fig. 3 shows earth and stone masonry, which are the materials considered to present
the reference models where the different traditional earthquake re- nonlinear behavior, is a standard isotropic Total Strain Rotating Crack
sistant solutions will be implemented in order to assess their influence Model (TSRCM). The model describes the tensile and compressive be-
on the seismic behavior of the buildings. Moreover, as previously havior of the material with one stress-strain relationship and assumes
mentioned, additional reference models needed to be built for assessing that the crack direction rotates with the principal strain axes. The
some of the techniques for the study to be more comprehensive. DIANA tension softening function selected is exponential and the compressive
software [47] was used for the construction of the models and to per- function selected to model the crushing behavior is parabolic. It has
form the pushover analyses. All the elements from the reference models been selected because of its robustness and simplicity, and has been
are solid 3D elements: ten-node isoparametric solid tetrahedron ele- already successfully applied in previous analysis of complex stone
ments (CTE30). There are two elements within the thickness of the masonry and earthen structures [25,28,29].
walls, which is established at 0.5 m. The height of the walls is set at 3 m Table 3 presents the material properties adopted for the three

4
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 3. Reference numerical models constructed for the quantitative assessment of the influence of traditional earthquake resistant techniques.

different materials used in the reference models. The material proper- introduction of ring beams is used to improve the connection between
ties required to define the constitutive model were based on data col- the orthogonal walls, tying the building to promote a global ‘box-be-
lected from different authors. An elasticity modulus of 10 GPa and a havior’. The positive effect of the ring beams highly depends on the
Poisson’s ratio of 0.2 were considered for the timber [17,31,51,16]. The connection between the timber elements, but if workmanlike con-
stone masonry compressive strength, elastic properties and specific structed, they can prevent the overturning of the wall and delamination
weight were based on the reference values provided by the Italian code in multiple-leaf masonry walls. Even though this technique is not meant
[39]. With respect to the rammed earth material properties, a com- to avoid the formation of cracks [33], which should initiate for a
pressive strength of 1 MPa, an elasticity modulus of 300 MPa and a loading similar to that observed in unreinforced structures, it can im-
Poisson’s ratio of 0.3 were adopted, based on the scattered values ob- prove the seismic performance of the building in its inelastic range,
served in the literature [5,22,4,17,1,14,32]. The remaining nonlinear sustaining larger deformations without collapsing.
properties of stone masonry and rammed earth are computed directly There are many possible configurations for the ring beams, which
from the compressive strength, based on recommendations given by can vary according to: (1) the dimensions of the timber elements
Lourenço [26]. The compressive fracture energy was obtained using a (a × b); (2) the distance between transversal connectors (s); (3) the
ductility factor d of 1.6 mm, which is the ratio between the fracture presence of diagonal bracing elements; (4) the number of longitudinal
energy and the ultimate compressive strength. The tensile strength was elements (n); or (5) their introduction at different levels within the
set at 1/10 of the compressive strength and an average value of height of the walls, namely at the floor/roof level, at the lintel level or
0.012 N/mm is adopted for the mode I fracture energy. It should be both. Fig. 4 shows the two main configurations evaluated: (A) ring
noted that the material properties shown in Table 3 are used in all the beams with two timbers in parallel at floor and lintel level; and (B) ring
analyses presented hereinafter, unless specified otherwise. beam at the floor level with single timber band and diagonal brace at
the corner. Fig. 4 also shows the abovementioned characteristics that
are modified within each configuration in order to evaluate the influ-
4. Numerical evaluation ence of this traditional technique.
The different configurations and dimensions were decided based on
4.1. Ring beams recommendations suggested by the Indian code [20,21], which con-
templates the use of timber ring beams for improving the earthquake
Ring beams were identified as one of the most widespread tradi- resistance of low strength masonry and earthen buildings. The re-
tional techniques used to improve the connections between structural cognition of the use of this type of traditional strengthening techniques
elements. They consist of a set of timber reinforcement elements, ar- as strengthening techniques, such as timber ring beams and buttresses,
ranged in horizontal planes embedded within the walls, which usually is not common in present-day codes, such as the Eurocode. The Indian
run continuously along the length of the walls. Timber ring beams are and the Nepalese Standards [37,38] are among the few codes ac-
commonly placed on top of the walls and are usually composed of a pair knowledging their potential. The guide for bhatar construction – a
of longitudinal beams located at both edges of the walls and joined traditional construction system from Pakistan that also makes use of
together with small transversal connectors, arranged like a ladder. The timber beams for reinforcing stone masonry walls – prepared by
Schacher [46], was also used as a reference for the specification of
Table 3 variations in the dimension of the different timber elements. All the
Mechanical properties adopted for the three materials used in the reference variations in the configuration of the ring beams were modeled in the
models.
four reference models shown in Fig. 3. Table 4 presents the summary of
Material E ν fc Gfc ft GfI W the 29 models finally constructed in order to assess the influence of the
(MPa) (MPa) (N/mm) (MPa) (N/mm) (kN/m3) ring beams in the seismic behavior of vernacular buildings, specifying
the variations in the dimensions and configuration of the timber ring
Stone masonry 1500 0.2 1.5 2.4 0.15 0.012 20
Rammed earth 300 0.3 1 1.6 0.1 0.012 20 beams elements. Fig. 5 shows some of the most representative models
Timber 10,000 0.2 – – – – 6 constructed and the direction assumed for the analysis (Y direction).

5
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 4. Ring beams characteristics subjected to variations showing two examples of the different configurations evaluated: (a) ring beams with two timbers in parallel
at floor and lintel level; and (b) ring beam at the floor level with single timber band and diagonal brace at the corner.

The embedded timber ring beams are simulated using three-node of vernacular buildings and for the comparative analysis shown below.
3D beam elements (CL18B), which also take shear deformation into
account. The ring beams are considered to be fully connected within the
4.1.1. Numerical results and discussion
wall. Thus, all beam elements share all common nodes with the solid
The models described in Table 4 were subjected to a pushover
elements of the walls. It is noted that considering such a perfect con-
analysis in the direction perpendicular to the walls presenting the
nection between the wall and the timber elements is not always ade-
maximum wall span (Y direction), since it is expected to be the most
quate for vernacular construction. However, this assumption is adopted
vulnerable direction of the building. The addition of a timber ring beam
in order to understand the capability of corner braces to improve the
significantly affects the failure mode of all the models, see Fig. 6. The
seismic resistance of vernacular buildings when they are implemented
reference models showed the typical out-of-plane failure of the exterior
in a workmanlike manner. The cross section of the different elements
back wall. The exterior back walls are more vulnerable in the direction
varies according to the values shown in Table 4.
selected for the analysis (Fig. 6a) due to the movement imposed by the
Concerning the material properties, the elastic mechanical proper-
pushover loads that leads them to rotate and bend outwards, which is a
ties shown in Table 3 were considered. Hitherto, all timber elements
less restricted movement than the inward overturning. However, when
considered in the models were analyzed assuming linear elastic beha-
the ring beams are included, they are able to activate the box-behavior
vior. However, in this case, since the timber elements are meant to
of the building and to transfer the load to the walls parallel to the
reinforce the wall, it was deemed necessary to assume that they are
horizontal load, which led to the development of relevant in-plane
subjected to high stress and can also eventually experience failure.
damage with characteristic diagonal shear cracking. Most of the da-
Thus, following the recommendations of Karanikoloudis and Lourenço
mage accumulates at the corners, where the transfer of forces takes
[23], upper bounds on the ring beam timber elements’ sectional re-
place. When there is also a ring beam at the lintel level, the out-of-plane
sistance were specified, by means of adopting an ideal nonlinear plastic
damage at the walls perpendicular to the loading direction is highly
material model for tension. The value of the maximum tensile strength
reduced and the in-plane resistance of the walls govern the global be-
(ft) adopted is 20.5 MPa and the Von Mises critical yield stress criterion
havior of the building.
was assumed. Only a tensile limit was applied because preliminary
The level of stress at the timber elements was also controlled at the
analyses showed that the highest stresses occur under tension. Even
ultimate limit state. Fig. 7 shows the axial forces (N) at the ultimate
though modeling embedded timber elements in masonry is a very
condition for the same two-floor buildings presented in Fig. 6. As it
complex issue – particularly given the anisotropic nature of wood and
could be expected, the maximum axial force values take place at the
the connections among the elements – this idealization of the nonlinear
middle of the wall span perpendicular to the seismic load, because of
behavior of the timber was considered enough for this study to un-
the outward bending of that wall. However, the tensile stresses are al-
derstand the effect that timber ring beams have on the seismic behavior
ways under the plastic limit adopted. Taking into account the

Table 4
Different models constructed for the evaluation of the influence of timber ring beams on the seismic behavior of vernacular buildings (dimensions in mm).
Model Ring beam timber elements Ring beam position

Name RmdEarth1 RmdEarth2 StoneMas2 StoneMas3 Longitudinal Transversal Diagonal Floor/ Lintel level
roof level
Number (n) Dimensions Dimensions Separation (s) Dimensions
(al × bl) (at × bt) (ad × bd)

Ref X X X X – – – – – – –
A1_Floor X 2 100 × 75 75 × 50 500 – X –
A2_Floor X X X X 2 100 × 75 75 × 50 1000 – X –
A2_FloorLintel X X X X 2 100 × 75 75 × 50 1000 – X X
A3_Floor X 2 75 × 38 50 × 30 500 – X –
A4_Floor X 2 75 × 38 50 × 30 1000 – X –
A5_Floor X X X 2 100 × 75 75 × 50 1000 100 × 75 X –
A5_FloorLintel X X X 2 100 × 75 75 × 50 1000 100 × 75 X X
B1_Floor X X 1 150 × 100 – – – X –
B2_Floor X X 1 150 × 100 – – 100 × 75 X –
B2_FloorLintel X X X X 1 150 × 100 – – 100 × 75 X X

6
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 5. Numerical models built in order to assess the influence of timber ring beams in the seismic behavior of vernacular buildings.

Fig. 6. Representative failure modes in the transversal Y direction obtained for the different models with different configurations of ring beams in terms of: (blue)
maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

dimensions of the beams cross section and the tensile strength plastic direct influence on the maximum capacity of the building. Fig. 9 pre-
limit considered, the ring beams with two bands could reach axial sents four-linear capacity curves constructed for the four buildings
forces up to 150 kN, while the ring beam with one band could reach under analysis. The variations of the load factors defining each limit
300 kN. Fig. 7 also shows that the use of a second ring beam at the lintel state are given in percentage, since they are normalized by the load
level helps to reduce the maximum stresses at the beam. factor obtained for the model with higher capacity (A1_Floor). Results
In order to clarify the influence of the axial capacity of the timber are also compared with the performance of the one-floor reference
beams in the global response of the building, a small parametric ana- model with no ring beam. Traditional timber ring beams can sometimes
lysis was carried out. Keeping the same ring beam configuration (model simply consist of rough timber pieces (trees trunks and branches)
A1_Floor), firstly the original plastic limit (ft = 20 MPa) was twice re- poorly connected and arranged [41]. The maximum capacity of the
duced two and four times. Secondly, the cross-section dimensions of the building increases approximately 2.5 times, when introducing a timber
beams were also reduced (model A3_Floor). Fig. 8 presents the axial beam assuming the original maximum tensile strength, whereas the
forces (N) at the ultimate condition for the four models constructed. It increase is around 1.7 times if the quality of the timber beam elements
shows how the maximum axial forces are also reduced according to the is considered poorer, assuming a tensile strength of around 5 MPa. The
reductions of the plastic limit and cross section dimensions. For in- graph presented in Fig. 9 also shows that the ring beams do not have the
stance, when the plastic limit is reduced four times, the maximum load same influence defining LS1, which means that they do not prevent the
that the beams can resist is 37.5 kN and this is the maximum value appearance of damage.
obtained from the analysis. The four-linear capacity curves constructed from the pushover
The reduction of the maximum capacity of the timber beams has a analyses performed in the two-floor rammed earth models are shown in

7
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 7. Axial forces (N) at the ring beams for the different models with different configurations at the ultimate limit state (scale in N).

Fig. 8. Axial forces (N) at the ring beams for the different one-floor models varying the plastic limit and the cross section dimensions of the ring beam elements, at the
ultimate limit state (scale in N).

Fig. 10a. Fig. 10b shows the variations of the load factors defining each the maximum capacity (RmdEarth_A2_FloorLintel). The graphs show
limit state obtained for the same set of models, taking into account the again the favorable effect of the use of timber ring beams, which, in the
normalization based on the load factor obtained for the model showing best case, can increase the maximum capacity of the reference building

Fig. 9. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the one-floor models; and (b) load factor variations
obtained for each limit state for the one-floor models.

8
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 10. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the two-floor models according to the timber ring beam
configuration; and (b) load factor variations obtained for each limit state for the two-floor models.

improving the connection between structural components by forming

close contours in horizontal planes [42]. To a greater or lesser extent,
the same type of enhancement of the structural behavior of the building
was achieved using the different ring beam configurations evaluated.

4.2. Corner braces

Traditional types of corner braces commonly observed in vernacular

buildings are timber stiffening elements placed usually diagonally at
the corners that are intended to reinforce the wall-to-wall connections
of the building. This technique also improves the post-elastic perfor-
mance of the walls because it aims at keeping the walls working to-
gether even when the joints between perpendicular walls crack during
an earthquake. They are easier to apply as a strengthening solution and
that is why their use is widespread in many seismic prone countries,
such as Chile, Morocco and Peru [1].
A new set of models was prepared aimed at evaluating the structural
Fig. 11. Variation of the load factor leading to the attainment of the maximum influence of this technique in the seismic behavior of vernacular
resistance (LS3) for the four set of models evaluated with different timber ring buildings. Among the many possibilities, three typical types of corner
beam configurations. braces (C) were explored, consisting of: (A) a pair of longitudinal timber
beams embedded within the walls at the corners, as a kind of partial
approximately three times. The best results are obtained for the model ring beam; (B) addition of diagonal timber members to the first type;
with ring beams at both the top and the lintel level but the difference is and (C) an independent diagonal corner brace, attached to the walls
not substantial with respect to the results from the building with the using timber wedges. The first two options were studied placing the
ring beam installed just at the top level. The ring beams with only one corner braces at the floor/roof (Floor) and at the floor/roof and lintel
timber band are slightly less efficient than the ones with two bands, level (FloorLintel), and the third option only at the top. The dimensions
particularly if the diagonal bracing elements are not included. Fig. 10b of the different timber elements were the ones adopted for the assess-
also shows that the use of these diagonal elements has a significant ment of the timber ring beams. Additionally, another set of analysis was
effect on the definition of the load factor leading to LS1 and they ac- performed on buildings presenting an initial deterioration of the wall-
tually help in delaying the initiation of damage. Very similar results to-wall connections integrity. Corner braces are assumed to be parti-
were obtained for the other set of models considering stone masonry as cularly effective for walls showing a poor connection and this extra set
the type of material for the walls and different number of floors. Fig. 11 of analyses is aimed at evaluating their efficiency in restoring the ori-
shows the variation of the load factor corresponding to the attainment ginal strength of the building. The compromised wall-to-wall connec-
of LS3 for the four set of reference models. tions were simulated by reducing the mechanical properties of the walls
Results showed that timber ring beams are very efficient in im- at the corners to 40% and 20%, in terms of percentage of the reference
proving the connection between the structural elements, since they are mechanical properties. This ratio (c) simulates the integrity of the wall-
able to engage all the walls in the structural response of the building to-wall connections.
under seismic loading, thus activating its in-plane response. That is why The three types of corner braces evaluated were modeled in three of
they can increase significantly the maximum capacity of the building, the reference models shown in Fig. 3 (RmdEarth1, RmdEarth2 and
by preventing the premature out-of-plane collapse of the walls. The StoneMas2). The modeling of the corner braces follows the same ap-
effect of timber ring beams is similar to that of proper diaphragms, proach used to evaluate the influence of ring beams using 3D beam
elements (CL18B), assuming a perfect connection between beam

9
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Table 5
Different models constructed for the evaluation of the influence of corner braces on the seismic behavior of vernacular buildings (dimensions in mm).
Model Corner brace type Ring beam position

Name RmdEarth1 RmdEarth2 StoneMas2 RmdEarth2 RmdEarth2 Partial ring beam Diagonal Floor/ Lintel level
(c = 40%) (c = 20%) strut roof level
Longitudinal elements Diagonal
elements

Length (l) Dimensions Dimensions Dimensions

(al × bl) (ad × bd) (ad × bd)

Ref X X X X X – – – – – –
A_Floor X X X X X 1250 100 × 75 – – X –
A_FloorLintel X X X X X 1250 100 × 75 – – X X
B_Floor X X X X X 1500 100 × 75 100 × 75 – X –
B_FloorLintel X X X X X 1500 100 × 75 100 × 75 – X X
C_Floor X X X X X – – – 100 × 75 X –

Fig. 12. Numerical models built in order to assess the influence of timber corner braces in the seismic behavior of vernacular buildings.

elements and solid elements and a plastic limit in tension. It is noted the great concentration of damage at the connection between perpendi-
that the effect of the corner braces on walls showing a degraded wall-to- cular walls is avoided with the use of braces, due to a better redistribution
wall connection was only evaluated for the RmdEarth2 reference model. of stresses. As a result, the out-of-plane bending damage at the center of
Table 5 presents the summary of the 30 different models constructed in the wall is more widespread and also partial in-plane damage takes place
order to assess the effect of corner braces in the seismic response of at the walls parallel to the seismic load, particularly at first floor.
vernacular buildings. Fig. 12 shows some of the most representative The four-linear capacity curves constructed from the pushover
models constructed and the direction adopted for the pushover analysis analyses performed in the two-floor rammed earth models are shown in
(Y direction). Fig. 15a. Fig. 15b shows the variations of the load factors defining each
limit state obtained for the same set of models, again in terms of per-
centage normalized using the load factors found for the model showing
4.2.1. Numerical results and discussion the maximum capacity. The abovementioned improved connection
The models described in Table 5 were tested again in the direction between orthogonal walls achieved with the corner braces is translated
perpendicular to the walls presenting the maximum wall span (Y direc- into a direct increment in the maximum capacity of the building. The
tion). The addition of the corner braces did not significantly affect the increment varies for the different types of braces but the maximum
failure mode of the buildings, which still present the typical out-of-plane almost reaches 30% for the corner brace consisting of a partial ring
overturning of the exterior wall. However, the corner braces help to im- beam with a diagonal stiffener (model RmdEarth2_B_Floor). Despite not
prove the connection between orthogonal walls. The failure modes found being a great increase, this technique can prevent failure at the con-
for the RmdEarth2 buildings, shown in Fig. 13, indicate that damage at the nection. It also has a greater influence in the attainment of LS1, because
corner is prevented and now starts approximately at the corner brace ends it avoids the early onset of damage due to a better redistribution of
in both orthogonal walls. Therefore, corner braces are able to engage the stresses at the corner of the building. The results were very similar for
walls parallel to the seismic load in the building response, but their effect the other two sets of reference models, as shown in Fig. 16, where the
is limited to their length. Thus, they are not sufficient to wholly activate variation of the load factor corresponding to the attainment of LS3 is
their in-plane resistance. The level of stress at the timber elements was displayed. The influence is greater for the two-floor stone masonry
also controlled at the ultimate limit state, but the stresses obtained are buildings due to the combination with the better material properties of
moderate and the timber elements are not prone to fail. the stone masonry, allowing the engagement of the in-plane walls in the
Fig. 14 shows the results in terms of damage pattern obtained for the seismic response of the building, increasing its capacity.
models showing poor wall-to-wall connections. Results show that the Fig. 17a shows the four-linear capacity curves constructed from the
introduction of corner braces effectively prevents the premature separa- pushover analyses on the RmdEarth2 set of models with and without the
tion exhibited by the walls in the reference model with no braces. Besides,

10
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 13. Representative failure modes in the transversal Y direction obtained for the different models with different types of corner braces in terms of: (blue)
maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 14. Damage crack pattern (crack width scale in m) obtained for the different models with deteriorated wall-to-wall connections and with or without corner
braces. Results at the ultimate limit state (LS4).

first type of corner brace (models A_floor). The influence of the corner connection, even though they are not able to activate in-plane resisting
braces on the buildings with a deteriorated connection is clear, since mechanisms in the walls parallel to the pushover load. Thus, braces
they are able to restore almost completely the original strength of the have a more moderate effect in increasing the maximum resistance of
building, and they even outperform the seismic performance of the the building. Results also showed that braces are efficient in restoring
reference model assuming good wall-to-wall connection in terms of the integrity of poor wall-to-wall connections.
maximum capacity. Fig. 17b shows the variation of the load factor
corresponding to the attainment of LS3 when reducing the integrity of 4.3. Quoins
the wall-to-wall connection, with and without the different types of
corner braces studied. The variations are studied for the same Quoins are a typical traditional technique applied to strengthen the
RmdEarth2 set of models, and are normalized using the results from the buildings at the corners, by means of using the best quality large
reference model with no corner brace. The previously discussed bene- squared stone ashlars for the construction of the corners with an effi-
ficial effect is very clear. The corner braces effectively ensure a proper cient interlocking with the orthogonal walls. Quoins can be also made
wall-to-wall connection even if the original joint was compromised. of other materials, such as brick masonry, when used together with
Moreover, if the connection between walls was already good, corner earthen walls. Quoins are a traditional solution that can be commonly
braces also significantly enhance the seismic performance of the found in many stone masonry vernacular buildings composing the
building. This effect highlights their potential as a seismic strength- historic city centers of many seismic prone regions, such as Italy, Greece
ening technique. and Algeria. They are meant to reinforce wall-to-wall connections,
In summary, results showed that corner braces have a clear influ- promoting a proper transfer of forces between orthogonal walls.
ence on the quality of the wall-to-wall connections. They proved to be However, their efficacy can be limited when coupled with poor fabric or
efficient in engaging orthogonal walls and avoiding damage at the internally unconnected masonry which tend to become loose [10].

11
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 15. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the two-floor models according to the corner brace
configuration; and (b) load factor variations obtained for each limit state for the two-floor models.

reference models constructed, marking the elements simulating the

quoins in red and indicating the direction selected for the pushover
analysis (Y direction).

4.3.1. Numerical results and discussion

The models were tested in the direction perpendicular to the walls
presenting the maximum wall span (Y direction). The presence of
quoins does not modify the out-of-plane failure mode observed for all
the models, consisting of the overturning of the exterior wall. The
quoins are nevertheless able to improve the wall-to-wall connection by
preventing the onset of damage at the corner. In this case, the damage
starts at the connection between the quoin and the rest of the wall. The
failure modes for the RmdEarth2 set of models are shown in Fig. 19. The
quoins thus engage the walls parallel to the seismic load, but, when
coupled with materials with low tensile strength, their effect is limited.
Results for the RmdEarth2 set of models are shown in Fig. 20a, in
terms of four-linear capacity curves constructed from the pushover ana-
lyses, and in Fig. 20b, in terms of the variations of the load factors de-
fining each limit state. The load factor variation is normalized using the
Fig. 16. Variation of the load factor leading to the attainment of the maximum results from the reference model in order to quantify the increase
resistance (LS3) for the three set of models evaluated with different corner achieved. There is a clear increment in the maximum capacity of the
brace configurations. building, reaching around 25% when good quality stone masonry is as-
sumed at the quoins (500% increment of the material properties). How-
The modeling strategy adopted to quantitatively evaluate the ever, the greatest variations occur for the load factor defining LS1, which
structural influence of this technique in the seismic behavior of ver- almost doubles for the model with ashlar masonry at the corners. Quoins
nacular buildings consisted of increasing the mechanical properties of are thus efficient in delaying the onset of damage and preventing failure
the walls at the corners. The four reference models shown in Fig. 3 were at the connection. Similar increasing trends and variations in the damage
used to assess the influence of this technique. For the rammed earth patterns were observed on the other sets of reference models. Fig. 21
buildings, the mechanical properties of the walls at the corners are shows the variations of the load factor corresponding to the attainment of
increased twice 200% and 500%, in terms of percentage of the rammed LS3 for every set, which lay within a similar range.
earth material properties shown in Table 3. This intends to simulate the Results showed that the effect of quoins addresses directly the
use of rough stone masonry and ashlar masonry at the corners, re- quality of wall-to-wall connections. Similarly to the previously studied
spectively. For the stone masonry buildings, the properties at the cor- corner braces, they are efficient in engaging orthogonal walls and
ners were increased only 200% of the reference stone masonry prop- avoiding damage at the connection. As a result, they are able to increase
erties from Table 3, simulating the use of ashlars at the corners. The the maximum resistance of the building. Their performance is also
quoins are considered again fully connected to the rest of the masonry comparable to corner braces quantitatively, as the increase achieved in
wall. It is noted that no interface elements were used for the analysis. the maximum capacity of the building with the use of quoins also lies
Despite the simplification of modeling the quoins as areas of the walls within the 20–30% range.
with enhanced material properties, they simulate an efficient inter-
locking between orthogonal walls and an efficient overlapping of the 4.4. Ties
quoin ashlars with the rest of the wall, which can prevent the com-
monly observed failure of the building at the corner. Fig. 18 shows the Ties are one of the most often adopted techniques used in

12
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 17. (a) Results in terms of four-linear capacity curves built based on the computed limit states for the RmdEarth2 models varying the wall-to-wall connection
quality (with and without corner brace A_Floor); and (b) load factor variations obtained for LS3 for the RmdEarth2 models with and without different corner braces.

Fig. 18. Numerical models built in order to assess the influence of quoins in the seismic behavior of vernacular buildings.

vernacular architecture in order to hold together the different structural perpendicular walls. For that matter, a simple steel tie rod of 20 mm
elements of the building. The widespread use of ties is likely because diameter was modeled at the top part of the corners at a height of 2.7 m
they can be easily implemented in existing structures. They can be ty- from the ground or the floor level. Only elastic properties were con-
pically located at different locations of the building, intended to im- sidered, since the failure of the rod was not considered, adopting a
prove the connection among different structural elements, namely: (1) common elasticity modulus of 210 GPa. The ties were modeled as
between perpendicular load bearing walls; (2) between load bearing trusses, using three-node 3D truss elements (CL9TR), in order to be only
walls and interior walls; (3) between parallel load bearing walls; (4) subjected to axial forces. The embedment length of the tie was set at
between walls and floors; and (5) between walls and roofs. Ties con- 1.25 m. The truss elements are considered fully connected to the solid
necting parallel walls or ties connecting floors or roofs to walls are elements from the walls, sharing all common nodes. The ties were
intended to achieve a joint behavior from all the walls subjected to out- connected at the exterior part of the wall using a cross-shaped steel
of-plane actions. They are usually placed at the floor and roof levels. anchor modeled using 3D beam elements (CL18B). This anchor was also
Actually, a common vernacular practice is the use of the floor timber considered to be fully connected to both the walls and the tie.
joists to connect parallel walls by means of using ties to reinforce their Additionally, the same mechanical properties (only elastic) used for the
connection with the walls. This application of ties allows considering tie rods are adopted for the anchor, not considering its possible failure.
that the diaphragm achieve the desired coupling effect between parallel Fig. 22 shows the location of the ties at the one-floor and two-floor
walls, which is shown to improve the seismic resistance of vernacular models. Preliminary analyses showed that ties barely have any influ-
buildings [42]. Ties have to be well fastened at the ends typically using ence in the seismic behavior of the buildings in terms of maximum
steel anchor plates. Otherwise, they can be counterproductive and in- capacity or damage pattern. Thus, in a second step, the ties were
duce significant stress concentration, causing cracking [48]. modeled in buildings showing an initial deterioration of the integrity of
The present paper will only evaluate the influence of ties connecting the wall-to-wall connections, aiming at evaluating if ties can help in

13
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 19. Failure modes in Y direction obtained for the RmdEarth2 models with quoins presenting different increases of the corners material properties (c), in terms of:
(blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 20. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2 models; and (b) load factor variations
obtained for each limit state for the RmdEarth2 models.

partially restoring their original strength. The compromised wall-to- introduction of ties, damage still mostly concentrates at the connection
wall connections were simulated using the approach used for the as- between perpendicular walls, showing their separation. Nevertheless,
sessment of the corner braces. The mechanical properties of the walls at they are able to involve a bigger portion of the wall parallel to the
the corners were reduced to 40% and to 20%. The effect of ties linking seismic load, showing also significant damage at the location where the
walls showing a degraded connection was evaluated using three out of tie rod ends.
the four reference models shown in Fig. 3: RmdEarth1, RmdEarth2 and This redistribution of stresses achieved by the ties has a beneficial
StoneMas2. effect in the results in terms of maximum capacity. Fig. 24a shows the
four-linear capacity curves constructed from the pushover analyses for
the RmdEarth2 set of models. Fig. 24b shows the variation of the load
4.4.1. Numerical results and discussion
factor corresponding to the attainment of LS3 when reducing the in-
The models were tested in the direction perpendicular to the walls
tegrity of the wall-to-wall connection, with and without ties, normal-
presenting the maximum wall span (Y direction). The presence of tie
ized using the results from the reference model with no ties. Even if the
rods affected slightly the failure mode of the models since they helped
increment in the maximum capacity of the building is low (around
in partially redistributing the stresses at the corner. Fig. 23 presents
10–15%), there is a noticeable beneficial effect. Ties mainly affect the
representative failure modes for the two-floor buildings. After the

14
J. Ortega et al. Engineering Structures 173 (2018) 1–27

[50]. Timber bands should be inserted within the entire wall thickness,
which allows avoiding the separation of the leaves in multiple-leaf
masonry walls and prevents crack propagation. The horizontal re-
inforcement of load bearing masonry walls with timber elements is one
of the most common vernacular practices in many seismic prone re-
gions throughout the world, such as Greece [50], Turkey, where the
construction system receives the name of ‘hatil’ [19], and India, where
is known as ‘taq’ [24].
Nonetheless, the same modeling approach used for the simulation of
the ring beams was followed to assess the efficiency of this timber-laced
masonry structural type. New numerical models were prepared in-
serting continuous timber bands along the height of the walls with a
vertical spacing of 1 m. Only one configuration was considered. The
horizontal reinforcement was arranged using two longitudinal beams
joined together with small transversal connectors. The cross-section
dimensions of the timber elements were 100 mm × 75 mm for the
longitudinal elements and 75 mm × 50 mm for the transversal ele-
Fig. 21. Variation of the load factor leading to the attainment of the maximum
ments, which were placed each 1 m. This configuration was introduced
resistance (LS3) for the five set of models evaluated with different type of
in the same four reference models presented in Fig. 3.
quoins.

Fig. 22. Position of the ties (marked in red) at the numerical models built in order to assess the influence of ties connecting perpendicular walls in the seismic
behavior of vernacular buildings. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

seismic performance of the building in its inelastic range, as they in- 4.5.1. Numerical results and discussion
crease the capacity of undergoing larger deformations sustaining the The models were tested in the direction perpendicular to the walls
same load-carrying capacity of the building. Only the variations in the presenting the maximum wall span (Y direction). As expected, the
load factor defining LS3 are evaluated because the ties are not effective presence of the timber reinforcement forming the timber laced walls
in avoiding the formation of cracks and do not have a significant in- had a crucial influence on the seismic behavior of the rammed earth
fluence for LS1. Similar increasing trends and variations in the damage building and its failure mode, see Fig. 26. The timber elements are able
patterns can be observed in Fig. 25, which shows the variations of the to tie the walls, leading to an almost monolithic response. That is why
load factor defining LS3 for the other two sets of reference models, with the greatest damage concentrates at the base, promoting the rotation of
and without ties. In conclusion, ties connecting perpendicular walls the building as a block. At ultimate condition there is, in any case,
have shown to be efficient in partially restoring the integrity of the extensive in-plane and out-of-plane damage widespread throughout the
wall-to-wall connections that are deficient or degraded. building, which remarks the efficiency of this technique to redistribute
the stresses along the structure. The level of stress at the timber ele-
4.5. Timber elements within the masonry ments is also shown at Fig. 26 at the ultimate state. The maximum axial
force values occur at the mid span of the wall perpendicular to the
Besides the use of timber ring beams at the top of the load bearing seismic load, because of its outward bending, reaching around 75 kN,
walls, timber elements can be inserted along the height of the wall as a far below the plastic limit considered for the analysis.
horizontal reinforcement, constituting a timber-laced masonry that can The influence of this technique in the seismic behavior of the
be understood as a different type of construction system. The insertion building is also very clear in terms of maximum capacity. Fig. 27a
of timber elements systematically along the height of the wall is aimed shows the four-linear capacity curves constructed from the pushover
at stabilizing the masonry walls and not only at improving the wall-to- analyses for the RmdEarth2 model and Fig. 27b shows the variation of
wall connection. That is why this technique is analyzed independently the load factor defining each limit state. The introduction of the timber
of the ring beams. Timber elements are applied due to their excellent reinforcement increases the maximum capacity of the building up to
tensile properties and because, by confining portions of masonry walls, approximately three times. Also, the use of timber lacing elements also
it enhances their compressive strength and deformability properties helps in delaying the crack propagation, showing a notable influence on

15
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 23. Failure modes in Y direction obtained for the RmdEarth2 models with different level of deterioration of the wall-to-wall connection (c) with and without ties,
in terms of: (blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For
interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

the definition of LS1. The similarity in the results obtained when in- distinct structural typology.
troducing this reinforcement in the other three models can be observed
in Fig. 28, in terms of the variation of the load factor corresponding to
the attainment of LS3. 4.6. Wall subdivisions
The numerical evaluation of the effect of introducing timber ele-
ments as continuous horizontal reinforcement embedded within the Horizontal courses of brick masonry can sometimes be recognized in
load bearing walls has a significant impact on the seismic behavior of the Portuguese vernacular practice and in other European countries.
the building. Even though the application of this construction technique was most
The addition of timber laces is extremely efficient in tying the walls likely not originally conceived as a seismic resistant measure, the
together and thus making the building respond as a unit. The effect is subdivision of the wall can be an effective way to stabilize the masonry
directly reflected in an increase of the maximum resistance of the and earthen walls, as well as to help to arrest crack development. Thus,
building of approximately three times. The insertion of timber elements their efficiency in improving the seismic behavior of vernacular
completely alters the seismic response of the reference building and the buildings has been assessed.
structural response of the walls. Therefore, it can be considered as a The effect of the use of brick horizontal courses extending through
the thickness of the walls on the seismic behavior of vernacular

Fig. 24. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2 models; and (b) load factor variations
obtained for LS3 for the RmdEarth2 models with and without ties.

16
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 25. Variation of the load factor leading to the attainment of the maximum resistance (LS3) for: (a) RmdEarth1; and (b) StoneMas2 set of models; reducing the
integrity of the wall-to-wall connection with or without ties.

Fig. 26. Failure modes in the transversal Y direction obtained for the RmdEarth2 model with the timber reinforcement, in terms of: (blue) maximum total dis-
placements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the references to colour in this
figure legend, the reader is referred to the web version of this article.)

buildings was evaluated considering only one configuration, which was horizontal courses were simulated using solid 3D elements (CTE30),
introduced in the same four reference models (Fig. 3. Brick courses adopting typical mechanical properties corresponding to solid brick
were inserted each 1 m along the height of the walls assuming that the masonry [39], which are shown in Table 6. The brick bands are also
thickness of the courses is 0.2 m, simulating 2–3 layers of bricks. Brick considered to be fully connected to the rest of the wall, sharing all

Fig. 27. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2 models with and without the timber
elements; and (b) load factor variations obtained for each limit state for the RmdEarth2 models.

17
J. Ortega et al. Engineering Structures 173 (2018) 1–27

wall perpendicular to the seismic load. Nevertheless, the improvement

in the material properties of the walls at the bands helps to slightly
reduce the extension of the damage. Fig. 30 presents the failure modes
obtained for the two-floor rammed earth models with and without brick
horizontal courses and the two-floor brick masonry model. The pro-
gressive reduction of the damage as the mechanical properties of the
material composing the walls increase is evident.
Fig. 31a shows the four-linear capacity curves constructed from the
pushover analyses performed in the same three models and Fig. 31b
shows the variation of the load factor defining each limit state. The
influence of the introduction of brick bands is also clearly reflected in
these graphs. The model with brick bands shows a performance in be-
tween the other two models, in terms of capacity, stiffness and ductility.
However, since the rammed earth material is still predominant in the
walls the response is still more similar to the one of the reference
RmdEarth2 model. Focusing on the variation of the load factor defining
Fig. 28. Variation of the load factor leading to the attainment of the maximum LS1, the variation is marginal, which suggests that the bands are not
resistance (LS3) for the four set of models evaluated when introducing em- efficient in avoiding the development of cracks.
bedded timber elements within the walls.
Finally, Fig. 32 shows the results for the other three models, in terms
of the variation of the load factor corresponding to the attainment of
Table 6 LS3. The response of the building is always intermediate between the
Mechanical properties adopted for the brick masonry horizontal courses. response of a building with walls built solely with the original material
Material E ν fc Gfc ft GfI W and the response of the building with brick masonry walls. Therefore,
(MPa) (MPa) (N/mm) (MPa) (N/mm) (kN/m3) for the stone masonry model, whose material properties are more si-
milar to the brick masonry properties, the influence is considerably
Brick masonry 2000 0.2 2 3.2 0.2 0.012 18
smaller.

common nodes. It is noted that no interface elements were considered 4.7. Buttresses
between both elements. Fig. 29 shows the reference models con-
structed, marking the elements simulating the brick courses in red and Buttresses are one of the most common reinforcement techniques
the direction of the pushover analysis. It is noted that two extra models traditionally used in vernacular architecture. The main working re-
were prepared assuming that the whole building is built in brick ma- sisting principle consisting of counteracting the rotation of the façade
sonry: BrickMas1 and BrickMas2, in order to compare the performance has always been very intuitive for vernacular builders and, as a result,
of the three buildings, with and without brick bands and using the two buttresses were widely applied in most seismic prone regions.
different materials composing the walls. Buttresses typically consist of pier-like massive local additions of ma-
sonry elements that are built at the same time as the building or added
as a reinforcement measure. When adding a buttress to an existing
4.6.1. Numerical results and discussion construction, special attention should be given to its connection with
The models were tested in the direction perpendicular to the walls the original walls. If they are not properly connected, their effect in the
presenting the maximum wall span (Y direction). The brick bands are global behavior of the building is reduced or null, acting independently.
not as efficient as the previously studied embedded timber elements in They can show different configurations and locations. Therefore, the
lacing the building together and they do not change the failure mode of evaluation of the effect of buttresses on the seismic behavior of the
the building, which is driven by the out-of-plane failure of the exterior building is a complex matter, as there are many possible variables

Fig. 29. Numerical models built in order to assess the influence of subdividing the wall using brick horizontal courses in the seismic behavior of vernacular buildings.

18
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 30. Failure modes in the transversal Y direction obtained for the RmdEarth2 model with and without brick horizontal course and for the BrickMas2 model, in
terms of: (blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For inter-
pretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

defining these elements. This section tries to provide a comprehensive models and thus, these properties were kept the same for the rest of the
analysis of the influence of buttresses focusing on the variations of two parametric analysis.
main characteristics: (a) dimensions; and (b) position within the First of all, regarding the variations of the dimensions of the buttress
building. With respect to the material, it was assumed that the but- (B), the thickness and length are modified. The initial dimensions were
tresses are built using a low-quality irregular masonry class. Based on decided based on recommendations suggested by the Indian code [20],
reference values proposed by the Italian code [39], an elasticity mod- which suggests the use of buttresses for walls longer than 10 times their
ulus of 1000 MPa was thus initially considered. The compressive thickness. The thickness of the buttress (tb) is recommended to be equal
strength was estimated as 1/1000 of the elasticity modulus and the to the wall thickness (t). The length of the buttress at the bottom (lb) is
remaining nonlinear properties were computed directly from the com- suggested to be at least three times the thickness of the wall. This length
pressive strength, based on recommendations given by Lourenço [26], is kept constant until the buttress is 0.4 m high and then it reduces
following the same approach previously discussed. Preliminary ana- progressively until reaching zero at the buttress maximum height (hb).
lyses varying the initially considered material properties of the but- With respect to the variations in the position of the buttress, it is
tresses resulted in minimal variations of the seismic behavior of the noted that buttresses can show many different configurations. Even

Fig. 31. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2, RmdEarth2_BrickBands and BrickMas2
models; and (b) load factor variations obtained for each limit state for the RmdEarth2, RmdEarth2_BrickBands and BrickMas2 models.

19
J. Ortega et al. Engineering Structures 173 (2018) 1–27

possibilities are evaluated: (a) buttresses located at an asymmetric po-

sition (Asym); and (b) buttresses located at the exact mid-span (Mid).
Concerning buttresses located at the corners, aimed at preventing the
separation of the walls, four different configurations are studied: (c) one
buttress placed at one corner; (d) two buttress placed at both corners;
(e) one buttress placed at one corner diagonally; and (f) two buttress
placed at both corners diagonally. Fig. 33 shows some of the most re-
presentative models constructed.
The variations in the characteristics and configuration of the but-
tresses were modeled in three of the four reference models shown in
Fig. 3: RmdEarth1, RmdEarth2 and StoneMas2. However, a new re-
ference model was constructed with a longer wall of 10 m
(RmdEarth1_s10), in order to better evaluate the influence of buttresses
when located at different positions within the span of the wall. The in-
plan area of the new reference model is 11 m × 5.5 m and the rest of the
characteristics are kept the same. Table 7 presents the summary of the
Fig. 32. Variation of the load factor leading to the attainment of the maximum 29 different models constructed in order to assess the influence of
resistance (LS3) for the four set of models evaluated when introducing hor- buttresses in the seismic behavior of vernacular buildings, specifying
izontal brick masonry courses within the walls.
the variations in the dimensions and configuration of the buttresses. It
is noted that the buttresses are modeled using the same solid 3D ele-
though they can be more effective when placed at critical locations, ments (CTE30) adopted for the walls and they are considered to be
such as the mid-span of long walls, which are the most vulnerable perfectly connected to the walls. It is acknowledged that this is not
elements to out-of-plane movements, they are many times located always the case in vernacular constructions but the ideal condition of
asymmetrically within the wall span or at the corners, reinforcing the the functioning of a buttress is addressed within this study.
connection between the walls. It is noted that despite their undeniable
positive effect counteracting the earthquake horizontal loads, these
elements are many times originally devised to counteract the roof loads. 4.7.1. Numerical results and discussion
Thus, their location is not always the most appropriate concerning the The models described in Table 7 were always tested in the direction
seismic design. In order to be as comprehensive as possible, this study perpendicular to the walls presenting the maximum wall span (Y di-
aims at analyzing several possible configurations. Eventually, two lo- rection). First of all, it should be noted that when the buttresses were
cations were considered: (1) within the span of the wall (Span); and (2) modeled at the mid-span of the longest walls, the global failure of the
at the corners (Corner). Concerning buttresses located within the span building changed and was driven by the out-of-plane failure of the other
of the long walls, aimed at stopping their out-of-plane rotation, two wall perpendicular to the seismic load, which showed inward bending

Fig. 33. Representative numerical models built in order to assess the influence of buttresses in the seismic behavior of vernacular buildings.

20
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Table 7
Different models constructed for the evaluation of the influence of buttresses on the seismic behavior of vernacular buildings.
Models Buttresses configuration

Name RmdEarth1 RmdEarth1_s10 RmdEarth2 StoneMas2 Dimensions (m) Location

Thickness (tb) Length (lb) Height (hb) Span Corners

a b c d e f

Ref X X X X – – – – – – – – –
B1_SpanAsym X 0.5 1.5 2.4 X
a
B1_SpanMid X X X X 0.5 1.5 2.4 (5) X
B2_SpanMid X X 0.5 1 2.4 X
B3_SpanMid X X 1 1.5 2.4 X
B4_SpanMid X X X X 1 1 2.4 (5)a X
B1_Corner1 X X X 0.5 1.5 2.4 (5)a X
B1_Corner2 X X X 0.5 1.5 2.4 (5)a X
B1_Corner1D X 0.5 1.5 2.4 X
B1_Corner2D X 0.5 1.5 2.4 X
B4_Corner1 X X 1 1 5 X
B4_Corner2 X X 1 1 5 X

a
In parenthesis the height of the buttresses in the two-floor buildings.

and overturned towards the interior of the building. The inward out-of- maximum free spans resulting after the addition of the buttresses.
plane failure was not taken into consideration in the analysis because Buttresses do not change the initial out-of-plane failure of the wall, but
this study is focused on evaluating the efficiency of the buttresses. are effective in avoiding the overturning of the exterior wall observed
Buttresses are significantly more efficient when counteracting the ro- when the free span is 10 m. The failure mode involves more bending
tation of the façade outwards. Thus, this initially obtained failure was damage widespread throughout the wall, not only accumulated at the
prevented by not modeling the wall failing inwards. Disregarding the connection between the walls and at the base.
inward rotation allowed isolating the effect of the buttresses in the wall The effect of the addition of the buttress is also clearly reflected on the
that has been reinforced with its addition. This approach also allows maximum capacity of the building. Fig. 35 presents the four-linear ca-
comparing the results between the different models for different but- pacity curves constructed for the RmdEarth1_s10 set of models and the
tresses configuration. Nevertheless, the fact that the buttress already variations of the load factors defining each limit state, in terms of per-
prevents the typical outward out-of-plane failure mode observed in the centage normalized using the results from the model with the buttress
reference model already shows an improvement in the global seismic located at the mid-span of the wall. Results confirmed the efficiency of the
response of the building. buttress in avoiding the overturning and improving the seismic response
Fig. 34 presents representative failure modes for the RmdEarth1_s10 of the building. The location of the buttress at the mid-span almost
model with the different positions of the buttresses along the wall span. doubles the capacity of the reference model. The buttress at the mid-span
The results are also compared with the failure modes of buildings with is able to improve the seismic capacity of the building to a greater extent
no buttresses presenting a maximum span of 5 and 7 m, which are the than simply reducing the wall span, since the addition of the buttress

Fig. 34. Representative failure modes in the transversal Y direction obtained for the different RmdEarth1_s10 models with different locations of the buttress in terms
of: (blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of
the references to colour in this figure legend, the reader is referred to the web version of this article.)

21
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 35. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth1_s10 models; and (b) load factor
variations obtained for each LS for the RmdEarth1_s10 models.

Fig. 36. Failure modes in the transversal Y direction obtained for the RmdEarth2 models with buttresses of different size at the mid-span, in terms of: (blue) maximum
total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article.)

changes the boundary conditions and the confinement of the wall. The loss, as shown by model RmdEarth1_s10_B4_SpanMid, whose length is
outward out-of-plane movement imposed by the pushover load on a wall two times the wall thickness but whose thickness is also twice the wall
spanning 5 m is more restricted with the presence of the buttress. This thickness (Fig. 35b).
leads to an increase in the maximum capacity of the building with but- The effect of buttresses located within the span of the wall was
tress (model RmdEarth1_s10_B1_SpanMid) with respect to a building with a further explored on the remaining reference models. Fig. 36 shows the
maximum wall span of 5 m and no buttress (RmdEarth1_s5) of around failure modes for the RmdEarth2 models with buttresses of different size
20%, see Fig. 34c and d. The same occurs with models placed at the mid-position of the wall. Reducing the maximum wall
RmdEarth1_s10_B1_SpanAsym and RmdEarth1_s10_Ref. The maximum wall span with a buttress is translated into a much stiffer response of the
span of both models is 7 m, but the maximum capacity of the model with building without modifying the original out-of-plane bending failure
the buttress is around 20% greater, see Fig. 34a and b. mode of the wall perpendicular to the seismic load. Nevertheless, an
In terms of the buttress dimensions, results show that the variation incipient out-of-plane collapse mechanism of the corner developed.
in the response is not critical. There are four models with Fig. 37 presents the four-linear capacity curves constructed for the
buttress of different dimensions located at the mid-span RmdEarth2 buildings and the variations of the load factor corresponding
(models RmdEarth1_s10_B1_SpanMid, RmdEarth1_s10_B2_SpanMid, to the attainment of LS3 for the different sets of reference models. These
RmdEarth1_s10_B3_SpanMid and RmdEarth1_s10_B4_SpanMid). The load results confirm that the effect of buttresses is similar for the different
factor corresponding to the maximum capacity of the building obtained buildings.
is similar for all of them but slightly decreases when reducing the di- Buttresses located at the corners of the building have a different
mensions of the buttress (Fig. 35b). Results show that the re- structural role and are mainly intended to avoid the separation of
commendations provided by the Indian code [20] are adequate. A perpendicular walls. They are indeed effective in limiting the damage at
minimum thickness of the buttress equal to the wall thickness and a the connection, proving to be a successful solution, see Fig. 38. Their
length of three times the wall thickness should be considered. Model influence on the seismic behavior of the building has also an impact in
RmdEarth1_s10_B2_SpanMid, which presents a buttress with a length of the capacity of the building and the stiffness of its response, as it can be
two times the wall thickness, already reduces around 10% the increase observed in the capacity curves resulting from the analysis in Fig. 39a.
in the maximum capacity (Fig. 35b). However, if the thickness of the Placing the buttresses at the corners leads to an increase in the max-
buttress is greater, the length of the buttress can be reduced with no imum capacity of around 40%, see Fig. 39b. The addition of only one

22
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 37. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2 models; and (b) variation of the load
factor leading to the attainment of the maximum resistance (LS3) for the different sets of models evaluated with buttresses of different size at the mid-span.

Fig. 38. Representative failure modes in the transversal Y direction obtained for the RmdEarth2 models with buttresses of different size at the corners, in terms of:
(blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the ultimate limit state (LS4). (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

buttress at one corner already has a positive impact on the global at both corners.
seismic behavior of the building. The damage at the wall-to-wall con- The influence of placing buttresses at the corners was also evaluated
nection where the buttress is located is reduced and the maximum ca- on the remaining reference models. The use of diagonal buttress, for
pacity of the building increases around 10–20%, depending on the di- instance, was explored in the RmdEarth1 model. Results showed that
mension of the buttress. Nevertheless, as expected, the best their effect was the same as using buttresses perpendicular to the wall,
performance is shown by the building where the buttresses are placed see Fig. 40a, which shows the capacity curves constructed for the

Fig. 39. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth2 models according to the buttress
configuration; and (b) load factor variations obtained for each limit state for the RmdEarth2 models.

23
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 40. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth1 models; and (b) variation of the load
factor leading to the attainment of the maximum resistance (LS3) for the RmdEarth2 and StoneMas2 sets of models evaluated with buttresses of different size at the
corners.

RmdEarth1 models. Also, the results obtained for the StoneMas2 models wall mainly contributes to reduce effectively the maximum free span of
were very similar to the results obtained for the rammed earth models, the walls subjected to out-of-plane loading. Nevertheless, the effect
as it can be observed in Fig. 40b, which shows the comparison of the observed is stronger than simply reducing the maximum wall span
variation of the load factor corresponding to the attainment of LS3 for because of the new boundary conditions. With respect to the buttresses
both models. The influence of the size of the buttress is greater when placed at the corners of the building, their effect directly addresses the
they are placed at the corners, particularly concerning their thickness. quality of the wall-to-wall connections and they are able to improve the
Finally, another model was constructed based on some of the most maximum capacity of the building.
typical retrofitting elements identified in Portuguese vernacular archi-
tecture [9]. A particular configuration observed was the simultaneous
use of buttresses and stone plinths at the outer base of the walls. These 4.8. Walls thickening
traditional constructive elements were added to model RmdEarth1 in
order to evaluate their influence on the seismic behavior of the The final reinforcement that was assessed was the traditional
building. Fig. 41 shows the model and the results in terms of capacity thickening of the walls by adding mass to construct scarp walls with
curves, comparing the performance of the building with and without varying thickness, intended to lower the center of gravity of the
the new elements. The maximum capacity of the building almost dou- buildings. This technique also increases the resisting area of the wall
bles the original capacity. This assessment exemplifies how common and reduces their height-to-thickness ratio, which improves their out-
traditional solutions, if well executed, can significantly reduce the of-plane resistance and reduces the possibility of overturning. To assess
seismic vulnerability of vernacular architecture. the efficiency of this technique, the walls of the reference models were
In conclusion, the effect of buttresses located within the span of a thus modified geometrically to become inclined with the thickness
decreasing from the bottom to the top. Three configurations were

Fig. 41. (a) Numerical model built in order to assess the influence of a typical configuration of different reinforcement elements observed in Portuguese vernacular
architecture; and (b) results in terms of four-linear capacity curves constructed based on the computed limit states.

24
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Fig. 42. Failure modes in the transversal Y direction obtained for the reference RmdEarth1 with constant wall thickness (t = 0.5 m) and the model with scarp walls
with varying thickness (t = 0.6–0.4 m), in terms of: (blue) maximum total displacements (scale in m); and (red) crack pattern (crack width scale in m). Results at the
ultimate limit state (LS4). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

evaluated varying the wall thickness at the bottom (t1) and the wall building by changing the wall height-to-thickness and thus the resisting
thickness at the top (t2): (a) t1 = 0.6 m and t2 = 0.4 m; (b) t1 = 0.6 m area of the wall, while reducing the mass of the wall at the top.
and t2 = 0.5 m; and (c) t1 = 0.5 m and t2 = 0.4 m. These variations
were introduced in the one-floor rammed earth building reference
model (RmdEarth1) and in a new reference model considering stone 5. Summary
masonry walls (StoneMas1).
The results from the individual numerical studies are finally sum-
marized in this section with a particular focus on the particular con-
4.8.1. Numerical results and discussion struction and structural aspects of the buildings that can be improved
The models were again analyzed in the direction perpendicular to with the use of the different traditional earthquake resistant techniques.
the walls presenting the maximum wall span (Y direction). The results Table 8 presents this summary and also provides a direct comparison
showed clearly the beneficial effect of using scarp walls. The failure between the techniques, which are arranged according to the range of
mode remains unchanged with respect to the reference building, as it is increase of the load factor corresponding to the attainment of the
shown in Fig. 42, but it occurs at a higher load. Fig. 43a presents the maximum resistance of the building (LS3). Table 8 also indicates the
capacity curves constructed for the RmdEarth1 models, showing clear applicability of the different techniques for strengthening existing
variations in terms of maximum capacity and stiffness. Results are also buildings. The applicability is divided into two possibilities, namely
compared with the performance of buildings with constant wall thick- ‘difficult’ when is more appropriate to implement the technique at the
ness of 0.4 and 0.6 m. The best results are obtained from the model that time of the construction or during partial reconstruction, and ‘practical’
presents the greatest variation of the thickness along the height, which when the technique can be and have been successfully applied for
is due to the fact that the resisting cross section increases, while the strengthening [41].
mass at the top of the wall is reduced. The increment in the maximum The influence of ring beams is qualitatively and quantitatively
capacity of the building with this configuration with respect to the comparable to the influence of horizontal diaphragms, since they are
building with walls with a constant thickness of 0.6 m is around 20%. very effective in providing a proper diaphragmatic action by engaging
Results are practically the same for the stone masonry model, see in-plane walls in the seismic response of the building, activating their
Fig. 43b, which presents the variation of the load factor corresponding in-plane resistance. This can lead to a significant increase of the max-
to the attainment of LS3 for both models. Results thus confirm that imum capacity of the reference building, up to 150%. However, the
scarp walls are effective in increasing the seismic capacity of the installation of ring beams as a strengthening technique in existing

Fig. 43. (a) Results in terms of four-linear capacity curves constructed based on the computed limit states for the RmdEarth1 models; and (b) variation of the load
factor leading to the attainment of the maximum resistance (LS3) for the RmdEarth1 and StoneMas1 sets of models evaluated.

25
J. Ortega et al. Engineering Structures 173 (2018) 1–27

Table 8
Summary of the influence of the traditional earthquake resistant techniques on the seismic performance of vernacular buildings.
Technique Improved structural aspect Range of LS3 increase rate Applicability as strengthening technique Section

Difficult Practical

1. Techniques improving the connection between structural elements

Ring beams Diaphragm 195–249 X 4.1
Corner braces Wall-to-wall connections 127–246 X 4.2
Quoins Wall-to-wall connections 110–120 X 4.3
Ties 4.4
– between perpendicular walls Wall-to-wall connections 105–108 X

2. Techniques stabilizing structural elements and buildings

Timber elements within the masonry Walls typology 251–305 X 4.5
Wall subdivision Walls typology 107–132 X 4.6

3. Techniques counteracting horizontal loads

Buttresses 4.7
– Within the span of the wall Reduction of wall span 180–199 X
– At the corners Wall-to-wall connections 108–136 X
Walls thickening Reduction of wall slenderness 112–131 X 4.8

buildings is difficult, since it would require the raising and removal of effect is not taken into consideration by the numerical models. The
the roof. Corner braces are easier to install and are also efficient in main drawback of these techniques is that they are hardly applicable
connecting orthogonal walls. They have a more moderate effect in in- other that at the time of construction or during partial reconstruction.
creasing the maximum resistance of the building (within the 25–40% The effect of the buttress located within the span of the wall is
range), but are very effective in restoring the integrity of deteriorated or stronger than simply reducing the maximum wall span. However, it is
poor wall-to-wall connections. All the different corner braces config- acknowledged that the analyses performed always considered a perfect
urations evaluated achieved a similar improvement in the structural connection between buttress and wall, which is not always the case.
behavior of the building under seismic loading, in terms of both damage Nevertheless, they are able to almost double the overall resistance of
pattern and maximum capacity. Nevertheless, the importance of the the building. Placing buttresses at the corners of the building directly
good state of conservation of the timber elements for their proper be- affects the quality of the wall-to-wall connections. Although the re-
havior as earthquake resistant techniques should be highlighted. The sponse of the building varied according to the size of the buttress, they
analyses presented have not taken into account the material decay were able to effectively improve the maximum capacity of the reference
because they were focused on assessing the efficiency of these techni- building within a 10–30% range, even when only one buttress was
ques when they are properly built and effectively maintained, showing modeled. Buttresses can also easily added to existing constructions, as
their full potential. In the case of decayed elements, their influence on well as intuitive, which explains their high popularity. Finally, the ef-
the seismic behavior of vernacular buildings should be neglected since fect of thickening the wall is directly associated with a change in the
they will most likely fail in providing any extra resistance either be- wall slenderness. Results showed that the performance of the building is
cause of a degraded connection or a loss of strength capacity. better when considering an inclined wall than having a wall with
Quoins also address wall-to-wall connections. They efficiently en- constant thickness and can increase the maximum capacity of the
gage orthogonal walls and avoid the concentration of damage at the building up to 30%.
connection. In quantitative terms, the increase obtained in the max-
imum resistance of the building after the introduction of quoins is si- 6. Final comments and general recommendations
milar to the increase obtained with the use of corner braces, lying
within the 20–30% range. However, this technique is practically im- The present paper presented a numerical assessment of the effi-
possible to be implemented other than at the time of construction or ciency in reducing the seismic vulnerability of vernacular constructions
during partial reconstruction. Ties connecting perpendicular walls do of different traditional earthquake resistant solutions identified from
not improve so significantly the maximum capacity of the building (less literature review as the result of a local seismic culture. The paper has
than 10%), but results showed that they are efficient in partially re- mainly targeted a representative Portuguese vernacular rammed earth
storing the integrity of the wall-to-wall connections that are degraded. construction type, but has also evaluated the effect of these earthquake
At the same time, they can be easily install in existing buildings, which resistant techniques on stone masonry buildings. The study is thus also
has led them to be one of the most widespread traditional strengthening restricted to the geometric and constructive characteristics that are
techniques observed throughout the world. common for this type of buildings. It should be highlighted that, in spite
The use of timber elements within the masonry and the subdivision of the particularities of the vernacular construction types used as a
of the wall are techniques that address directly the type of material reference, they share many characteristics with other vernacular con-
composing the walls and completely alter the structural typology, but struction systems throughout the world, especially in the South
both of them are only practically applicable at the time of construction Mediterranean region. This is particularly evident at a structural level,
or during partial reconstruction. The insertion of continuous horizontal since the structural system is in most cases conceptually the same and
timber elements within the walls ties the walls together and achieves consists of load bearing walls as the main vertical resisting elements.
the desired monolithic behavior of the building. This results in in- Therefore, the results obtained in this work may be extrapolated to
creasing the maximum resistance of the building up to three times. The other similar structures outside the Portuguese context, such as adobe
introduction of brick courses is not as efficient, but also allows en- masonry or cob constructions in the case of earthen buildings.
hancing the seismic behavior of the building a middle way between Detailed FE models were prepared simulating the different techni-
what would suppose constructing the walls entirely with the material ques and possible variations. Pushover analyses were then conducted
used for the horizontal courses, reaching an increase in the maximum on the resulting model so that the influence of each technique could be
capacity up to 30%. It is acknowledged that both techniques also help quantitatively evaluated. The extensive numerical study presented in
in reducing the risk of delamination of multiple-leaf walls, but this this paper mainly contributes to the deep understanding of the seismic

26
J. Ortega et al. Engineering Structures 173 (2018) 1–27

behavior of these empirically developed traditional techniques. Results [18] Homan J. Seismic cultures: myth or reality? In: Proc. of second international con-
have primarily helped to validate the efficiency of the techniques. Some ference on post-disaster reconstruction: planning for reconstruction, Coventry
University, UK; 2004.
of them are extraordinarily efficient in increasing the capacity of ver- [19] Inan Z. Runner beams as building element of masonry walls in Eastern Anatolia,
nacular buildings, particularly those meant to improve the connections Turkey. In: Correia M, Carlos G, Rocha S, editors. Vernacular heritage and earthen
architecture: contributions for sustainable development. London, UK: Taylor &
between structural elements, such as ring beams, which, by tying the Francis Group; 2014. p. 721–6.
building and improving its ‘box-behavior’, led to increments of the [20] IS 13827. Improving earthquake resistance of earthen buildings – Guidelines.
building maximum capacity of almost three times. Bureau of Indian Standards (BIS), New Delhi, India; 1993.
[21] IS 13828. Improving earthquake resistance of low strength masonry buildings –
The results obtained were finally summarized according to the ef- guidelines. Bureau of Indian Standards (BIS), New Delhi, India; 1993.
ficiency of the different techniques evaluated and to the structural as- [22] Jaquin P. Analysis of historic rammed earth construction. PhD Dissertation. Durhan
University; 2008.
pect that can be improved with their application. This summary is
[23] Karanikoloudis G, Lourenço PB. Seismic assessment of Kuño Tambo Church
aimed at understanding which techniques are more effective in redu- (strengthening proposal), Peru. University of Minho, Technical report 2016-DEC/E-
cing the seismic vulnerability of the building and how, in terms of 10, Guimarães, Portugal; 2016.
[24] Langenbach R. Don’t tear it down! Preserving the earthquake resistant vernacular
which construction aspect of the building do they address. This can architecture of Kashmir. New Delhi, India: UNESCO; 2009.
eventually be helpful in making decisions on seismic strengthening [25] Lourenço PB, Krakowiak F, Fernandes FM, Ramos LF. Failure analysis of Monastery
strategies of vernacular constructions. The validation of these techni- of Jerónimos, Lisbon: how to learn from sophisticated numerical models. Eng Fail
Anal 2007;14:280–300.
ques is important because it can help to reintroduce them in the ver- [26] Lourenço PB. Recent advances in masonry structures: micromodelling and homo-
nacular building culture for the preservation of the vernacular heritage. genization. In: Galvanetto U, Ferri Aliabadi MH, editors. Multiscale modeling in
solid mechanics: computational approaches. Imperial College Press; 2009. p.
Gaining confidence on the use of these techniques can be also helpful in 280–300.
preventing the abandonment of vernacular buildings that are many [27] Lourenço PB, Mendes N, Ramos LF, Oliveira DV. On the analysis of masonry
times considered unsafe, since the loss of knowledge on traditional structures without box behavior. Int J Archit Herit: Conserv Anal Restorat
2011;5(4–5):369–82.
materials and construction techniques has commonly led to the de- [28] Lourenço PB, Karanikoloudis G, Mendes N, Corallo C. Assessment of the South aisle
molition and reconstruction of the buildings using new modern mate- in Canterbury. In: Proc. of the 2nd international conference on preservation,
rials, with the consequent invaluable loss of the heritage. maintenance and rehabilition of historical buildings and structures, Porto, Portugal;
2015.
[29] Lourenço PB, Karanikoloudis G, Greco F. In situ testing and modeling of cultural
Acknowledgments heritage buildings in Peru. In: Proc. of the 10th international conference on struc-
tural analysis of historical constructions, SAHC 2016, Leuven, Belgium; 2016.
[30] Mallardo V, Malvezzi R, Milani E, Milani G. Seismic vulnerability of historical
The work presented in this paper was partly financed by FEDER masonry buildings: a case study in Ferrara. Eng Struct 2008;30:2223–41.
funds through the Competitivity and Internationalization Operational [31] Mendes N, Lourenço PB. Seismic vulnerability of existing masonry buildings: non-
linear parametric analysis. In: Psycharis I, Pantazopoulou S, Papadrakakis M, edi-
Programme – COMPETE and by national funds through FCT – tors. Seismic assessment, behavior and retrofit of heritage buildings and monu-
Foundation for Science and Technology within the scope of the project ments. Comput Methods Appl Sci, vol. 37. Cham: Springer; 2015.
[32] Miccoli L, Oliveira DV, Silva R, Müller U, Schueremans L. Static behaviour of
POC1-01-0145-FEDER-007633. This work was also financed in the
rammed earth: experimental testing and finite element modelling. Mater Struct
framework of the Portuguese Public Procurement Code, LOTE 3EC5 – 2014;48:3443.
Escola Secundária da Anadia e Gafanha da Nazaré. [33] Michiels T. Seismic retrofitting techniques for historic adobe buildings. Int J Archit
Herit: Conserv Anal Restor 2015;9(8):1059–68.
[34] Mogollón J. Bahareque: a local seismic culture of the colombian coffee región. In:
References Proc. of international workshop on the role of bamboo in disaster avoidance,
INBAR, Guayaquil, Ecuador; 2002.
[35] Mouyiannou A, Rota M, Penna A, Magenes G. Identification of suitable limit states
[1] Angulo-Ibáñez Q, Mas-Tomás Á, Galvañ-Llopis V, Sántolaria-Montesinos JL. from nonlinear dynamic analyses of masonry structures. J Earthq Eng
Traditional braces of earth constructions. Constr Build Mater 2012;30:389–99. 2014;18(2):231–63.
[2] Audefroy JF. Haiti: post-earthquake lessons learned from traditional construction. [36] Naderzadeh A. Application of seismic base isolation technology in Iran. Menshin
Environ Urban 2011;23(2):447–62. 2009;63(2009.2):40–7.
[3] Betti M, Vignoli A. Numerical assessment of the static and seismic behaviour of the [37] NBC-203. Guidelines for earthquake resistant building construction: low strength
basilica of Santa Maria all'Impruneta (Italy). Constr Build Mater masonry. Government of Nepal, Kathmandu, Nepal; 1994.
2011;25(12):4308–24. [38] NBC-204. Guidelines for earthquake resistant building construction: earthen
[4] Braga A, Estevão J. Os Sismos e a Construção em Taipa no Algarve. In: Proc. of building. Government of Nepal, Kathmandu, Nepal; 1994.
Sísmica 2010 – 8°Congresso de Sismologia e Engenharia Sísmica, Aveiro, Portugal; [39] NTC. Norme Tecniche per le Costruzioni. D.M. 14 gennaio 2008, S.O. No. 30, G.U.
2010. No. 29/2008. Rome, Italy; 2008.
[5] Bui Q-B, Morel J-C, Hans S, Meunier N. Compression behaviour of non-industrial [40] Okubo T. Traditional wisdom for disaster mitigation in history of Japanese
materials in civil engineering by three scale experiments: the case of rammed earth. Architectures and historic cities. J Cult Herit 2016;20(July–August 2016):715–24.
Mater Struct 2008;42(8):1101–16. Cultural HELP 2014 Special Issue.
[6] Correia M, Merten J. Preliminary report of the local seismic culture in Portugal. In: [41] Ortega J, Vasconcelos G, Rodrigues H, Correia M, Lourenço PB. Traditional earth-
Taversism project – atlas of local seismic cultures, EUCCH, Ravello, Italy; 2001. quake resistant techniques for vernacular architecture and local seismic cultures: a
[7] Correia M. A Habitação Vernácula Rural no Alentejo, Portugal. In: Memorias del IV literature review. J Cult Herit 2017;27:181–96.
Seminario Iberoamericano sobre vivienda rural y calidad de vida en los asenta- [42] Ortega J, Vasconcelos G, Rodrigues H, Correia M. Assessment of the influence of
mientos rurales, Santiago del Chile, Chile; 2002. horizontal diaphragms on the seismic performance of vernacular buildings. Bull
[8] Correia M. Taipa no Alentejo. Lisboa, Portugal: Argumentum; 2007. Earthq Eng 2018.
[9] Correia M, Carlos G. Cultura sísmica local em Portugal/Local seismic culture in [43] Pierotti P, Uliveri D. Culture sismiche local. Pisa, Italy: Edizioni Plus, Università di
Portugal. Lisboa, Portugal: Argumentum; 2015. Pisa; 2001.
[10] D’Ayala DF, Paganoni S. Assessment and analysis of damage in L’Aquila historic city [44] Rota M, Penna A, Magenes G. A methodology for deriving analytical fragility curves
centre after 6th April 2009. Bull Earthq Eng 2011;9(1):81–104. for masonry buildings based on stochastic nonlinear analyses. Eng Struct
[11] Ferreira F, Vicente R, Varum H, Costa A, Costa AA. Out-of-plane seismic response of 2010;32(5):1312–23.
stone masonry walls: experimental and analytical study of real piers. In: Proc. of the [45] Saloustros S, Pelà L, Roca P, Portal J. Numerical analysis of structural damage in the
15th world conference on earthquake engineering, Lisbon, Portugal; 2012. church of the Poblet Monastery. Eng Fail Anal 2014;48:41–61.
[12] Ferrigni F. S. Lorenzello, à la recherché des anomalies qui protégent. Conseil de [46] Schacher T. Bhatar construction: an illustrated guide for craftsmen. Mansehra,
l’Europe et Centre Universitaire Européen pour les Biens Culturels, Ravello, Italy; NWFP, Pakistan: Swiss Agency for Development and Cooperation SDC; 2007.
1990. [47] TNO. DIsplacement method ANAlyser. User's manual, release 9.4.4. Netherlands;
[13] Foufa AA, Benouar D. Atlas of earthquake-resistant traditional techniques in 2011.
Algeria: the case of the casbah of Algiers. Eur Earthq Eng J 2005;2(5):23–39. [48] Tolles EL, Kimbro E, Webster F, Ginell W. Seismic stabilization of historic adobe
[14] Gallego R, Arto I. Evaluation of seismic behaviour of rammed earth structures. In: structures: final report of the Getty Seismic Adobe Project. GCI Scientific Program
Mileto, Vegas, García-Soriano, Cristini, editors. Earthen architecture: past, present Reports. Los Angeles, USA: The Getty Conservation Institute; 2000.
and future. London, UK: Taylor & Francis Group; 2014. [49] Touliatos PG. Traditional aseismic techniques in Greece. In: Mendes Victor L, edi-
[15] Gautam D, Prajapati K, Paterno KV, Bhetwal KK, Neupane P. Disaster resilient tors. Proc. of the international workshop ‘Les systemes nationaux faces aux seismes
vernacular housing technology in Nepal. Geoenviron Dis 2016;3(1). majeurs’, Lisbon, Portugal; 1992.
[16] Giongo I, Dizhur D, Tomasi R, Ingham J. Field testing of flexible timber diaphragms [50] Vintzileou E. Effect of timber ties on the behavior of historic masonry. J Struct Eng
in an existing vintage URM building. J Struct Eng 2014;141 Special issue: Field 2008;134(6):961–72.
testing of bridges and building. [51] Wilson A, Quenneville P, Ingham J. Natural period and seismic idealization of
[17] Gomes MI, Lopes M, de Brito J. Seismic resistance of earth construction in Portugal. flexible timber diaphragms. Earthq Spectra 2013;29(3):1003–19.
Eng Struct 2011;33(3):932–41.

27