You are on page 1of 22

International Journal of Architectural Heritage

Conservation, Analysis, and Restoration

ISSN: 1558-3058 (Print) 1558-3066 (Online) Journal homepage: https://www.tandfonline.com/loi/uarc20

Seismic Load Capacity of Historical Masonry


Mosques by Rigid Body Kinetics

Irfan Kocaman, Ilker Kazaz & Emriye Kazaz

To cite this article: Irfan Kocaman, Ilker Kazaz & Emriye Kazaz (2020) Seismic Load Capacity of
Historical Masonry Mosques by Rigid Body Kinetics, International Journal of Architectural Heritage,
14:6, 849-869, DOI: 10.1080/15583058.2019.1570389

To link to this article: https://doi.org/10.1080/15583058.2019.1570389

Published online: 03 Feb 2019.

Submit your article to this journal

Article views: 306

View related articles

View Crossmark data

Citing articles: 2 View citing articles

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


https://www.tandfonline.com/action/journalInformation?journalCode=uarc20
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE
2020, VOL. 14, NO. 6, 849–869
https://doi.org/10.1080/15583058.2019.1570389

Seismic Load Capacity of Historical Masonry Mosques by Rigid Body Kinetics


a
Irfan Kocaman , Ilker Kazaza, and Emriye Kazaz b

a
Department of Civil Engineering, Erzurum Technical University, Erzurum, Turkey; bFaculty of Architecture and Design, Atatürk University,
Erzurum, Turkey

ABSTRACT ARTICLE HISTORY


Kinetic analysis methods based on linear and nonlinear rigid body dynamics are used to evaluate Received 10 October 2018
earthquake safety of masonry structures. In this study, the formulas used to calculate the in-plane Accepted 11 January 2019
and out-of-plane load capacities of masonry load-bearing walls were evaluated and a procedure KEYWORDS
based on rigid body mechanism was proposed to calculate the out-of-plane load capacities of the Kinetic analysis; Lala Pasha
walls of Ottoman period masonry mosques. New aspects of the method with respect to existing Mosque; load capacity;
formulations is the inclusion of dynamic axial load and definition of the collapse limit spectral masonry mosque;
acceleration on the overturning wall. The calculated capacities of the mosque and individual walls mechanism; nonlinear
were compared with the results of nonlinear pushover analysis and time history analyses per- analysis; pushover analysis;
formed under 1.0 and 0.5 scaled forms of nine different 3-component ground motion records. It time-history analysis
was displayed that the seismic load capacity estimated by the proposed method is very close to
the values calculated by pushover and time history analyses. The method was developed on Lala
Pasha Mosque, and the reliability and applicability of the proposed methodology is verified on
a different historical masonry mosque in comparison to finite element analyses results.

1. Introduction structural systems. This structural form, which was


born in the vicinity of the Mediterranean, has been
Being home to many civilizations over the course of
used in the early Christian structures and the Sassanid,
history, Anatolia is also a geography where the seis-
Armenian, Byzantine, and Roman architectures (Tuluk
mic activity is very high. For this reason, determina-
2006). The Ottomans developed their own architectural
tion of the earthquake load capacity of historical
style characterized by large central rotundas (sometimes
structures correctly, taking strengthening measures
surrounded by multiple smaller domes), pencil-shaped
in case of inadequacies and preserving these struc-
minarets, and open facades.
tures for the future generations safely is a crucial task
In this system, the main dome is seated on the
for Anatolians. It is also important to accurately
structure that is constituted with pillars anchored each
determine the damage that occurred or may occur
other with supporting arches. The spatial composition
on the structures, in order to take preventive mea-
is developed around this domed baldachin structural
sures and to keep the structural health under control.
system in the mosques. This type of domed structures is
Historic mosques constitute a significant portion of
classified as “longitudinally developed,” “transversely
the cultural heritage in Anatolia. The mosque is the
developed,” and “centrally located” according to the
Muslim gathering place for prayer. In the Islamic
direction of the expansion of space. Despite the choice
world, the mosque in its many forms is the quintessen-
of the longitudinal space usage in Christian architec-
tial Islamic building. The architecture of a mosque is
ture, the space development in Ottoman mosques is
shaped most strongly by the regional traditions of the
central. Although such central structures have been
time and place where it was built. As a result, style,
found in Roman architecture in Byzantium, Anatolia,
layout, and decoration can vary greatly.
and Syria long before the Ottomans, there is no other
The dome and domed structure, one of the indispen-
place where the central planned type of space advanced
sable symbolic and functional formations of the history
to a level as in the Ottoman mosques (Kuban 1958).
of world architecture, formed the cover systems of the
The Ottomans introduced central dome mosques in
big space structures during the centuries between the
the 15th century. Since the 14th century, the Ottomans
Middle Age and the emergence of the contemporary
started the development process of the type of domed

CONTACT Ilker Kazaz ilkerkazaz@erzurum.edu.tr Department of Civil Engineering, Erzurum Technical University, Erzurum 25050, Turkey
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/uarc.
© 2019 Taylor & Francis
850 I. KOCAMAN ET AL.

space in Anatolia by combining the structure culture in depending on the geometry, construction technique,
Islam and Seljuks with the local materials of Anatolia and and accuracy expected from the model (de Felice
their own culture. Until the middle of the 15th century, 2011). Linear response spectrum analysis, linear and
this form of space is encountered in the Early Period nonlinear single and multimode static (pushover) ana-
structures in Iznik and Bursa, where the dome is carried lysis, time-history analysis are frequently used in the
by the walls. From the middle of 15th century, the need to structural evaluation of masonry structures. Being the
enlarge the continuous space centered by dome has arisen main load bearing elements of historical buildings,
the idea of covering the main space with a dome as large there are various studies to determine the lateral load
as possible and stipulated the desire of growing and capacities of masonry walls. Many of these studies
spreading in every direction to provide the transverse contain experimental work and theoretical approaches
development of space. Due to the fact that the boundaries for the estimation of load and displacement capacities
of the central space to be covered by a single dome over (Candeias et al. 2017; Costa 2012; Karimi et al. 2016;
a certain dimension is technically very difficult for that Silva et al. 2014; Vasconcelos and Lourenço 2006).
period, this problem has been solved by moving the body Kanıt and Döndüren (2010) experimentally and theo-
walls bounding the space outwards by almost always retically examined the behavior of the masonry wall
determining a rectangular form and supporting the cen- sample under cyclic loading. The experimental results
tral dome on a four-, six-, or eight-footed baldachin were compared with the numerical results obtained by
creating a central space. These mosques have a large using ANSYS software. It was determined that ANSYS
dome centered over the prayer hall. In addition to having software is a reliable tool for numerical studies on
a large central dome, a common feature is smaller domes masonry walls.
that exist off-center over the prayer hall or throughout the Kinetic analysis methods based on linear and nonlinear
rest of the mosque, where prayer is not performed. This rigid body dynamics are also used to evaluate earthquake
system, which can be composed of four, six, or eight safety of masonry structures (Ferreira, Costa, and Costa
pillars, led to come into existence of original and rich 2015; Sorrentino et al. 2017). By assuming a certain col-
mosque tradition of classical period Ottoman architecture lapse mechanism in the calculation, expected vertical and
(Tuluk 2006). horizontal loads are placed in the system and the horizon-
Tuluk (2006) grouped the four-pillar baldachin form tal force compatible with the mechanism is calculated.
Ottoman mosques under six headings as described in Some studies investigated the applicability of displacement
Figure 1. The richest spatial trials of these baldachin based methods, taking into account the reserve strength
structures were on the type sitting on four pillars. Half- due to rocking and bending to determine the behavior of
dome, dome, or vault-covered spaces, façades and gal- masonry walls under dynamic loading (Doherty et al. 2002;
leries, which are sometimes placed on only two sides, Melis 2002). In some approaches, a part of the structure is
sometimes on four sides, both structurally supported modeled as pin-connected rigid blocks (no-tension mate-
this carrier system and contributed to the spatial expan- rial with infinite compressive strength) by neglecting frac-
sion of the plan determined by this central dome. ture energy and friction between the blocks (Ordinance
This study aims to investigate the seismic behavior 3274, 2003; Ordinance 3431, 2005). The local collapse
of masonry mosques and introduce simplified assess- mechanism was also studied through a kinetic limit analy-
ment tools for the load capacity estimation. The diffi- sis based on rigid block rotation (Garofano and Lestuzzi
culties and variability of methods in the determination 2016). Such kinetic approaches were verified with different
of load capacity of historic structures are addressed in methods such as experimental study, pushover analysis,
different studies. The modeling approach and analysis time-history analysis, static analysis, etc. in different works
methods are important for determining behavior of (Cennamo et al. 2011; Galasco, Lamomardino, and Penna
structure. The selection of this approach may vary 2006; Milani and Valente 2015; Pavlovic, Reccia, and

Type I Type II Type III Type IV Type V Type VI

Figure 1. Four-pillar baldachin variations in Ottoman mosques and spatial expansion (Tuluk 2006). The thick black square in the
center displays the central space covered by main dome.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 851

Cecchi 2016). The dependence of the out-of-plane seismic model was verified against available experimental data
capacity of masonry walls on the morphology of the wall at the first stage. The necessary components of the new
section is investigated using the distinct element method method were developed and accuracy was evaluated in
and it was reported that the effective reduction in strength consideration with historical Lala Pasha Mosque, which
and displacement capacity of masonry sections depend on is a typical classical period Ottoman mosque con-
the transverse bond provided by stones (de Felice 2011). structed under the supervision of the great architect-
Derakhshan et al. (2017) proposed a displacement-based engineer Sinan. Seismic capacity and performance of
assessment procedure, which uses effective SDOF system the Lala Pasha Mosque were calculated using nonlinear
representation of the actual structure based on the pushover and time-history analyses under nine differ-
observed damage mechanism, to predict the results of ent 3-component ground motion records scaled to 1.0
shake table testing of two unreinforced masonry buildings. and 0.5 times of the original amplitude. The horizontal
These studies show that different kinematics approaches in- and out-of-plane load capacities of each wall and
can be used safely to evaluate the seismic capacity of entire structure were obtained from the dynamic ana-
masonry walls. lyses and the pushover analyses were compared with
Besides kinetic approaches, there are various studies for the proposed method estimations and among each
the determination of the seismic capacity of masonry struc- other. The usability of each and in particular proposed
tures with pushover analysis. Özkaya, Kazaz, and Okuyucu method was evaluated.
(2015) studied the seismic behavior of single span masonry The reliability of the method was further verified on
bridges in their work. They have compared the displace- another historical structure, Erzurum Narmanlı Mosque.
ments and local strains obtained from nonlinear dynamic The seismic capacity of Narmanlı Mosque was determined
and static analyses on the finite element model. Bocciarelli with time-history analyses under three different severe
and Barbieri (2017) performed nonlinear pushover analysis ground motions and pushover analysis. Comparison of
in the seismic investigation of the historical Gabbia Tower. the capacities estimated with proposed method and finite
The structure was modeled according to the Euler- element analyses for both structures displayed that the
Bernoulli beam theory and a numerical approach for the proposed method safely estimates the seismic load capa-
pushover analysis of masonry towers, having hollow arbi- cities of classical period Ottoman masonry mosques.
trary sections, was proposed. D’Ambrisi, Mariani, and
Mezzi (2012) compared the results of nonlinear static and
dynamic analysis of a historical tower and underlined the 2. Erzurum Lala Pasha Mosque
importance of the nonlinear dynamic analysis with refined
2.1. History and architectural features
models in the evaluation of the actual seismic capacity of
a masonry structure. They reported that the base shear-top Lala Pasha Mosque is located in the city center of
displacement curves from dynamic analyses compose an Erzurum between the Yakutiye Madrasah on the west
envelope for the static pushover curve. Bartoli, Betti, and and two important traffic routes of the city on the east
Vignoli (2016) was evaluated the seismic vulnerability of and south, as shown in Figure 2. The mosque is the first
a tower structure on three levels of evaluation, depending Ottoman period structure in Erzurum and it is now
on different degrees of knowledge of the structure, as pro- under the responsibility of the General Directorate of
posed by the Italian “Guidelines for the assessment and Foundations. Architect Sinan is the architect and engi-
mitigation of the seismic risk of the Cultural Heritage.” neer of the mosque. Lala Pasha Mosque enters the group
The seismic demand and damage was evaluated using the of square planned and central domed mosques as
capacity spectrum method. İstanbul Şehzade Mosque, Gebze Çoban Mustafa Pasha
In this study, available equations calculating the in- Mosque, Payas Sokullu Sarı Selim Mosque, which are
and out-of-plane load capacities of masonry walls were works of Architect Sinan. The main dome, which is
examined to determine the total lateral load capacity of placed on four octagonal stone columns in center of
mosque type structures composed of these elements. the mosque, is supported with half dome and buttresses
A new method that is based on the kinetics of out-of- on four sides. The mosque was built symmetrically with
plane overturning collapse mechanism and incorpo- small domes on the sides of the main dome. The main
rates dynamic axial load on the rocking wall as novelty dome is 13 m in diameter, 17.45 m in height, and
was proposed to determine the out-of-plane load capa- approximately 50 cm thick. Main dimension of the
city of masonry walls. The method’s estimations were mosque is 26 × 26 m2. The narthex on the south side
compared with time-history analyses results employing is 26 × 5 m2. The narthex consists of 6 columns support-
nonlinear finite element analysis method, so the accu- ing 5 small domes in-between. The thickness td of the
racy of utilized finite element procedure and material four main walls (load bearing walls) of the structure is
852 I. KOCAMAN ET AL.

(a) (b)

(c) (d)

Dimensions in centimeters.

Figure 2. (a) Erzurum Lala Pasha Mosque, (b) current plan, (c) section AA, and (d) section BB.

approximately 180 cm. The walls increased to the roof Table 1. Material properties of Erzurum Lala Pasha Mosque.
level is at a height of hd = 8.4 m. The ld/hd and hd/td Compressive Tensile Elasticity
Strength, fck Strength, fctk Modulus,
ratios on the walls of the mosque are 3.13 and 4.67, Mosque Part (MPa) (MPa) E (MPa)
respectively. Here, ld is the length of walls. There are Walls 3.92 0.3 784
two rows of windows on the four faces of the mosque. Columns and Arches 5.63 0.3 1126
Domes 3.00 0.3 600
Minaret is located on the north-west corner of the
mosque.
The length/height (ld/hd) ratio of the load bearing walls used by taking into consideration the previous investiga-
varies between 1 and 5 (usually 2 and 3) in historical tions made on similar structures. Since destructive test
mosques in Turkey. The wall thickness (td) varies accord- methods are not allowed on the structure, the necessary
ing to the span and the vertical loads transferred from the compressive strength of the stone was extracted by
roof cover. In the buildings covered with small domes Schmidt hammer tests. Details of the method used to
(madrasa, caravanserais cells), the walls are 65–85 cm determine material properties can be found in Kocaman
thick and wall thickness increases as the diameter of the (2017). In the Turkish Seismic Code (DBYBHY 2007),
dome increases. As a different example, Izmit Pertev compressive strength and elasticity modulus relation is
Pasha Mosque with a dome diameter of 16.75 m has suggested as E = 200fck for the masonry elements, where
a wall thickness of 203 cm. Approaches in this study are fck is the homogenized compressive strength of the three-
usable in case of stone masonry mosques with ld/hd ratios leaf walls. On the other hand, the elastic modulus defined
of 3–4 and hd/td < 4–5. for the wall material is given as 550fck in FEMA356
(FEMA356 2000) and 1000fck in EC6 (Eurocode 6 1995).
If E = 200fck was used, it was observed that horizontal load
2.2. Material properties
capacity does not change much, but the nonlinear displa-
For the mechanical properties of the materials used in the cement demand is about four to five times higher than
finite element model, the values given in Table 1 were 1000fck case (Kazaz and Kocaman 2018). In this study, the
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 853

elastic modulus under compression is assumed to be capable of cracking in tension and crushing in com-
E = 200fck, which may be considered as a representative pression. A five-parameter Willam-Warnke concrete
lower value of elastic modulus for such type of structures model (CONC) was used as the material model, for
according to available literature (Magenes and Penna which the element is also applicable to geological mate-
2009). rials such as rock (Willam and Warnke 1975). In this
model, the failure stress surface of the material is
formed by using the tensile and compressive strength
2.3. Finite element model of the mosque
of the stone. The behavior of the material is linear up to
Load resistance systems of historic masonry structures are failure in cracking and crushing. If the material at an
quite different from the load bearing systems of conven- integration point fails in uniaxial, biaxial, or triaxial
tional structures. Since different structural forms such as compression, the material is assumed to crush at that
vaults, arches, pendants, and domes are used in mosques, point. In SOLID65, crushing is defined as the complete
it is difficult to create the numerical models and deter- deterioration of the structural integrity of the material
mine the behavior of masonry mosques using traditional (e.g., material spalling). Under conditions where crush-
elements. The complexity arising from decorative and ing has occurred, material strength is assumed to have
ornamental features may lead to problems in the model- degraded to an extent such that the contribution to the
ing and meshing of these structures. In this study, archi- stiffness of an element at the integration point can be
tectural decorations which are thought to have no effect ignored. Numerous numerical applications for the use
on structural behavior and consists of insignificant of SOLID65 and Willam-Warnke material model in the
amount of mass were ignored in the finite element mod- finite element analyses of reinforced concrete and
els. Nevertheless, as shown in Figure 3a, a detailed solid masonry structures in the ANSYS are given in detail
model of Lala Pasha Mosque was produced using in Kazaz, Yakut, and Gülkan (2006) and Kazaz (2010).
SolidWorks (2013) software. The generated model was
transferred to ANSYS APDL (2015) software and the
2.4. Verification of finite element tool
model was meshed with 17,860 nodes and 66,521 solid
elements, as shown in Figure 3b. The brick only mesh Now, it is demonstrated that the SOLID65 element
cannot be achieved due to treatment of mosque model as and the Willam-Warnke material model can be used
single solid part in order to prevent contact elements and safely in dry-joint masonry walls by validating avail-
to enable continuous mesh among different parts of the able test results. In previous studies, William-Warnke
model such as walls, arches, and domes. So, tetrahedron material model and SOLID65 element in ANSYS
elements were used together with bricks elements in the software was commonly used for comparative finite
model. The boundary conditions at the base of the walls element analysis on the seismic behavior and assess-
were taken as fixed. Further details about the finite ele- ment of historic masonry buildings with experimental
ment model can be found in Kocaman (2017). data (Betti, Galano, and Vignoli 2014, 2016; Bartoli
The legacy SOLID65 element was used in ANSYS et al. 2017). It was reported that the finite element
models in this study. SOLID65 has 8 nodes and 3 model is able to predict the damaged areas and the
translational degree of freedoms per node. SOLID65 is incipient collapse mechanism with respect to the
used for the three-dimensional model of solids with or damage experienced by the masonry specimen during
without reinforcing bars (rebars). The solid element is the shaking tests. These studies also mentioned about

(a) (b)

Figure 3. Lala Pasha Mosque (a) SolidWorks model and (b) ANSYS model.
854 I. KOCAMAN ET AL.

the effects of the epistemic uncertainties that are number of parameters used to define the failure surface
related to the different models and numerical techni- in the material model are the crushing strength, tensile
ques in structural modeling. strength, shear transfer coefficients for open and closed
Oliviera (2003) conducted experimental study in cracks, which were assigned the values of 7 MPa,
order to determine behavior of dry joint masonry walls 0.05 MPa, 0.05 and 0.8, respectively. For the elastic
under compression and lateral force. In the experiment, modulus of walls, experimental values derived under
the effect of increased axial load on the lateral displace- pre-load application on the walls as reported in Oliviera
ment capacity of the walls was investigated. Figure 4a (2003) was assigned to models.
gives the details of the experimental setup. The wall Figure 5 plots the comparison of lateral load-
dimensions are 1000 × 1000 × 200 mm3. We are con- displacement curves from the experiment, the numer-
cerned with the results of tests under 30 kN, 100 kN, ical study by Oliviera (2003) and the finite element
200 kN, and 250 kN axial load values, for which the model in this study. Estimated force capacities are
applied axial load produces axial stress levels of 0.15, seen to be quite consistent with experimental values,
0.50, 1.00, and 1.25 MPa, respectively. Except 250 kN which are the primary concern of this study. The dis-
load case, two specimens were tested for each case. These placements up to initiation of strength degradation,
stress values are typical in masonry structures. It was which can be attributed to material crushing, were
noted that all the stones used in specimen had a smooth also predicted well. Since in the utilized material
surface. It was also reported that elastic behavior of the model, the total strength diminishes beyond failure
walls was observed up to 30% of the maximum load. surface, the numerical model suddenly loses strength
Finite element model of the walls were created by due to crushing. Using plasticity formulations can over-
using ANSYS APDL software as displayed in Figure 4b. come this problem, yet as the calculations by Oliviera
In the model, the axial load was applied using mass employing such a technique has demonstrated, the
elements creating the same level of compression on the actual behavior cannot be captured correctly. In fact,
wall under gravitational acceleration as in the experi- the behavior in this region after peak load is question-
ment. The SOLID65 solid element in the ANSYS able on the experimental curve as well, since under
element library was used to model the wall. William- continuously imposed monotonic displacement loading
Warnke concrete model was used to model stone fail- fluctuations on the load curve indicates to complex
ure surface under three-axial stress state. The least disintegration phenomenon of masonry.

(a) (b)
Mass Elements

Concrete
Beam

Figure 4. (a) Dry-joint masonry wall dimensions and test setup and (b) ANSYS model.

Figure 5. Comparison of experimental and calculated lateral load-displacement curves (SW30 indicate to specimen under 30 kN
axial load).
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 855

2.5. Modal analysis where Γn is the modal participation factor, Øn is the


modal displacement vector of the nth mode, [M] is the
Mode shapes and fundamental vibration frequencies of
mass matrix, and Sa is the maximum acceleration of
Lala Pasha Mosque were obtained as a result of modal
a SDOF oscillator with period Tn and damping ratio ξn.
analysis. For the material properties, values given in
Application of this force will create the same maximum
Table 1 were defined. Total mass of the mosque was
response of the nth mode. Since the spectral accelera-
calculated as 5679 tons. The modal analysis was done in
tion and the modal participation factor only has the
20 modes and the first 2 modes came out to be the
effect of scaling the load vector in Equation (1), in
minaret’s modes. Since these two modes are local
pushover analysis the load-displacement curve of the
modes that do not represent the entire structure, only
system can be obtained by a constant and monotoni-
the modal quantities of the main mosque body were
cally increasing lateral load pattern defined as
reported. Table 2 gives the modal frequencies and mass
participation ratios for the first four main body modes. 
fPn g ¼ ½M: ϕn (2)
While modes 3 and 4 are mixed translational-torsional
modes, modes 5 and 6 are torsional modes. on the premise that the dynamic characteristics of the
Kocaman (2017), using the operational modal ana- nonlinear (damaged) structure remains unchanged as
lysis method, determined the experimental dynamic the linear (undamaged) system. In case of uniform
characteristics of Lala Mustafa Pasha Mosque. mass distribution along the height, such as the conven-
Experimental measurements were carried out using tional buildings, the lateral load pattern for pushover
pedestrian and traffic induced environmental vibrations analysis can be represented only with modal displace-
on the mosque. Close values of mode shapes and fre- ment vector, Øn. In irregular structures, this force dis-
quencies were obtained between experimental and tribution can be defined as the combination of more
numerical models taking the elastic modulus of than one modal inertia force vector as in case of multi-
masonry material as 1000fck in the finite element mode pushover analysis.
model as reported in Table 2. In this study, E = 200fck Application of pushover analysis on historical
was adopted in nonlinear static and dynamic analyses masonry structures possesses difficulties, since the
considering that this value better characterizes the large tensile strength of material is very low, which changes
amplitude vibration behavior of the homogenized wall the load transfer path and force redistribution imme-
material in account of the reduction in stiffness under diately after cracking, the distribution of mass in ele-
earthquake effect. vation and the configuration of structural system in
plan and elevation is irregular. In general, three lateral
loads distributions are used for pushover analysis: (a)
2.6. Pushover analysis modal pattern, proportional to forces consistent with
Static pushover analysis is a simple yet effective method the nth mode shape in the loading direction; (b) mass
for the determination of seismic performance of struc- proportional pattern, based on lateral forces propor-
tures. The nonlinear force-displacement relationship of tional to mass regardless of its position in elevation;
structures in earthquake effect can be determined using and (c) mass and mode proportional pattern, where
a lateral load pattern mimicking the modal force dis- a force distribution proportional to the nth mode dis-
tribution of the mode with the largest mass participa- placement shape multiplied by the mass distribution,
tion ratio under the assumption that the dynamic as defined in Equation (2). Contrary to expectations,
behavior is dominated by this mode. From structural in a previous study Endo, Pela and Roca (2017)
dynamics, the equivalent static load that creates the reported that the mass proportional loading pattern
same maximum effect of the nth mode of a linear sys- yielded the closest load capacity to nonlinear dynamic
tem is given by analyses than the other loading patterns. Interestingly,
 in that study more convenient loading patterns sig-
fFn g ¼ Γn :½M: ϕn :Sa ðTn ; n Þ; (1) nificantly underestimated the load capacity of the

Table 2. Mass participation ratios (MPR) and modal frequency values for the main body modes of Lala Pasha Mosque.
Freq. (Hz) Freq. (Hz) Freq. (Hz) MPR MPR MPR MPR
Mode (Exp.) E = 1000fck E = 200fck (x dir. Horz.) (y dir. Vert.) (z dir. Horz.) (y Rotation)
3 9.42 8.44 3.77 0.06 0.48E-5 0.56 0.56
4 10.55 8.5 3.80 0.55 0.79E-4 0.06 0.64
5 10.85 9.3 4.16 0.57E-3 0.25E-4 0.36E-4 0.64
6 11.82 11.74 5.25 0.25E-4 0.30E-3 0.84E-3 0.66
856 I. KOCAMAN ET AL.

structure. For conventional buildings, mass propor- was calculated as 1.12. We simply take it unity in our
tional loading pattern corresponds to uniform rectan- analysis. As input to the program, the simplified accel-
gular lateral loading, which envelops the upper bound eration profile, which was defined as the average
estimation of load capacity. acceleration on the wall and dome parts, was applied
Equation (2) is especially vital for historical masonry on the walls and the dome as separate values, as shown
structures, such as mosques, where the mass is not uni- in Figure 6a. In this way, the acceleration applied to
form in horizontal and vertical directions. It is difficult the dome part of the structure is about 2.2 times
to create a lateral load profile as described by Equation greater than the acceleration applied to the lower
(2) in analysis packages, since mass in masonry struc- load bearing members (walls and columns).
tures is not lumped at certain levels as in floors of It is known that the load profile used in pushover
ordinary buildings. Therefore, instead of calculating analysis changes the calculated capacity curves. The lat-
and applying force profile acting on nodes of the struc- eral force-displacement curves calculated by uniformly
ture in ANSYS, pushover analysis is performed by using distributed acceleration profile (mass proportional load-
an acceleration field defined as ϕ € ¼ ϕ Sa . Effective ing) and by idealized acceleration profile (simplified
n n
earthquake forces are determined as the mass of each mass and modal shape proportional loading) are given
element multiplied by this acceleration field. in Figure 7a. The idealized load pattern resulted in less
For Lala Pasha Mosque variation of acceleration stiffer and lower capacity response curve than the uni-
along the height of the structure was determined by form case as expected. Capacity values calculated in the
Response Spectrum Analysis (RSA). Turkish Seismic North-South (N-S) direction and capacity values in the
Code (DBYBHY 2007) design spectrum for firm soil East-West (E-W) direction are almost same because
and effective peak ground acceleration of 0.4 g was the structure is symmetrical except for the narthex
used as seismic input. Figure 6 displays the amplified and the minaret. According to the pushover analysis,
acceleration profile along the height of the structure the lateral load capacity of the structure is around
on the central axes obtained from spectrum analysis. 25,000 kN. The structure behaves linear up to 10 mm
However, due to its structural form and properties, the of dome top displacement and after 25 mm the structure
amplification is not the same at all parts of the entire reaches maximum load capacity. Because of force-
structure and in particular on rocking walls as shown controlled continuous incremental loading used in push-
in Figure 6b. As we move to the edges of the walls over analysis, load reduction couldn’t be achieved after
restrained by perpendicular walls, the acceleration maximum load capacity was reached. Therefore, the
profile changes significantly. In historical masonry analysis was interrupted at the point where it reaches
buildings such as mosques, baths, and caravansaries maximum load capacity. So, the displacement capacity of
the modal participation factor Γn, which is a measure the structure may probably attain slightly larger values. It
of this amplification in displacements and accelera- should be noted that this is an inadequacy of the existing
tions, is recommend to be taken as 1.1 (TYDK 2017). force-controlled analysis method.
In the mosque model, taking the north wall top mid- Selection of the control node for the displacements
point as control node, the modal participation factor representing the horizontal axis of the capacity curve

(a) 18 (b)
Simplified acceleration
16 profile
14
12 Acceleration profile obtained
Height (m)

10 by spectral analysis
8
6
4 Wall top
2 level
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Acceleration (g)

Figure 6. (a) Distribution of the accelerations along the height on the middle axisand (b) modal combination of acceleration
response of the Lala Pasha Mosque from RSA.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 857

significantly effects the equivalent single degree of repre- Although the elastic modulus of the material is
sentation of the structure. This conversion, which is the reduced from 1000fck to 200fck, it is seen that the
essence of simplified nonlinear analysis methods, is based calculated drift ratio limit values came out to be con-
on the comparison of both the structural capacity siderably lower than the performance limits recom-
(pushover) curve and the demand spectra in spectral- mended in the TYDK guideline according to the
acceleration vs. spectral-displacement coordinates. How- results of the pushover analyses. In case of linear ana-
ever, this study is aimed to investigate the load and lysis, calculated displacements should be even lesser.
deformation capacity of the mosque using direct meth- For this reason, a performance assessment with pro-
ods, which requires no such capacity conversion of the posed guideline limits will yield unsafe estimations.
system. So, roof top point and wall-top midpoints is
selected to represent the displacement response of the
2.7. Nonlinear time-history analyses
entire structure and the walls, respectively.
In Figure 7b, the base shear force is plotted against Due to close proximity to active faults, Eastern Anatolia
the displacements obtained at the top mid-points of Region was exposed to many large and small earth-
the four walls and the main dome top. Comparing quakes throughout history. According to historical
these pushover curves, it is seen that the displacement sources, there were earthquakes caused loss of life and
of north wall top point in the push direction is sig- property in the years of 1790, 1843, 1850, 1852, 1859,
nificantly larger than other points as a result of out- 1868 (Erzurum Environmental Status Report 2011).
ward overturning action. It is also notable in Figure 7b According to the information received from the
that drift ratios at the maximum load for walls acting Regional Directorate of Foundations in Erzurum, the
out-of-plane is significantly larger than the walls act- mosque did not suffer any serious damage during these
ing in-plane. The drift ratios for north and south walls earthquakes. In this study, the time history analyses of
are 0.55% and 0.17%, respectively. For east and west the Lala Pasha Mosque were carried out using nine
walls acting in-plane parallel to earthquake direction different ground motions and their half-scaled forms,
the drift ratio is 0.05% approximately. Static drift ratio which were previously compiled by Kazaz (2016). The
damage limits were proposed in the guideline on ground motions were not selected to conform the
the Management of Earthquake Risks for Historic regional seismic hazard, but rather to excite the struc-
Buildings (TYDK 2017) by the General Directorate ture on every level of performance scale. In the analyses
of Foundations. Similar values were also proposed in seismic excitation was applied in all three directions
other studies (Vanin et al. 2017). According to this (two horizontal and vertical) except the Erzincan
guideline, the drift ratio limits for limited damage and record. Only horizontal components of Erzincan
collapse prevention is recommended as 0.3% and 1%, ground motion was used in the analysis. The character-
respectively. However, the calculated drift ratio limits istics of the earthquake records are summarized in
are significantly lower than the guideline recommen- Table 3. In the table, horizontal and vertical compo-
dations. The global drift ratio capacity at the top of the nents of the ground motion were named with the
dome is 0.15% in the analysis. corresponding model axes in the analysis platform.

Drift Ratio (‰) Drift Ratio (‰)


0 0.6 1.2 1.8 2.4 0 1 2 3 4 5
30000 30000
E&W S N
25000 25000
Displacement
20000
Force (kN)

20000
Force (kN)

control point Displacement


control points
15000 15000
W
10000 10000
S N
5000 Uniform Acel. 5000
Idealized Acel. E
0 0
0 10 20 30 40 0 10 20 30 40 50
(a) Displacement (mm)
(b) Displacement (mm)

Figure 7. (a) Effect of different load patterns on force-displacement relation and (b) the force-displacement curves for different
control points on mosque under idealized acceleration.
858 I. KOCAMAN ET AL.

The damping was assigned in the form of Rayleigh Figure 9 shows the principal total strain on the
damping yielding a damping ratio of 3% in the signifi- structure as a result of pushover and time-history ana-
cant modes. lyses and a current photograph from the east wall. As
The maximum values of dome top displacement seen in Figure 9a,b, as a result of outwards overturning
and base shear force pairs obtained from time history tendency of north wall, there is significant possibility of
analyses are plotted in Figure 8. The figure contains crack development extending between the nearest win-
data points, plotted all as positive, of maximal values dow opening and the wall top on the perpendicular
on both positive and negative sides of the time- walls (east wall window). This damage was predicted
response curves. The static pushover curve and the by both pushover analysis and time-history analysis. As
dynamic base shear-top displacement curve from displayed in Figure 9c, the existence of different pattern
Erzincan earthquake analysis were also drawn on the and color of stones in this area indicates to a possible
graph, because the response curve derived under this earthquake damage and a repair action afterward,
record reveals the structure’s force and displacement throughout the lifetime of the mosque.
capacity explicitly. Erzincan earthquake simulation In Figure 8, it is seen that the pushover analysis that
disclosed that the mosque experiences partial collapse uses the force profile compatible with the modal dis-
of dome as a result of out-of-plane overturning and placement vector multiplied with the mass as given by
toppling of the north wall as shown in Figure 9b. This Equation (2) and the time-history analysis yields close
earthquake record, which effectively reflects the char- estimations of load capacity and stiffness. Figure 8 also
acteristics of near-field ground motion, is an appro- displays that displacement limit for elastic behavior is
priate seismic input for time-history analysis of around 10 mm, whereas strength starts decreasing after
masonry structures for obtaining the collapse mechan- 20 mm verifying the limits determined on the pushover
ism and load-displacement relationship. Among the curve. According to the hysteresis curve obtained from
selected ground motion inputs, collapse was also the Erzincan earthquake, it can be interpreted that the
observed in Cape Mendocino and Kocaeli earthquake strength degradation roughly beyond 40 mm of dome
analyses. top displacement is unrecoverable. This value can be

Table 3. Peak ground values of ground motions used in time-history analyses (Y direction corresponds to vertical in the analyses platform).
PGA (cm/sec2) PGV (cm/sec)
Earthquake Year Mw Station X dir. Z dir. Y dir. X dir. Z dir. Y dir.
Imp. Valley 1979 6.5 Keystone Rd., El Cento Array #2 365 309 104 32.8 33.7 12.0
Kocaeli 1999 7.4 Düzce 374 315 480 52.6 59.5 20.9
Erzincan 1992 6.8 Center 382 460 - 100.2 189.4 -
Northridge 1994 6.7 Pacoima-Kagel Canyon 295 424 177 31.1 50.8 14.1
W. Narrows 1987 6.1 7420 Jaboneria, Bell Gardens 216 90 220 27.9 3.9 17.3
Cape Men. 1992 7.1 89324 Rio Dell Overpass-FF 378 539 191 45.6 43.7 9.6
Northridge 1994 6.7 LA — Fletcher Dr. 241 167 109 25.2 14.5 10.3
Northridge 1994 6.7 24283 Moorpark — Fire Station 286 189 143 20.4 20.1 6.6
Northridge 1994 6.7 24389 LA — Century City CC N. 217 217 113 25.2 25.2 8.7

30000 30000
25000 25000
20000 20000
Base Shear Force (kN)
Base Shear Force (kN)

15000 15000
10000 10000

5000 5000

0 0
-20 -10 0 10 20 30 40 -20 -10 0 10 20 30 40
-5000 -5000

-10000 -10000
Dynamic Analyses
-15000 Erzincan Hysteresis -15000
Pushover Analysis -20000
-20000
(a) Top Displacement (mm) (b) Top Displacement (mm)

Figure 8. Dynamic capacity curve of the structure developed from peak values of top displacement and base shear pairs of dynamic
analyses results in (a) x direction and (b) z direction.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 859

(a) Pushover Analysis (b) Dynamic Analysis (ERZ) (c) East Wall
Figure 9. Damage distribution determined by the analyses and the current photo showing that the stones in the same area have
changed.

taken as the displacement capacity of the structure at in- and out-of-plane shear forces were calculated by
the top of the dome. Considering the height of this including and excluding the intersection regions at the
point from ground is 17.45 m, it was concluded that edges of the walls. It is seen in all the graphs that the
the load capacity is reached at a drift ratio of 0.11% and load capacities increase when the intersection areas of
the displacement capacity is limited at a drift ratio of the walls included. While displacement limit was
0.23%. At this point a serious decrease in the load found to be around 6 mm on the walls (at ~8.4 m
capacity develops due to overturning of the north wall elevation) for the elastic in-plane behavior (0.07% drift
and declares the initiation of collapse, as shown in ratio), nonlinear behavior was observed to initiate
Figure 9b. after 15 mm in the out-of-plane behavior (0.18%
Figure 10 displays the in- and out-of-plane dynamic drift). Considering these values, it can be concluded
load capacity curves of four walls developed from the that masonry structures behaves as rigid until the
peak values of top displacement and base shear force strength is exceeded under earthquake loading and
pairs of dynamic analyses results. The displacements once the strength is exceeded, cracks and damage
were derived in the middle-top of the wall. The progress very quickly.

with intersection areas without intersection areas idealized capacity

East Wall West Wall North Wall South Wall


10000 10000 6000 6000
a) in-plane b) in-plane c) out-of-plane d) out-of-plane inwards
7500 7500
4000 4000
5000 5000
x direction (N-S)
Force (kN)

2000 2000
2500 2500
outwards
0 0 0 0
-2500 -2500
-2000 -2000
-5000 -5000
-4000 -4000
-7500 -7500 inwards outwards
-10000 -10000 -6000 -6000
-2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 -4.0 -2.0 0.0 2.0 4.0 -4.0 -2.0 0.0 2.0 4.0
6000 6000 10000 10000
e) out-of-plane f) out-of-plane g) in-plane h) in-plane
inwards 7500 7500
z direction (E-W)

4000 4000
5000 5000
Force (kN)

2000 2000
2500 2500
outwards
0 0 0 0
outwards -2500
-2000 -2500
-2000
-5000 -5000
-4000 -4000
inwards -7500 -7500
-6000 -6000 -10000 -10000
-4.0 -2.0 0.0 2.0 4.0 -4.0 -2.0 0.0 2.0 4.0 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0
Drift Ratio (‰) Drift Ratio (‰) Drift Ratio (‰) Drift Ratio (‰)

Figure 10. Dynamic capacity curves of four walls developed from peak values of top displacement and base shear pairs of dynamic
analyses results.
860 I. KOCAMAN ET AL.

As shown in Figure 10a–h, displacements calculated of this rocking motion. Simultaneously, the parallel
on the walls resisting earthquake loads by in-plane opposite wall bends inward the structure, and the ver-
action can only reach low values of lateral displace- tical walls at the edges act as support against this
ment. Although the lowest values proposed in literature bending motion and restrict the movement. The out-
for the elastic modulus of the masonry material were of-plane load capacities of the walls on the compression
used in this study, these drift values stipulate the careful and tension sides are 4000 kN and 2400 kN in average,
use of proposed damage limits for such structures and respectively, without intersection regions. Adding up
walls. We believe that this discrepancy raises from two these values, the total horizontal load capacity for the
limitations: using scaled single isolated wall tests that four walls is calculated as 20,400 kN approximately,
cannot represent the complex global interaction of dif- which is very close to actual capacity of 21,767 kN as
ferent elements in the entire structure and deriving given in Figure 8a. Therefore, it can be stated that in
drift limits particularly from test under monotonic the seismic load capacity evaluation of thick-walled
and quasi-static cyclic loading that drifts the specimen masonry structures, not only the walls parallel to earth-
to imposed displacement, which cannot reflect the iner- quake excitation direction but also the out-of-plane
tial hysteric effect on masonry disintegration. behaving vertical walls should be taken into considera-
The in-plane load capacities of the four walls have tion. The out-of-plane response depends not only on
similar values, which is 7000 kN approximately includ- the seismic loading intensity and wall geometry but also
ing intersection regions (Figure 10a–h). Figure 11 illus- on the static and kinematic boundary conditions of the
trates the areas used in shear force calculation on walls. walls. The location of openings on the perpendicular
For out-of-plane behavior, two different behaviors and restraining in-plane working walls, vertical restraint
load capacities emerge depending whether the wall’s that is provided by the roof structure, the stiffness of
rocking (overturning) motion is inside or outside the the horizontal connections determines the rocking
structure. As shown in Figure 10c,d, the load and dis- mechanism on the out-of-plane rocking walls.
placement capacity of the walls on the compression side
of the structure increase due to increasing pressure on
the wall as the wall deflects outward. The wall including 3. A new approach for calculation of
the intersection regions separates from the supporting earthquake load capacity
vertical walls on both sides and topples outwards. The Determination of load capacity of walls of a structure is an
ductile behavior shown in Figure 10c,d occurs because important step in determining the earthquake

Figure 11. Illustration of shear area with and without intersection regions.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 861

performance of the structure. Horizontal load capacity of Depending on the earthquake excitation direction,
masonry walls can be determined according to Equations the calculation model given in Figure 12b is proposed
(3) and (4) analytically. Total length of the walls is to determine out-of-plane lateral load capacity of thick
26.30 m. It is assumed that window openings, which walls with low hd/td ratio (hd/td < 5) based on rigid
have a width of 1.25 m in average, do not carry shear body dynamics and kinetic analysis principles (Kazaz
force. Considering that there are 5 window openings on and Kocaman 2018; Melis 2002). In the model, the
the east and south facades, the net wall length is ld external forces consist of the vertical loads Wup trans-
= 20.05 m, if the intersection areas are included and mitted from the upper structural elements and resulting
16.45 m if the intersection areas are excluded. The wall from inertial effects Ne, and the normal reactions and
thickness is td = 1.8 m. In the Turkish Seismic Code, the base shear forces at the walls depending on the rocking
wall shear strength, fvk, is determined by Equation (3): direction. The vertical forces affect the tipping resis-
tance of the wall. The earthquake force Fco that leads to
fvk ¼ fvko þ 0:4σ d  0:10fb (3) toppling of the wall and therefore is the load capacity of
wall is calculated as
 
td W d
Vwall ¼ ld  td  fvk : (4) Fco ¼ þ Wup þ Ne : (5)
he 2
Here, the value of characteristic shear strength fvko of In the equation, Wd is self-weight of the wall in
stone masonry is proposed as 0.1 MPa according to static condition, Wup static vertical load transmitted
the same code. From the static analysis, vertical load from roof structure, Ne is the dynamic axial load on
on each wall was calculated as 11,280 kN. Under this wall due to the earthquake load reversals, he is
load, the average vertical compressive stress, σd, was height of effective total earthquake load form
found as 0.313 MPa. When these values are used, the the ground level, and td is the wall thickness.
wall shear strength is obtained as fvk = 0.225 MPa. It is assumed that the contact between the wall and
Using this value in Equation (4), in-plane load capa- the floor is rigid and pointed. In fact, contact at the base
city of a single wall is determined as 8120 kN including of the wall during rocking is not at a point, but on
and 6662 kN excluding intersection areas. However, the a certain area. However, since the width of the contact
collapse mechanism obtained from dynamic analysis area obtained with similar models based on rigid body
displayed that one of the intersection areas at the dynamics is very small (<100 mm) as compared to
ends of in-plane working walls remains inside the thickness of the wall and this assumption is supported
volume of out-of-plane overturning wall (north wall) by the finite element calculations, it is assumed that the
as shown in Figure 12a. Therefore, the net wall length ld contact is along the outer base line to simplify the model.
should be calculated by excluding window openings In some models, material flexibility of the pivot point (or
and one of the intersection regions at the ends. By area) is also considered. Such a contact definition may
this way, the in-plane load capacity of the east and affect both the displacement and the load capacity. Since,
west walls is calculated as 7391 kN. the material of the dry stone masonry outer leaf of the

Figure 12. (a) Erzincan earthquake simulation principal strain plot on main walls at t = 2.98 sec: the north wall starts to tip outside of
mosque as a block, and (b) rigid block model of the north wall.
862 I. KOCAMAN ET AL.

three-leaf walls has the adequate compressive strength αMTSa attached at the center of mass and yc is the vertical
against crushing, such a phenomenon is also ignored. coordinate of the center of mass. Taking moment about
We will focus on the determination of he and Ne in the contact point on the south wall, dynamic axial load is
Equation (5) since these values directly affect the result. derived as
The dynamic axial load (Ne) in the model was deter-
αMT Sa ðTÞyc
mined initially by dynamic analyses results. Figure 13 Ne ¼ : (6)
displays the force and displacement time-histories of le
the north and south walls obtained from the Erzincan It is recommended to use the effective period Te instead
earthquake analysis. In the graphs, dynamic axial force of the initial elastic period Ti for the building period in
variation (wall static axial load was removed), wall out- the calculation of spectral acceleration in Equation (6) by
of-plane shear force, and scaled displacement were several codes and previous studies, in order to account
plotted to investigate the phase relation between these for the reduction in stiffness under earthquake effect.
quantities. It was observed that when the north wall The effective period can be taken as 1.5 times of initial
overturns outward, displacement and lateral load (base period, Te = 1.5Ti (Kadas, Yakut, and Kazaz 2011).
shear) increases simultaneously as the dynamic axial Now, the application of this method is displayed for
compressive load (Ne) on the wall also increases. The Erzincan earthquake. Maximum axial dynamic force
dynamic axial load on the opposite sides of the excita- from the dynamic analysis of Erzincan ground motion
tion direction takes opposite signs and considering the is 5483 kN. As a result of the modal analysis of the
rocking phases indicated in Figure 13, the absolute structure, the dominant mode frequency in x direction
value of the dynamic axial load may be taken as same was determined to be 3.80 Hz (Ti = 0.26 sec). Sa(Te)
both in compression and tension. In light of this beha- value was obtained as 6.90 m/s2 from Erzincan ground
vior, the model given in Figure 14 was developed. motion acceleration spectrum. In this mode, the mass
Dynamic axial load Ne given in Equation (6) varies participation ratio in x direction is α = 0.55 and the
depending on the intensity of the ground motion acting total mass is MT = 5679 tons. When the values given
on the structure. The dominant modal shape and mass above are input to Equation (6), the dynamic axial load
participation ratio in the direction of earthquake excita- is Ne = 6062 kN (with yc = 6.3 m, le = 22.4 m). As it is
tion are also factors affecting the determination of Ne. In seen, the estimated dynamic axial load is very close to
the free body diagram given in Figure 14, the structure the analysis value. Comparison of the Ne values
and walls are in translation under horizontal earthquake obtained from dynamic analyses under different earth-
forces, vertical loads due to the weight of the structure are quakes with the Ne values calculated by Equation (6) is
excluded since they cancel out with static vertical reaction given in Figure 15. The correlation indicates that the
forces at the wall base, and only the dynamic axial tensile- proposed relation can be used to calculate Ne.
compressive force couple Ne caused by inertial effects of North wall is under compression in the earthquake
the horizontal earthquake load are illustrated. In Figure cycle that causes the mosque to collapse as shown Figure
14, the effective earthquake forces reduce to the vector 12a, previously. The wall is exposed to axial force of

8000
x800 magnified
(+) Tension

Phase I Phase II Phase III


6000 horz. displacement
4000
Force (kN)

2000
0
-2000
Compression (-)

Ne (North Wall)
-4000
Ne (South Wall)
-6000 Vx,base (North Wall)
-8000
1 2 3 4 5 6
Time (sec)

Figure 13. Variation of dynamic axial and lateral load on the North and South walls. In Phase I North wall is moving inside and Ne is
tensile; in Phase II North Wall is rocking outside and Ne is compressive; in Phase III the intensity of the ground motion leads to
toppling of the wall outward.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 863

In fact, it is useful to assign a greater value for td in this


calculation. Because the 0.5 m projections of the arches
joining to wall and 1.2 m thick bedplate of the entrance
door increases the distance of the vertical loads to the
pivot point, as can be seen in Figure 11. The models
shown in Figure 12b can be arranged in detail accord-
MT.Sa
ing to the geometry of the structure to obtain more
precise results.
h=8.4 m

CM
yc=6.3 m Assuming that the distribution of accelerations is
triangular and the mass is uniformly distributed along
the height of the walls, the resultant of earthquake load
le=22.4 m Ne
Ne on the north and south walls can be calculated as

Figure 14. Determination of dynamic axial force on the out-of- Md  Sa


plane acting walls caused by lateral seismic force. Fo ¼ : (7)
2

In the dynamic cycle, where the north wall toppling


8000 outside of the structure and leading to collapse, the
south wall is bending inside the structure. The hor-
Equation (6), Ne (kN)

y = 1.0011x
R² = 0.7448
6000 izontal earthquake load on the south wall can be
simply calculated by Equation (7). Comparison of
4000 seismic loads estimated using Equations (5) and (7)
with the lateral seismic loads (base shear force)
2000 obtained from dynamic analyses for north and south
walls are given in Figure 16. The resulting graphs
0 demonstrated that the proposed model estimates the
0 2000 4000 6000 8000 dynamic horizontal force with acceptable accuracy
Dynamic Analyses, Ne (kN) and can be used to determine the horizontal earth-
quake loads on historical mosques and similar four-
Figure 15. Comparison of the axial dynamic load on the north
walled box type structures.
wall from Equation (6) and dynamic analysis.
At this point it is useful to note that the structure
and walls did not go through severe damage or col-
lapse in all earthquake analyses. In fact, the structure
10804 kN (Wd) due to its own weight and 2847 kN (Wup)
behaved linearly or close to linear in the analyses with
due to dome. Using Equation (6), the dynamic axial force
some full scale (1.0) and all half-scaled (0.5) records.
(Ne) was calculated as 6062 kN. As shown in Figure 12a,
In this case, Equation (5), which is proposed to deter-
the intersection areas and even small parts of the east and
mine the out-of-plane load capacity of the walls, cal-
west walls breaking from the window openings participate
culates the earthquake load applied on the north wall,
in the overturning motion of the north wall. The west and
which has not reached to the capacity, excessively,
east walls which are perpendicular to the north and south
since Equation (5) was derived from a kinetics model
walls, behaves as restraining support when the north and
representing the ultimate load limit state of the rock-
south walls move inwards the structure. But the west and
ing walls. Therefore, in the region left in the box in
east walls loses their restraining effect when the north and
Figure 16a, the model gives greater values than the
south walls move outwards the structure, as displayed in
analyses results and Ne is required to be limited after
Figure 12a. Knowing that mass is uniformly distributed
yielding of the wall. Amplitude of the ground accel-
along the height of the walls, distribution of dynamic
eration or in other words the spectral acceleration,
forces acting on the north wall can be regarded as trian-
which will cause the walls to collapse out of plane,
gular. It is possible to get he = 0.67hwall in this way. Using
can be calculated with the help of the existing equa-
these values, the out-of-plane load capacity to the north
tions. By equating Equations (5) and (7) and distribut-
wall can be calculated by Equation (5) as
  ing Ne expressed by Equation (6) inside the relation
1:8 10804 and arranging for Sa, the lower limit of spectral accel-
ðFco ÞNW ¼ þ 2847 þ 6062
0:67  8:4 2 eration that will lead to toppling of the wall is calcu-
¼ 4577kN : lated as:
864 I. KOCAMAN ET AL.

6000 6000
R² = 0.7894 R² = 0.6409

Analyses, Fo (kN)

Analyses, Fo (kN)
4000 4000

2000 2000

0 0
0 2000 4000 6000 0 2000 4000 6000
(a) (b)
Equation (5), Fo (kN) Equation (7), Fo (kN)

Figure 16. Horizontal force correlation: (a) North wall and (b) South wall.

ðWd þ 2Wup Þ three different analysis methods: dynamic analyses, sta-


Sa;c ¼  : (8) tic analyses, and the proposed kinematic analyses. The
Md he
td  2 αMleT yc
horizontal seismic load capacity of the mosque was
Here, Md is the mass (self and roof) of the north wall obtained as 21,767 kN, 25,248 kN, and 22,250 kN for
(1392 ton). In the Erzincan case, Sa,c that will start the dynamic, pushover, and kinetic analyses approaches,
collapse mechanism in the structure is 6.36 m/sec2. respectively. It was seen that the kinetic method yields
This explains the destructive effect of Erzincan earth- reliable and safe results for the seismic load capacity
quake [SaERZ(Te) = 6.9 m/sec2] on the structure. In the evaluation of masonry mosques.
calculation of dynamic axial load Ne using Equation (6),
the largest value of spectral acceleration Sa should be 4. Case study of Erzurum Narmanlı Mosque
capped with Sa,c, since the structure will be in a yield
state reaching maximum load. For values of Sa smaller 4.1. Historical and architectural features
than Sa,c, the actual spectral acceleration is used in the Narmanlı Mosque, which is located in the east of
calculation of Ne. In such a case, the horizontal seismic Double Minaret Madrasah in the Tabriz Gate district
load on all walls is calculated using Equation (7). in Erzurum city center, was built by Hadji Yusuf Efendi
We now have all the equations needed to calculate the in 1738 according to the four-line inscription on the
structural load capacity. In the case of Sa = 6.36 m/sec2, door. The mosque is one of the important Ottoman
the capacity of the north wall is calculated by using monuments in Erzurum and under the responsibility of
Equations (5) and (6) as: the General Directorate of Foundations. The view and
 plan drawings of the mosque are given in Figure 17.
1:8 10804
Fco ¼ þ 2847 Made of smooth cut stone, the mosque attracts atten-
0:67  8:4 2
  tion with its workmanship.
0:55  5679  6:36  6:3 Narmanlı Mosque is a main single-domed mosque
þ
22:4 with a square plan. There are four quarter domes around
a single main dome. The dome sitting on the four tromps
at the corners inside is reflected to outside by a 16-sided
¼ 4425 kN:
drum at the perimeter, preventing the lateral expansion of
Effective earthquake load on the south wall (no col- dome. The main dome has 15.2 m diameter, 17 m height,
lapse) using Equation (7) is and approximately 65 cm thickness. The main dimen-
sions of the mosque are 19 × 19 m2. The narthex is
957  6:36
Fo ¼ ¼ 3042 kN: 19 × 4.5 m2 in the southern part. The narthex is consist
2 of five domes and six columns. Thickness of four main
Previously, the in-plane load capacity of two walls walls of the mosque is approximately td = 170 cm. The
parallel to the excitation direction was obtained as walls has a height of hd = 8.66 m to the roof level. The ld/hd
7391 kN for each. By kinetic method, the total load and hd/td ratios of the mosque walls are 2.19 and 5.1,
capacity of the structure is calculated as 22,250 kN by respectively. There are four windows on the east and
summing the in- and out-of-plane load capacities of west facades of the masque and two windows on the
four walls. In summary, the earthquake load capacity of north and south facades. The minaret is located on the
the historic Lala Pasha Mosque was determined using north-west corner of the mosque.
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 865

Figure 17. Erzurum Narmanlı Mosque and its plan.

4.2. Finite element model Table 4. Mass participation ratio (MPR) and frequency values
for the first four modes of Narmanlı Mosque.
As shown in Figure 18, detailed solid model of Narmanlı MPR MPR MPR
Mosque was produced using SolidWorks software. The Frequency (x direction (y direction (z direction
Mode (Hz) Horizontal) Vertical) Horizontal)
generated model was transferred to ANSYS APDL soft-
3 3.86 0.48 0.60E-4 0.19
ware and model was meshed with 15,657 nodes and 62633 4 3.89 0.19 0.82E-4 0.48
elements using SOLID65 element. The same material 5 4.82 0.42E-2 0.21E-4 0.11E-2
6 4.92 0.35E-3 0.60E-5 0.39E-2
properties of Lala Pasha Mosque given in Table 2 was
also used for Narmanlı Mosque.
first four main body modes are emphasized. Table 4
4.3. Modal analysis shows the mass participation ratios and frequency values
for the dominant four modes.
Mode shapes and vibration periods of Narmanlı Mosque
was obtained by modal analysis. The total mass of the
mosque was calculated as 3747 tons. The analysis was
4.4. Calculation of the seismic capacity of
made in 20 modes and the first 2 modes came out of the
Narmanlı Mosque with proposed approach
minaret’s modes as in the case of Lala Pasha Mosque. In
fact, minarets are the most vulnerable parts of mosque The proposed kinetic analysis method was used to deter-
structures against seismic action and susceptible to exten- mine the seismic capacity of Narmanlı Mosque. Equation
sive damage under severe earthquake loading. These (4) was used to calculate the in-plane load capacity of
modes were excluded from the model and analysis, and masonry walls. Length of the walls is 19.00 m. Window

(a) (b)

Figure 18. Narmanlı Mosque: (a) SolidWorks model and (b) ANSYS model.
866 I. KOCAMAN ET AL.

openings, which are 1.50 m in average, were subtracted Effective earthquake load on the south wall (no col-
from the length of the walls. Considering that there are 3 lapse) was given by Equation (7) as
window openings on the east and south facades, the net
762:8  5:93
wall length ld is calculated as 15.55 m if the intersection Fo ¼ ¼ 2262 kN:
2
areas are included and 12.15 m if the intersection areas are
excluded. However due to assumed overturning mechan- The in-plane load capacity of the walls parallel to the
ism of the walls as discussed previously, one of the inter- earthquake direction was obtained as 4831 kN. The total
section region was included in the calculation of the wall load capacity of the structure was found as 15,224 kN
length and it was found as ld = 13.85 m. The wall thickness summing the individual in- and out-of-plane load capa-
is td = 1.7 m. From the static analysis, vertical compressive cities as 2x4831 kN + 3300 kN + 2262 kN.
stress σd was found as 0.263 MPa. Using these values, the
wall characteristic shear strength was obtained as fvk 4.5. Pushover analysis
= 0.2052 MPa. When this value is inserted in Equation
(4), the in-plane load capacity of the each wall parallel to To verify the estimated load capacity, nonlinear pushover
the earthquake direction was obtained as 4831 kN. and time-history analyses were performed on the
The free-body diagram given in Figure 19 was used Narmanlı mosque. The acceleration profile acting on the
for determining Ne value. As a result of the modal middle axis of the structure was calculated as shown in
analysis of the structure, the dominant mode frequency Figure 20 using the modal combination under acceleration
in x direction was determined to be 3.86 Hz (Ti = 0.259 response spectrum suitable for ground and location of the
sec). In this mode, the mass participation ratio in structure. In order to easily input the acceleration field that
x direction is α = 0.48 and the total mass is MT is applied on the structure, the profile was idealized as
= 3747 tons. When the values given above were written average acceleration acting on the body and roof part of
in Equation (6), the dynamic axial load was found as the structure as shown in Figure 20a. Using the force-
Ne = 4105 kN with yc = 5.16 m and le = 15.6 m. controlled pushover analysis, the lateral load capacity of
Spectral acceleration that causes the toppling of walls can the structure was calculated as 18,500 kN approximately.
be calculated by Equation (8). Md is the mass carried by the
north wall (1113.5 tons). In this case, Sa,c that initiates the
4.6. Time-history analysis
collapse mechanism in the structure is 5.93 m/s2.
Using Sa = 5.93 m/s2, the out-of-plane load capacity of The time-history analyses of Narmanlı mosque was per-
the north wall was calculated employing Equation (5) as formed under 1992 Erzincan, 1992 Cape Mendocino and
 1999 Kocaeli earthquake records, for which the peak
1:7 10923 values are given in Table 3. In Figure 21, the base
Fco ¼ þ 2274
0:67  8:66 2 shear force-displacement dynamic hysteresis curves are
  plotted for these earthquake scenarios. While displace-
0:48  3747  5:93  5:16
þ ment was found to be around 4 mm on the main dome
15:6
(at 16.95 m elevation) for the limit of elastic in-plane
behavior of walls (0.025% drift), nonlinear behavior
¼ 3300 kN: initiation was observed after 16 mm of dome displace-
ment in the out-of-plane behavior of walls (~0.1% drift).
Total load capacity of the mosque was calculated
approximately as 18,385 kN. Comparing this value with
the estimated seismic load capacity of 15,224 kN, it can
be concluded that the proposed method safely and clo-
sely estimates the seismic load capacity. The error in the
prediction is around 17%, when compared to Lala Pasha
Me.Sa Case.
h=8.66 m

CM

yc=5.16 m
5. Discussion and conclusion

le=15.6 m
This study aims to derive simple relations for the seismic
Ne Ne
load capacity evaluation of Ottoman period masonry
Figure 19. Determination of dynamic axial force on the out-of- mosques. These domed structures seated on four thick
plane acting walls caused by lateral seismic force. walls and pillars anchored each other with supporting
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 867

17.5 20000
Simplified
15 Acceleration Profile

12.5 15000
Acceleration Profile

Height (m)

Force (kN)
obtained by Spectral
10 Analysis
10000
7.5 displacement
control point
5 Wall top
level 5000
2.5

0 0
0 0.2 0.4 0.6 0.8 1 0 20 40 60
(a) (b)
Acceleration (g) Displacement (mm)

Figure 20. (a) Calculated structure height-acceleration distribution relation used MCM and (b) force-displacement graph.

Erzincan Kocaeli Cape Men. Calculated Capacity


20000

15000
Base Shear Force (kN)

10000

5000

0
-90 -60 -30 0 30 60 90
-5000

-10000

-15000

-20000
Displacement (mm)

Figure 21. Top displacement-base shear force graphs of the mosque in comparison to estimated capacity.

arches are suitable to analyze in analogy to rocking The predictions of the proposed method for the seis-
response of rigid blocks. Besides using conventional mic load capacity evaluation of mosques were verified
code shear strength formulations for the in-plane-load against the results of nonlinear pushover and time-
capacity of masonry walls, a methodology was developed history analyses on two different structures in Erzurum,
for the out-of-plane load capacity calculation based on Turkey. Selected mosques reflects different architectural
rigid body dynamics. The effect of axial load variation on forms. While in the Lala Pasha Mosque case the dome is
the load capacity of out-of-plane rocking walls were supported on four pillars and perimeter walls, in
incorporated into the proposed method. The behavior Narmanlı Mosque case the dome is carried by walls. For
of inwards and outwards bending walls were differen- both cases estimated seismic load capacity was in good
tiated. Since not every seismic action leads to collapse, agreement with the analyses results and remained on the
different relations were proposed for the determination safe side. For Lala Pasha Mosque case the errors in load
of seismic load acting on the wall and capacity evalua- capacity prediction are −12% and +2.2% with respect to
tion of the wall. pushover analyses and time-history analyses, respectively.
Assuming the spectral acceleration value is a good For the Narmanlı Mosque case the prediction errors are
indicator of the seismic intensity and considering its −19% and −17% for pushover analyses and time-history
direct relation with the seismic load acting on the total analyses, respectively.
mass, an equation giving the lower limit of spectral Comprising a significant portion of this study,
acceleration that will lead to toppling of the wall is other significant outcomes are in regards to the static
derived. It was shown by the proposed models and and dynamic finite element analyses of the masonry
the equations that the out-of-plane resistance of the structures. For pushover analyses comparing different
thick walls (hd/td < 5) contributes significantly to the lateral load profiles, it was determined and theoreti-
total structural load capacity. cally displayed that the most appropriate load profile
868 I. KOCAMAN ET AL.

for seismic load capacity curve calculation is the mass buildings with flexible diaphragms. Engineering Structures
and mode proportional pattern, where a force distri- 61:195–208. doi:10.1016/j.engstruct.2013.12.038.
bution proportional to the nth mode displacement Betti, M., L. Galano, and A. Vignoli. 2016. Finite element
modelling for seismic assessment of historic masonry
shape is multiplied by the mass distribution as defined buildings. Earthquakes and Their Impact on Society
in Equation (2). The stiffness and maximum load 377–415. doi:10.1007/978-3-319-21753-6_14.
capacity that were evaluated using pushover and Bocciarelli, M., and G. A. Barbieri. 2017. A numerical procedure
dynamic analyses yielded very close results when for the pushover analysis of masonry towers. Soil Dynamics
used as defined here. It was also demonstrated that and Earthquake Engineering 93:162–71. doi:10.1016/j.
soildyn.2016.07.022.
the damage pattern and collapse mechanism could be
Candeias, P. X., A. Campos Costa, N. Mendes, A. A. Costa,
obtained by both analyses methods effectively. and P. B. Lourenço. 2017. Experimental assessment of the
As a final and very important conclusion, it was observed out-of-plane performance of masonry buildings through
that the calculated displacement capacity of masonry mos- shaking table tests. International Journal of Architectural
ques (or structures) was very low, although very low values Heritage 11 (1):31–58. doi:10.1080/15583058.2016.1238975.
of elastic modulus as low as 600–700 MPa for walls and Cennamo, C., B. M. Chiaia, S. D’Angelo, and D. Ferretti.
2011. Seismic assessment and rehabilitation of a historical
domes were used in the models. The drift ratio on the model theater based on a macro-element strategy. International
depends on the choice of control point in masonry struc- Journal of Architectural Heritage 5 (3):264–95. doi:10.1080/
tures due to lack of diaphragm effect. For the Lala Pasha 15583050903560256.
Mosque the dome top and wall out-of-plane ultimate drift Costa, A. A. 2012. Seismic assessment of the out-of-plane
ratios can only reached to 0.25% and 0.5%, respectively. For performance of traditional stone masonry walls. PhD
diss., Faculty of Engineering of the University of Porto.
Narmanlı Mosque the dome top drift ratio reached to 0.5%
D’Ambrisi, A., V. Mariani, and M. Mezzi. 2012. Seismic
at ultimate. These values are very low according to existing assessment of a historical masonry tower with nonlinear
drift limits for masonry walls. The larger drift capacity of the static and dynamic analyses tuned on ambient vibration
Narmanlı Mosque can be attributed to its smaller size and tests. Engineering Structures 36:210–19. doi:10.1016/j.
architectural form, where the dome is only carried by walls engstruct.2011.12.009.
and the drum at the perimeter of the dome provides more DBYBHY. 2007. Code on structures to build in earthquake regions.
General Directorate of Disaster Affairs. (in Turkish).
resistance against lateral deflections. We believe, the research de Felice, G. 2011. Out-of-plane seismic capacity of masonry
on masonry structures needs to be move forward to dis- depending on wall section morphology. International Journal
placement capacity evaluation by conducting dynamic tests of Architectural Heritage 5 (4–5):466–82. doi:10.1080/1558
on structural system level rather than on component speci- 3058.2010.530339.
mens and employing reliable nonlinear finite element ana- Derakhshan, H., Y. Nakamura, J. M. Ingham, and M. C. Griffith.
2017. Simulation of shake table tests on out-of-plane masonry
lysis procedures for validation.
buildings. Part (I): Displacement-based approach using sim-
ple failure mechanisms. International Journal of Architectural
Heritage 11 (1):72–78. doi:10.1080/15583058.2016.1237590.
ORCID Doherty, K., M. C. Griffith, N. Lam, and J. Wilson. 2002.
Displacement-based seismic analysis for out-of-plane bending
Irfan Kocaman http://orcid.org/0000-0002-1774-7114 of unreinforced masonry walls. Earthquake Engineering &
Emriye Kazaz http://orcid.org/0000-0002-3885-1885 Structural Dynamics 31 (4):833–50. doi:10.1002/(ISSN)1096-
9845.
Endo, Y., L. Pelà, and P. Roca. 2017. Review of different
pushover analysis methods applied to masonry buildings
References and comparison with nonlinear dynamic Analysis. Journal
of Earthquake Engineering 21 (8):1234–55. doi:10.1080/
ANSYS 16.1. 2015. Finite element analysis program.
13632469.2016.1210055.
Canonsburg, PA, USA: ANSYS Inc.
Erzurum Environmental Status Report. 2011. Erzurum
Bartoli, G., M. Betti, P. Biagini, A. Borghini, A. Ciavattone,
Governorship. Erzurum: Environmental and Urban Planning
M. Girardi, G. Lancioni, A. M. Marra, B. Ortolani,
Provincial Directorate.
B. Pintucchi, et al. 2017. Epistemic uncertainties in struc-
Eurocode 6: Design of masonry structures. 1995. Part 1.1: General
tural modeling: A blind benchmark for seismic assess-
rules for reinforced and unreinforced masonry structures,
ment of slender masonry towers. Journal of Performance
European Committee for Standardization, ENV 1996-1-1:1995.
of Constructed Facilities 31 (5):04017067. doi:10.1061/
Ferreira, T. M., A. A. Costa, and A. Costa. 2015. Analysis of
(ASCE)CF.1943-5509.0001049.
the out-of-plane seismic behavior of unreinforced masonry:
Bartoli, G., M. Betti, and A. Vignoli. 2016. A numerical study
A literature review. International Journal of Architectural
on seismic risk assessment of historic masonry towers:
Heritage 9 (8):949–72. doi:10.1080/15583058.2014.885996.
A case study in San Gimignano. Bulletin of Earthquake
Galasco, A., S. Lamomardino, and A. Penna. 2006. On the
Engineering 14:14575–1518. doi:10.1007/s10518-016-9892-9.
use of pushover analysis for existing masonry buildings.
Betti, M., L. Galano, and A. Vignoli. 2014. Comparative
Paper presented at the 1st European Conference on
analysis on the seismic behavior of unreinforced masonry
INTERNATIONAL JOURNAL OF ARCHITECTURAL HERITAGE 869

Earthquake Engineering and Seismology, Geneva, failure mechanisms. Engineering Failure Analysis 47
Switzerland, September 3. (A):129–61. doi:10.1016/j.engfailanal.2014.09.016.
Garofano, A., and P. Lestuzzi. 2016. Seismic assessment of NEHRP. 2000. Guidelines for the seismic rehabilitation of
a historical masonry building in Switzerland: The “Ancien buildings. FEMA 356. Federal Emergency Management
Hôpital De Sion”. International Journal of Architectural Agency, Washington, DC.
Heritage 10 (8):975–92. doi:10.1080/15583058.2016.1160303. Oliveira, C. D. V. 2003. Experimental and numerical analysis
Kadas, K., A. Yakut, and I. Kazaz. 2011. Spectral ground motion of blocky masonry structures under cyclic loading. PhD
intensity based on capacity and period elongation. Journal of diss., Minho University.
Structural Engineering 137 (3):401–09. doi:10.1061/(ASCE) Ordinance 3274. 2003. Technical rules for the design, evalua-
ST.1943-541X.0000084. tion and seismic retrofitting of buildings. Italy: Italian Prime
Kanıt, R., and M. S. Döndüren. 2010. Investıgatıon of using Minister Ordinance.
ANSYS software in the determınatıon of stress behaviors Ordinance 3431. 2005. Last modifications and integrations to
of masonry walls under out of plane cyclıng load. the Ordinance 3274. 2003. Italy: Italian Prime Minister
International Journal of the Physical Sciences 5 (2):97–108. Ordinance.
Karimi, A. H., M. S. Karimi, A. Kheyroddin, and Özkaya, S. G., I. Kazaz, and D. Okuyucu. 2015. Investigation
A. A. Shahkarami. 2016. Experimental and numerical of seismic behavior of arched masonry bridges by finite
study on seismic behavior of an infilled masonry wall elements method. Paper presented at the 5th International
compared to an arched masonry wall. Structures 8 Symposium on Conservation and Consolidation of
(1):144–53. doi:10.1016/j.istruc.2016.09.012. Historical Structures, Erzurum, Turkey November 1. (In
Kazaz, İ. 2016. Seismic deformation demands on rectangular Turkish).
structural walls in frame-wall systems. Earthquakes and Pavlovic, M., E. Reccia, and A. Cecchi. 2016. A Procedure to
Structures 10 (2):329–50. doi:10.12989/eas.2016.10.2.329. investigate the collapse behavior of masonry domes: Some
Kazaz, İ. 2010. Dynamic characteristics and performance meaningful cases. International Journal of Architectural
assessment of reinforced concrete structural walls. PhD Heritage 10 (1):67–83. doi:10.1080/15583058.2014.951797.
diss., Middle East Technical University. Silva, B., M. D. Benetta, F. D. Porto, and C. Modena. 2014.
Kazaz, İ., and İ. Kocaman. 2018. Seismic load capacity eva- Experimental assessment of in-plane behavior of three-leaf
luation of stone masonry mosques. Journal of the Faculty stone masonry walls. Construction and Building Materials
of Engineering and Architecture of Gazi University 33 53:149–61. doi:10.1016/j.conbuildmat.2013.11.084.
(2):543–58. (in Turkish). SolidWorks. 2013. Modeling Software.
Kazaz, İ., A. Yakut, and P. Gülkan. 2006. Numerical simula- Sorrentino, L., D. D’Ayala, G. de Felice, M. C. Griffith,
tion of dynamic shear wall tests: A benchmark study. S. Lagomarsino, and G. Magenes. 2017. Review of out-of-
Computers & Structures 84 (8–9):549–62. doi:10.1016/j. plane seismic assessment techniques applied to existing
compstruc.2005.11.002. masonry buildings. International Journal of Architectural
Kocaman, İ. 2017. Determination of the necessary material proper- Heritage 11 (1):2–21. doi:10.1080/15583058.2016.1237586.
ties for the calculation of dynamic behavior of historic masonry Tuluk, Ö. İ. 2006. Variations on the square based baldachin
structures. MS diss., Erzurum Technical University. (In in ottoman mosques for the concept of space (15th-17th
Turkish). Centuries). Journal of the Faculty of Engineering and
Kuban, D. 1958. Osmanlı dini mimarisinde iç mekan Architecture of Gazi University 21 (2):275–84. (in Turkish).
teşekkülü-rönesansla bir mukayese. İstanbul: Güven Basım TYDK. 2017. Management of Earthquake Risks for Historic
ve Yayınevi. (in Turkish). Buildings. General Directorate of Foundations. (in Turkish).
Magenes, G., and A. Penna 2009. Existing masonry buildings: Vanin, F., D. Zaganelli, A. Penna, and K. Beyer. 2017.
General code issues and methods of analysis and Estimates for the stiffness, strength and drift capacity of
assessment. Eurocode 8 Perspectives from the Italian stone masonry walls based on 123 quasi-static cyclic tests
Standpoint Workshop, 185–98, Napoli, Italy. reported in the literature. Bulletin of Earthquake Engineering
Melis, G. 2002. Displacement-based seismic analysis for out of 15 (12):5435–79. doi:10.1007/s10518-017-0188-5.
plane bending of unreinforced masonry walls. MS diss., Vasconcelos, G., and P. B. Lourenço. 2006. Assessment of the
University of Pavia. in-plane shear strength of stone masonry walls by simplified
Milani, G., and M. Valente. 2015. Comparative pushover and models. Structural Analysis of Historical Constructions 1:843–59.
limit analyses on seven masonry churches damaged by the Willam, K. J., and E. D. Warnke. 1975. Constitutive model for
2012 Emilia-Romagna (Italy) seismic events: Possibilities the triaxial behavior of concrete. International Association
of non-linear finite elements compared with pre-assigned for Bridge and Structural Engineering 19 (3):1–30.

You might also like