See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272423441

Investigation of a historic masonry structure by numerical and operational

modal analyses

Article  in  The Structural Design of Tall and Special Buildings · February 2015

DOI: 10.1002/tal.1213

CITATIONS READS

22 277

5 authors, including:

Halil Nohutcu Ali Demir

Manisa Celal Bayar University Manisa Celal Bayar University
15 PUBLICATIONS   60 CITATIONS    25 PUBLICATIONS   80 CITATIONS   

SEE PROFILE SEE PROFILE

Emre Ercan Emin Hökelekli

Ege University Bartin University
12 PUBLICATIONS   67 CITATIONS    11 PUBLICATIONS   36 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

New Exact Method for Discrete Optimization Problems View project

Mafsal Sürtünme Tipi Sismik Sönümleyici Tübitak 1512 Spherical Friction Damper Design View project

All content following this page was uploaded by Emin Hökelekli on 01 February 2018.

The user has requested enhancement of the downloaded file.

THE STRUCTURAL DESIGN OF TALL AND SPECIAL BUILDINGS
Struct. Design Tall Spec. Build. 2015
Published online in Wiley Online Library (wileyonlinelibrary.com/journal/tal). DOI: 10.1002/tal.1213

Investigation of a historic masonry structure by numerical and

operational modal analyses

H. Nohutcu1, A. Demir1*,†, E. Ercan2, E. Hokelekli3 and G. Altintas1

1
Department of Civil Engineering, Celal Bayar University, Manisa, Turkey
2
Department of Civil Engineering, Ege University, Izmir, Turkey
3
Vocational High School, Afyon Kocatepe University, Afyon, Turkey

SUMMARY
This paper presents the results of model calibration conducted on a historical mosque called Hafsa Sul-
tan in Manisa, Turkey. The finite element model of the mosque was calibrated by the use of the results
obtained from ambient vibration tests of the structure. In order to develop a solid model of the structure,
the dimensions of the structure, defects such as cracks and material degradations in the structure, and the
materials used in different parts were identified. For the evaluation of the material properties of the
structure, nondestructive and destructive testing methods were used. The numerical and experimental
modal parameters of the structure were obtained by finite element method (FEM) and Operational Modal
Analysis (OMA), respectively. The natural frequencies and corresponding mode shapes were obtained
from both FEM and OMA and compared with each other. While a good compatibility was achieved be-
tween mode shapes, some differences between natural frequencies occurred. It was thought that the dif-
ferences resulted from variations in the Young’s modulus of masonry, cracks in elements or boundary
conditions. Therefore, the finite element model was calibrated by changing material parameters. Finally,
a more realistic numerical model of the mosque was put forward and the results were discussed in detail.
Copyright © 2015 John Wiley & Sons, Ltd.

Received 25 May 2014; Revised 5 November 2014; Accepted 24 December 2014

KEY WORDS: historical masonry structure; dynamic parameters; ambient vibration test; Operational Modal Analysis;
finite element method; model calibration

1. INTRODUCTION

Historical structures are the treasures remaining from the thousands of years’ cultural heritage.
They are the world’s cultural heritage, and have to be protected. Today, earthquakes are the main
threat to historical structures. Masonry is the oldest building and unique technique for historical
structures. In order to understand the earthquake behavior of masonry structures, many studies have
been conducted in the literature and earthquake behaviors were investigated using the finite element
method (FEM) (Kocak, 1999; Lourenco, 2002; Ayala and Speranze, 2003; Apostolopoulos and
Sotiropoulos, 2008; Betti and Vignoli, 2008; Bayraktar et al., 2010a; Bağbancı, 2013; Foraboschi,
2013; Gonen et al., 2013; Parisi and Augenti, 2013; Parisi et al. 2013).
The difficulties in FEM analyses of the masonry structures can be listed as an evaluation of ma-
terial properties, boundary conditions, the lack of insight and models for the complex behavior of
units, mortar, joints and masonry as a composite material. Furthermore, existing calculation proce-
dures are mainly of an empirical and traditional nature and the use of numerical tools for the anal-
ysis or design of masonry structures is rather incipient as discussed by Lourenco (2002). Historical

*Correspondence to: Ali Demir, Department of Civil Engineering, Celal Bayar University, PO Box 45140, Muradiye,
Manisa, Turkey.
†
E-mail: ali.demir@cbu.edu.tr

Copyright © 2015 John Wiley & Sons, Ltd.

 H. NOHUTCU ET AL.

masonry structures have been particularly vulnerable to earthquakes and wind loads for thousands
of years. The preservation of historical structures is considered to be a fundamental issue in the cul-
tural life of modern societies as a consequence of bad experiences gained in past decades (Binda
et al., 1992). For this reason, conservation and the structural safety assessment of historical struc-
tures have become an increasing concern. There are many studies concerning this topic in the lit-
erature including both analytical and experimental investigations of these structures (Ercan and
Nuhoglu, 2014).
Because the load-bearing system cannot be determined accurately enough, evaluating the dy-
namic characteristics of historical structures is a very hard task in theory. The behavior of struc-
tures under dynamic loads includes various uncertainties. Along with this, uncertainties in
parameters affecting dynamic behavior complicate a realistic determination of the structure’s dy-
namic behavior. Assessment of natural frequency, mode shape and damping ratio, which are all re-
ferred to as dynamic characteristics in experimental methods, was undertaken to allow a more
realistic achievement of the properties of the available structure (Bayraktar et al., 2010b). Dynamic
structure characteristics were used in a more realistic way of calculating earthquake forces on the
structure and control of the accuracy of the established analytic models. For these reasons, a real-
istic determination of the dynamic characteristics by applying vibration tests on the structures is
rather important. At present, to define the dynamic parameters of engineering problems experimen-
tally, Experimental Modal Analysis (EMA) and Operational Modal Analysis (OMA) methods are
used. To obtain the acceleration data in the EMA method, an external vibration should be applied.
As this is a complicated work to do for big and historical structures, OMA has been preferred more
than EMA. In the OMA method, the vibrations caused on the structure were recorded by acceler-
ometers and the parameters were revealed by analyzing the results (Júlio et al., 2008; Bayraktar
et al., 2009; Bayraktar et al., 2010c; Aras et al., 2011; Foti et al., 2012; Osmancikli et al.,
2012; Bartoli et al., 2013; Gentile and Saisi, 2013).
Bayraktar et al. (2009) investigated a Turkish-style reinforced concrete minaret. The finite element
model of minarets was prepared with ANSYS (Anyss, Inc., Canonsburg, PA, USA; ANSYS, 2006)
and analytical dynamic characteristics such as analytical frequencies and mode shapes. Later, the
experimental natural frequencies, mode shapes and damping ratios were determined via the OMA
method. The analytical model of the minaret was updated to minimize the differences between analyt-
ical and experimental modal properties by changing material properties and boundary conditions. Aras
et al. (2011) determined modal properties of a historical masonry palace in Istanbul with both an
experimental and numerical studies. The experimental study was based on an ambient vibration sur-
vey, while numerical analysis was based on a finite element analysis of the structure. The finite element
model of the structure was updated to minimize the differences between finite element and experimen-
tal modal properties by changing the Young’s modulus of masonry. Calibrated model’s frequencies
and the mode shapes were in accordance with experimental results. Gentile and Saisi (2013) investi-
gated the dynamic properties of two historical structures using nondestructive (NDT) dynamic test
results. One of the structures was a masonry bell tower; the other was an iron arch bridge. The dynamic
characteristics of the structures at different levels of ambient excitation and constant temperature were
determined with operational modal testing in order to check the invariance of the modal parameters.
In the study, firstly, the dimensions of the structure, defects such as cracks and material degradations
on the structure, and the materials used in different parts were determined in detail by making a mea-
surement survey. Secondly, the material properties of the components of the masonry were determined
for the FEM using destructive tests such as density, uniaxial compressive and indirect tension tests.
The NDT methods, ultrasonic pulse velocity and Schmidt hammer testing methods were also used
to evaluate material properties. The solid model of the structure was then developed, and the modal
analysis of the structure was carried out using the finite element program ABAQUS (ABAQUS Inc.,
Palo Alto, CA, USA; ABAQUS, 2010). The NDT dynamic test and OMA were then carried out,
and the dynamic properties of the masonry were extracted using ARTeMIS Modal Pro (AMP; Struc-
tural Vibration Solutions A/S, Aalborg East, Denmark, SVS, ARTeMIS Modal Pro 3.0, 2014). The
results from OMA and FEM were compared, and according to the feedback, the FEM model was
calibrated by changing the material parameters and boundary conditions. Finally, the real material
properties were determined.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

2. DESCRIPTIONS OF THE STRUCTURE, HISTORY INVESTIGATION AND

MEASUREMENT SURVEY

The Hafsa Sultan Mosque is one of the most important landmarks of Ottoman architecture. The
mosque was built for (in memory of) the mother of Suleiman the Magnificent, Ayse Hafsa Sultan,
in the 16th Century. The mosque has an importance for the city of Manisa and for Turkey as a whole.
The mosque is the main building of the Ottoman social complex Hafsa Sultan. The mosque was con-
structed using andesite stone walls and granite columns. The mean thickness of the masonry walls is
1.40 m. The mosque has a great dome and four small domes at the edges of the great dome. The width
of the great dome is 11.8 m, and the width of the small domes is 5.25 m. The plan dimensions of the
mosque are 29.25 × 21.33 m. In addition, there are five 4.2-m diameter domes with six columns in
the courtyard as seen in Figure 1. The domes are supported by walls and arches, and the arches transfer
the load to the foundation via two cylindrical stone columns as shown in Figure 2.

3. MATERIAL TESTS

The structural constituents of the mosque are pink and gray andesite face stones, granite columns and
mortar. In order to determine the parameters needed for finite element modeling, material NDT and
destructive tests were applied on several elements of the mosque’s masonry. Fallen stones from the

Figure 1. External view of Hafsa Sultan Mosque.

Figure 2. Interior view of Hafsa Sultan Mosque.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

structure were taken, and samples that were strong enough to cut out core samples of diameter 54 mm
were selected. In the laboratory, cylinder core samples of diameter 54 mm were drilled out. The heads
of samples were cut using a machine. Stone test samples for indirect tension test (Brazilian test) with a
height of 27–54 mm and samples for uniaxial compression test with a height of 108–120 mm were pre-
pared as seen in Figure 3 (TS 699, 1987; Ulusay et al., 2001). The heads of the cylinder core samples
were grinded in order to have appropriate cylinder samples in the emery machine (TS 699, 1987). Af-
ter the preparation of samples, density, uniaxial compression and indirect tension tests were applied as
shown in Figure 3. The loading rate was set to 0.2 mm/min, which satisfies the failure time of 1–10 min
(TS 699, 1987). Average results are shown in Table 1.
For the NDT tests, before transferring the samples to the laboratory, L-type and LB-type Proceq
Schmidt hammers were used to determine the surface hardness values (rebound value) of the stone.
The stones belonging to the structure were also tested by Schmidt hammer test. Compressive strength
values were calculated from the ‘rebound value-compressive strength scheme’ of Ulusay et al. (2001).
In the laboratory, before indirect and uniaxial compression tests were applied on stone specimens, ul-
trasonic pulse velocity tests were conducted using pundit-type equipment (Figure 4).
The Horasan mortar samples were weak and too small for drilling so only point load tests could be
conducted on samples of arbitrary shape. Results of point load tests revealed that average uniaxial
strength of mortars was calculated to be 6.25 MPa. Tensile strength and Young’s modulus of mortar
were also estimated from the literature with the aid of point load test results (Kocak, 1999). Young’s
modulus and tensile strength were calculated to be 1100 MPa and 1.43 MPa, respectively. The density
of the mortar was calculated to be ρ = 1.34 g/cm3.
Masonry is a composite, and this composite material consists of two or more different constituent
materials. By the use of the homogenization approach, the mortar and stone were assumed to show
a coupled behavior; therefore, the overall behavior of the composite media was taken into account.
While determining the elastic parameters of the masonry mosque, the homogenization equations,
which depend on the strength parameters of constituents, were used. The mosque has two types of ma-
sonry: stone and mortar. The compressive strength of masonry was determined by Equation (1) as de-
scribed by Eurocode 6 (1996):
f k ¼ Kf 0:65
b f m
0:25
(1)

where K is a constant, fb is the compressive strength of stone and fm is the compressive strength of mor-
tar. K is in the range from 0.4 to 0.6 and depends on the morphology of the masonry as described by
Eurocode 6 (1996). K was taken to be 0.5 in this study.
The Young’s modulus of masonry was determined by the use of Equation (2) as described by
Lourenco (2001):
tm þ tu
E ¼ tm tu ρ (2)
Em þ Eu

where tm, tu, Em and Eu are the thicknesses and heights of the stone and mortar, respectively. The co-
efficient ρ varies with the bond between mortar and stone and was taken to be 0.5 for this study as

Figure 3. Drilling out core samples, cutting edges, uniaxial compression and Brazilian test.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

Table 1. Average test results of samples.

Test type Nondestructive tests Destructive tests
Sample Number of Density ρ Rebound Ultrasonic Compressive Tensile Young’s
samples (g/cm3) value R pulse velocity strength strength modulus
(m/s) (MPa) (MPa) (MPa)
Stone 30 2.20 54 1822 30 2.0 10 000
Granite 5 2.60 61 2200 100 8.5 40 000

Figure 4. Ultrasonic pulse velocity test on stone.

described by Lourenco (2001). The shear modulus can be taken to be 40% of the Young’s modulus as
described by Eurocode 6 (1996). The tensile strength of masonry can be taken to be 10% of compres-
sive strength as described by Kocak (1999). The Poisson ratio was taken to be 0.17 for masonry as de-
scribed by Kocak (1999). The elastic material parameters of masonry for the finite element model are
shown in Table 2.

4. NUMERICAL MODEL OF THE HAFSA SULTAN MOSQUE

First, the building survey of the mosque was prepared onsite and a three-dimensional solid model was
formed (Figure 5).
Later on, the finite element model of the structure was generated using the structural analysis soft-
ware ABAQUS (ABAQUS, 2010). In the generation of the finite elements model, solid elements were
used and three various elements were used as stone walls, granite columns and steel tie bars.
In the finite elements analysis of Hafsa Sultan Mosque, 673 527 tetrahedral (C3D4) solid elements
and 145 442 nodes were used to determine the structure’s behavior accurately. In the analysis, it is as-
sumed that the structure is supported by the base as fixed. On the lower and upper sides of the struc-
ture’s columns, there are metal rings acting as hinge. By these metal rings, the columns of the structure
act under pressure. In modeling, these areas were modeled as hinge. Hafsa Sultan Mosque’s three-
dimensional finite elements model in ABAQUS (ABAQUS, 2010) and the meshed version are shown
in Figure 6. In the study, linear perturbation–frequency module, which allows a value eigenvector anal-
ysis to be made about the natural frequencies, was used.
To determine the most convenient mesh range to be used in the model, a convergence analysis was
performed. In the convergence analysis, a frequency analysis was performed for every mesh range for
seed sizes between 1.2 m and 0.3 m, and the frequency values found are given in Table 3 and Figure 7.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

Table 2. Material parameters of masonry for finite element

model.
Material Stone masonry
Compressive strength (MPa) 7.42
Tensile strength (MPa) 0.74
Young’s modulus (MPa) 1500
Shear modulus (MPa) 600
Density (kg/m3) 2200
Poisson’s ratio 0.17

Figure 5. Three-dimensional solid model of Hafsa Sultan Mosque.

Figure 6. Three-dimensional finite element model of Hafsa Sultan Mosque.

As a result of the convergence analysis, optimum mesh range was found to be 0.3 m, and the fre-
quencies belonging to the first five modes found as a result of the frequency analysis performed are
given in Table 4.

5. OPERATIONAL MODAL ANALYSES

In order to define the dynamic parameters of the structures and objects in the analysis of the engi-
neering problems, two types of modal analysis are used. These are the EMA and OMA methods.
EMA is a good verifier of numerical analysis. But, it is not possible to apply EMA for extra-large
structures such as bridges and large-scale buildings. In the OMA method, the vibrations around the
structure due to environmental reasons are detected by accelerometers. This method can be applied
to many structures. The dynamic modal parameters are exposed by the analysis of the results.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

Table 3. First frequency values and number of elements according to the mesh ranges.
Seed size First mode frequency Number of Seed size First mode frequency Number of
ranges (m) values (Hz) meshes ranges (m) values (Hz) meshes
1.20 3.8039 41 992 0.70 3.4603 91 589
1.15 3.7814 42 989 0.65 3.4265 111 931
1.10 3.7546 45 113 0.60 3.3889 130 904
1.00 3.6554 54 224 0.55 3.3607 153 960
0.95 3.6208 56 104 0.50 3.3281 188 648
0.90 3.5958 61 001 0.45 3.2979 235 175
0.85 3.5490 68 067 0.40 3.2542 330 144
0.80 3.5285 71 968 0.35 3.2280 449 443
0.75 3.5059 78 973 0.30 3.1971 673 527

Figure 7. Convergence analysis.

Table 4. Frequency values found as a result of the numerical

analysis.
Mode Frequencies (Hz)
1 3.197
2 5.092
3 5.176
4 5.984
5 6.656

OMA is a special modal analysis method that is performed by collecting real time data from a
structure, which is already in service. The required stimulating effects for OMA are environmental
factors such as vehicle load, wind and machine vibration. Because the environmental effects cannot
be clearly identified, the modal parameters were revealed with different algorithms than EMA. In
this study, ‘the enhanced frequency domain decomposition (EFDD) technique by Brincker et al.
(2000), Peeters (2000) and Jacobsen et al. (2006)’ and ‘the stochastic subspace identification
(SSI) technique by Overschee and De Moor (1996), Peeters (2000), Bayraktar et al. (2009) and
Altunışık et al. (2010)’ are used.

5.1. Results and calibration of numerical model

The experimental measurements carried out on the historical Hafsa Sultan Mosque were obtained by
vehicle, wind, ground motions and vibrations using the OMA method. The possible movement points

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

were defined by the analysis of the structure’s numerical model prepared, and accelerometers were
placed on these points. In the experimental study, 16 uniaxial SENSEBOX-7021 accelerometers and
16-channel TESTBOX-6501 wireless data collection units were used. The collected data were trans-
ferred to the computer using the TESTBOX software. Dynamic characteristics were acquired using
the AMP 3.0 software. To consider the temperature and environmental changes on the structure, two
different tests were performed on different dates. In each measurement, three accelerometers were kept
as reference, and measurements were obtained for X, Y, Z directions from 15 different directions in total.
The points where the accelerometers were placed are shown in Figure 8. The locations on the structure in
ABAQUS (ABAQUS, 2010) of the accelerometers are shown in the numerical model (Figure 8(a)). In
Figure 8(b), the approximate model of the structure was prepared in the AMP software and data collected
from the accelerometers were assigned to the shown points. In Figure 8(c), the accelerometers placed on
the structure are shown. In Figure 8(b), the blue-colored arrows show the place of the reference
accelerometers.
Dynamic modal parameters of the historical Hafsa Sultan Mosque were obtained by the EFDD and
SSI methods in the AMP software. The singular values of spectral density matrices of data set acquired
by the EFDD method are shown in Figure 9(a), and the stabilization diagram acquired by the SSI
method is shown in Figure 9(b).
The results belonging to the first five modes of the structure obtained by experimental and theoret-
ical studies are presented in Table 5. The experimental frequencies were calculated using the EFDD
and SSI methods. Noncalibrated numerical frequencies were found to be approximately 10% greater
than the experimental ones. This difference is thought to be due to the existence of microcracks or
macrocracks on the structure, which is being used for hundreds of years, properties of materials or
the corrosion on the steel tie bars. The cracks on the main dome and the mosque’s walls are shown
in Figure 10.

Figure 8. Locations and directions of the accelerometers on the structure.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

Figure 9. Dynamic characteristics obtained from test setup using the EFDD and SSI methods. (a)
Singular values of spectral density matrices obtained from the EFDD method. (b) Stabilization dia-
gram of estimated state space models obtained from the SSI technique.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

Table 5. The dynamic characteristics acquired before and after the finite element model calibration.
Mode Frequencies Frequencies with Frequencies Maximum difference
with FEM EFDD (Hz) with SSI (Hz) between frequencies (%)
Before After Test 1 Test 2 Test 1 Test 2 Damping EFDD SSI
calibration calibration ratios (%)
1 3.197 2.908 2.902 2.908 2.902 2.903 1.542 0.2 0.2
2 5.092 4.631 4.577 4.580 4.574 4.590 1.070 1.2 1.2
3 5.176 4.668 4.648 4.675 4.669 4.680 1.525 0.4 0.3
4 5.984 5.462 5.510 5.494 5.502 5.480 1.502 0.9 0.7
5 6.656 6.073 6.179 6.133 6.098 6.112 1.848 1.8 0.7
FEM, finite element method; EFDD, enhanced frequencydomain decomposition; SSI, stochastic subspace
identification.

In order to decrease the difference among the numerical results and the experimental ones, the
Young’s modulus for the structure’s elements was defined to be 1210 MPa after numerous trials,
and the frequencies and the damping ratios of the calibrated model are presented in Table 5. The shear
modulus was calculated to be 509 MPa. It can be seen that the calibrated numerical frequencies be-
come very close to the experimental ones. The average differences between the frequencies obtained
with the EFDD and SSI methods and the numerical ones are 0.9% and 0.6%, respectively.
The calibrated numerical mode shapes are given in Figure 11. It is seen that mode shapes ac-
quired with both methods show good compatibility. The structure’s first mode, which is determined
both experimentally and numerically, was on Y direction (Figure 11(a)), second and third modes
were on X direction (Figure 11(b)), fourth mode was on Y direction (Figure 11(c)) and the fifth
mode was on X direction (Figure 11(d)).

5.2. Discussions
In this study, the dynamic characteristics of the historical masonry structure were calculated by using
OMA and numerical methods. The frequencies obtained from the numerical model have been

Figure 10. Structural cracks.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

Figure 11. Numerical and experimental mode shapes. (a) First mode. (b) Second and third modes. (c)
Fourth mode. (d) Fifth mode.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

calculated to be 10% greater than the frequencies obtained from OMA. This difference was thought to
be sourced by the Young’s modulus of the carrying walls and the cracks on the structure. After numer-
ous trials and adding these cracks on the model, the Young’s modulus in FEM was reduced to
1210 MPa from 1500 MPa and the numerical model of the structure was calibrated. The shear modulus
is calculated to be 509 MPa according to Eurocode 6 (1996). As the frequency values for first five
modes obtained with the numerical model were between 0 and 8 Hz, it was determined to be 0–6 Hz
after calibration. The frequencies obtained experimentally were between 0 and 6 Hz. The difference be-
tween frequencies obtained by OMA and the calibrated numerical model were found to be 0.9% with
the EFDD method and 0.6% with the SSI method. Additionally, a very good compatibility between the
mode shapes of the structure was achieved. The structure’s first and fourth modes were on Y, second,
third and fourth modes were on X direction, and all these were determined using the OMA and numer-
ical methods. A significant difference was not detected between the mode shapes acquired by the
EFDD and SSI methods.

6. CONCLUSIONS

In this study, the dynamic characteristics of Hafsa Sultan Mosque such as frequency, damping ratios
and mode shapes were determined by NDT methods. A solid model of the structure was prepared to
generate a finite element model of the structure by a building survey. The model was prepared in
the ABAQUS software for FEM analyses; the most appropriate mesh range was found to be 0.3 m.
Solid elements were used to construct the finite elements model. The mechanical properties of andesite
stone and Horasan mortar, granite columns and steel tie bars were obtained by using the literature stud-
ies as well as destructive and NDT methods. Frequency and mode shapes were obtained by numerical
analysis to achieve performance of the environmental vibration tests accurately, and the critical points
of the accelerometers to be placed were determined. Besides, the OMA method was used to determine
the dynamic characteristics of the structure experimentally. Accelerometers placed on the structure
were used to detect the environmental vibrations caused by vehicles load, wind load and so on. The
obtained data were used to extract dynamic characteristics of the structure by the AMP software. There
was approximately 10% difference between the frequency values obtained through the numerical and
OMA methods. This difference might be due to microcracks or macrocracks on the structure, boundary
conditions, properties of materials or the corrosion on the steel tie bars. To overcome this difference
and to be able to make a more realistic resolution, the Young’s modulus of the numerical model
was changed step by step and the model was calibrated. The calibrated model and the experimental
results are reasonably compatible. Among the frequency values, there were an average 0.9% difference
in the EFDD method and a 0.6% difference in the SSI method. In addition to this, the mode shapes for
both methods showed great compatibility.
Moreover, the results show that conventional methods which determine the material properties to
obtain estimated model of the masonry type structures are not reliable and can be more reliable by ap-
plying OMA for realistic FEM models.
In this study, the dynamic characteristics of the historical structure were defined in a realistic way by
performing destructive and NDT tests on the structure. It can be concluded that seismic studies, which
incorporate calibrated numerical models, should further be conducted for sustainable structures.

ACKNOWLEDGEMENTS

The research described in this paper was financially supported by the TUBITAK (Project No.
112 M093) and the Scientific Research Project Commission of Celal Bayar University (Project No.
MUH2013-59).

REFERENCES
ABAQUS. 2010. Dassault Systèmes Simulia Corp., Providence, Rhode Island, USA.
Altunışık AC, Bayraktar A, Sevim B, Kartal ME, Adanur S. 2010. Finite element model updating of an arch type steel laboratory
bridge model using semi-rigid connection. Steel Composite Structures 10(6): 543–563.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
OPERATIONAL MODAL ANALYSES, HISTORICAL MASONRY STRUCTURE AND DYNAMIC CHARACTERISTICS

ANSYS. 2006. Swanson Analysis System, USA.

Apostolopoulos C, Sotiropoulos P. 2008. Venetian churches of Lefkada, Greece Construction documentation and seismic behav-
iour “Virgin Mary of the Strangers”. Construction and Building Materials 22: 434–443.
Aras F, Krstevska L, Altay G, Tashkov L. 2011. Experimental and numerical modal analyses of a historical masonry palace.
Construction and Building Materials 25(1): 81–91.
Ayala D, Speranze E. 2003. Definition of collapse mechanisms and seismic vulnerability of historic masonry building. Earth-
quake Spectra 19: 479–509.
Bağbancı MB. 2013. Examination of the failures and determination of intervention methods for historical Ottoman traditional
timber houses in the Cumalıkızık Village, Bursa-Turkey. Engineering Failure Analysis 35: 470–479.
Bartoli G, Betti M, Giordano S. 2013. In situ static and dynamic investigations on the “Torre Grossa” masonry tower. Engineer-
ing Structures 52: 718–733.
Bayraktar A, Sevim B, Altunışık AC, Türker T. 2009. Analytical and operational modal analyses of Turkish style reinforced con-
crete minarets for structural identification. Experimental Techniques 33(2): 65–75.
Bayraktar A, Şahin A, Özcan DM, Yıldırım F. 2010a. Numerical damage assessment of Haghia Sophia bell tower by nonlinear
FE modeling. Applied Mathematical Modelling 34: 92–121.
Bayraktar A, Türker T, Altunışık AC, Sevim B, Şahin A, Özcan DM. 2010b. Determination of dynamic parameters of buildings
by Operational Modal Analysis. Tek Dergi 337: 5185–5205.
Bayraktar A, Sevim B, Altunışık AC, Türker T. 2010c. Earthquake analysis of reinforced concrete minarets using ambient vibra-
tion test results. Structural Design of Tall and Special Buildings 19: 257–273.
Betti M, Vignoli A. 2008. Modelling and analysis of a Romanesque church under earthquake loading: assessment of seismic re-
sistance. Engineering Structures 30: 352–367.
Binda L, Gatti G, Mangano G, Poggi C, Sacchi LG. 1992. The collapse of the Civic Tower of Pavia: a survey of the materials and
structure. Mason International 20: 11–20.
Brincker R, Zhang LM, Andersen P. 2000. Modal Identification from Ambient Responses Using Frequency Domain Decompo-
sition. Proceedings of the 18th International Modal Analysis Conference (IMAC-XVII).
Ercan E, Nuhoglu A. 2014. Identification of historical Veziragasi aqueduct using the Operational Modal Analysis.
Thescientificworldjo DOI:10.1155/2014/518608.
Eurocode 6. 1996. European Committee for Standardization, Brussels, Belgium.
Foraboschi P. 2013. Church of San Giuliano di Puglia: seismic repair and upgrading. Engineering Failure Analysis 33: 281–314.
Foti D, Diaferio M, Giannoccaro NI, Mongelli M. 2012. Ambient vibration testing, dynamic identification and model updating of
a historic tower. NDT and E International 47: 88–95.
Gentile C, Saisi A. 2013. Operational modal testing of historic structures at different levels of excitation. Construction and Build-
ing Materials 48: 1273–1285.
Gonen H, Dogan M, Karacasu M, Özbasaran H, Gokdemir H. 2013. Structural failures in refrofit historical murat masonry arch
bridge. Engineering Failure Analysis 35: 334–342.
Jacobsen NJ, Andersen P, Brincker R. 2006. Using enhanced frequency domain decomposition as a robust technique to har-
monic excitation in Operational Modal Analysis. Proceedings of ISMA2006, Belgium, September 18-20.
Júlio E, Rebelo C, Dias Da Costa D. 2008. Structural diagnosis of the Tower of the University of Coimbra by modal identifica-
tion. Engineering Structures 30(12): 3468–3477.
Kocak A. 1999. Linear and nonlinear analysis of historical masonry structures under static and dynamic loads case study Küçük
Ayasofya mosque. Ph.D. Thesis. Istanbul: Yıldız Technical University.
Lourenco PB. 2001. Assessment of the stability conditions of Cistercian cloister. 2nd International Congress on Studies in
Ancient Structures.
Lourenco PB. 2002. Computations on historic masonry structures. Progress in Structural Engineering and Materials 4:
301–319.
Osmancikli G, Uçak Ş, Turan FN, Türker T, Bayraktar A. 2012. Investigation of restoration effects on the dynamic characteris-
tics of the Hagia Sophia bell-tower by ambient vibration test. Construction and Building Materials 29: 564–572.
Overschee VP, De Moor B. 1996. Subspace Identification for Linear Systems: Theory, Implementation, Applications. Kluwer
Academic Publishers: London.
Parisi F, Augenti N. 2013. Seismic capacity of irregular unreinforced masonry walls with openings. Earthquake Engineering and
Structural Dynamics 42(1): 101–121.
Parisi F, Lignola GP, Augenti N, Prota A, Manfredi G. 2013. Rocking response assessment of in-plane laterally-loaded masonry
walls with openings. Engineering Structures 56: 1234–1248.
Peeters B. 2000. System identification and damage detection in civil engineering structures. Ph.D. Thesis. Belgium: Katholieke
Universiteit Leuven.
SVS, ARTeMIS Modal Pro 3.0. 2014. <http://www.svibs.com>
TS 699. 1987. Methods of Testing Natural Building Stones. Institute of Turkish Standards: Ankara.
Ulusay R, Gökçeoglu C, Binal A. 2001. Rock Mechanics Laboratory Experiments, Turkish Chambers of Geology Engineers,
Ankara, Turkey.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal
 H. NOHUTCU ET AL.

AUTHORS’ BIOGRAPHIES

Halil Nohutcu was born in 1969. He received his BS degree in Civil Engineering from Selcuk
University, Turkey, in 1992; his MS degree from Celal Bayar University, Turkey, in 1996, where
he is currently working as an Assistant Professor; and his PhD degree in Experimental Strengthening
of Prefabricated Structures in Civil Engineering from Osmangazi University, Turkey, in 2007. His re-
search interests include prefabricated structures, experimental study, finite element analyses, and oper-
ational modal analyses.

Ali Demir was born in Karabük in 1983. He received his BS degree in Civil Engineering from Celal
Bayar University, Turkey, in 2005; his MS degree from Celal Bayar University, Turkey, in 2008,
where he is currently working as an Assistant Professor; and his PhD degree in Experimental Strength-
ening in Civil Engineering from Celal Bayar University, Turkey, in 2012. His research interests in-
clude concrete structures, performance evaluation of buildings, experimental study, and operational
modal analyses.

Emre Ercan was born in İzmir in 1977. He received his BS degree in Civil Engineering from İstanbul
Techinical University, Turkey, in 2001; his MS degree from Ege University, Turkey, in 2004, where
he is currently working as a PhD Instructor; and his PhD degree in Civil Engineering from Ege
University, Turkey, in 2010. His research interests include impact-echo, operational modal analysis,
FEM, and steel structures.

Emin Hokelekli was born in Yozgat in 1976. He received his BS degree in Civil Engineering from
Balıkesir University, Turkey, in 2000 and his MS degree from Afyon Kocatepe University, Turkey,
in 2010, where he is currently working as an Instructor in Afyon Kocatepe University. His research
interests include finite element analyses, steel structures, and operational modal analyses.

Gokhan Altintas was born in 1971. He received his BS degree in Civil Engineering from Dokuz Eylul
University, Turkey, in 1992; his MS degree from Celal Bayar University, Turkey, in 1996, where he is
currently working as an Associate Professor; and his PhD degree in Vibration Mechanics in Civil
Engineering from Yildiz Technical University, Turkey, in 2002. His research interests include modal
analysis, voxel-based finite element methods, computer simulations, and reverse engineering
problems.

Copyright © 2015 John Wiley & Sons, Ltd. Struct. Design Tall Spec. Build. 2015
DOI: 10.1002/tal

View publication stats