Vous êtes sur la page 1sur 21

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS Earthquake Engng Struct. Dyn.

2012; 41:1939 Published online 1 April 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/eqe.1115

Minaret behavior under earthquake loading: The case of historical Istanbul


C. S. Oliveira1 , E. akt2 , D. Stengel2,3 and M. Branco1, ,
1 Instituto

Superior Tcnico, ICIST, Lisbon, Portugal University, Istanbul, Turkey 3 Karlsruhe University, Karlsruhe, Germany
2 Bogazii

SUMMARY Minarets are very slender structures with an old existence. The historical ones are made of cut-stone-block masonry assembled in peripheral cylindrical wall with an interior helicoidal stair supported on a central core and on the wall. They are spread throughout the Islamic world and constitute an important heritage not only of religious value, but also of great cultural interest. Throughout the times, these structures as part of a mosque, have suffered signicant damage during the earthquakes. Istanbul presents interesting characteristics to evaluate their dynamic behavior, as they are in great number, in an area where a large event in the next 30 years has been predicted. In this paper, we performed a series of in situ ambient vibration tests to old minarets of various sizes and compared results of frequencies with numerical modeling of the same structures. For the low-amplitude motion, the frequency values of the rst modes can be obtained from an empirical formulae function of the inertia of the cross-section and of the height of the main body. Damping ratios for these amplitudes are of the order of 0.51.0%. Dynamic linear analyses of these structures indicate that for most cases very high stresses develop for PGA above 0.5 g, an input with a reasonable chance of occurring in the next 30 years. These high stresses are expected to cause the toppling of the minarets in the form that has been observed in the recent past events. Copyright 2011 John Wiley & Sons, Ltd.
Received 5 July 2010; Revised 23 January 2011; Accepted 26 January 2011 KEY WORDS:

minarets; masonry structures; seismic performance; in situ frequency measurements; analytical modeling

1. INTRODUCTION The Historical Peninsula of Istanbul is full of very ancient mosques with one or more tall and slender minarets. These masonry minarets stand almost free or with the lowest section as part of the adjacent structure, the main mosque building, above ground. They have very different heights ranging from 15 to 70 m and different designs, with one, two, or three balconies. Along the several centuries of their existence, they have gone through a few earthquakes. For example, during the last earthquakes of 1999 (Kocaeli and Duzce) many minarets, with main structural system of either reinforced concrete or masonry, suffered extensive damage, while others survived with minor damage [1, 2]. Studies revealed high odds for a strong earthquake event to occur in the next 30 years [3]. Therefore, it is important to better understand how these structures behave during moderate to strong shaking, dening the levels of seismic action for which no harm may result and to propose some retrotting policies, for the case when these levels are exceeded.
Correspondence

E-mail:

to: M. Branco, Instituto Superior Tcnico, ICIST, Lisbon, Portugal. memb@civil.ist.utl.pt 2011 John Wiley & Sons, Ltd.

Copyright

20

C. S. OLIVEIRA ET AL.

Figure 1. Location of mosques with studied minarets in Istanbul (number of minarets between brackets): 1Hagia Sophia (#4); 2Yeni Cami (#1); 3Rstem Pasa (#1); 4Sleymaniye (#2); 5S ehzade (#1); 6Akbyk (#1); and 7Mihrimah (#1) (Google Earth).

Several papers have already dealt with the topic [46], but the collection of minarets in Istanbul is so huge in number and variable in character that a comprehensive study using analytical and ambient vibration testing cannot be extrapolated from previous studies. We selected a group of 11 minarets (Figure 1) spanning heights from 23 to 67 m. For all of them we performed in situ testing for identication of at least the rst eight modes and corresponding frequencies. For seven of these minarets we conducted linear numerical dynamic modeling (Figure 2). This selection was made with the aim of analyzing different geometric types and to evaluate how different levels of slenderness affect the seismic behavior of a minaret. The models were calibrated through the comparison of modal frequency values obtained by in situ measurements with the numerical values. We then proceeded to determine the structural response in terms of maximum displacements and stresses for the site-specic ground motion that can be expected in each location.

2. STRUCTURAL DESCRIPTION Classical Ottoman minarets have a standardized assembly of components or segments as shown in Figure 3. Minarets are slender structures, usually of cylindrical form and they can stand alone or be contiguous and integral with the mosque structure. There is essentially a masonry wall tube and an inner core surrounded by a helicoidal stairway going up counter-clock wise, made of single steps spanning from the inner core to the wall. These pieces penetrate half the thickness of the peripheral cylindrical wall, in the bodyshaft section. In some cases there are two parallel stairs as it happens in Hagia Sophia #1. The basic elements of the minaret are: footing, boot/pulpit (kaide), transition segment (kp), cylindrical or polygonal bodyshaft, stairs, balcony (serefe), upper part of the minaret body (petek), spire/cap (klah), and end ornament (alem). They may be built in cut-stone, brick, or a mixture of both. The top is usually a 3-D timber structure covered by 5-mm-thick lead sheets. Iron clamps hold wall blocks together. Above the upper balcony, the helicoidal stairway stops as well as the stone core. A wooden cylindrical column with slightly smaller diameter gives vertical continuity to the inner core until the base of the cap, also serving as a support to a rudimentary vertical wooden stair [7].

3. DAMAGE IN PAST EARTHQUAKES Looking into the past, there were several occasions where minarets in Istanbul suffered from ground shaking [8, 9]. In the 10th September 1509 earthquake, the Hagia Sophia Minarets collapsed [8].
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

21

Figure 2. Minarets modeled: (a) Hagia Sophia #2; (b) Hagia Sophia #3; (c) Hagia Sophia #4; (d) Yeni Cami; (e) Sleymaniye #3; (f) Akbyk; and (g) Mihrimah.

In the 22nd May 1766 and 10th July 1894 earthquakes the minaret of Mihrimah fell [10]. The damage to 75 minarets resulting from the August 17 and November 12, 1999 earthquakes was documented in [6]. The damage distribution reported for 45 minarets in Dzce, Bolu, and Kaynal gives an indication of the extent of damage to minarets in the region. The 19 out of 35 (54%) reinforced concrete (RC) minarets surveyed in Dzce, Bolu, and Kaynal, sustained damage of intensity severe to collapse. Seven out of ten (70%) masonry minarets surveyed in these three cities, sustained higher levels of damage including collapse. The location of the failure in the minarets that collapsed during the 1999 earthquakes was found to be at the region near the bottom of the cylinder, where a transition was made from a circular to a square section. In RC minarets, the reinforcing bars in this section were spliced. Inspection of the failure regions revealed no necking or fracture in the bars. The old masonry minarets were also observed to fail near the bottom of the cylinder, where the lateral stiffness and strength are smaller compared with those of the transition region or the minaret base and where the minaret connects to the adjacent building or is part of it at the lower section. Few cases of minor damage were also observed, such as the collapse of parts of the balcony. However, in a few places where large damage was present in buildings, minarets performed quite well as in the case of Glck during the 1999 Kocaeli earthquake (Figure 4).
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

22

C. S. OLIVEIRA ET AL.

Figure 3. Minaret component description (elevation) of a typical classical Ottoman minaret.

Figure 4. No damage to minaret in heavily damaged area in Glck during the 1999 Kocaeli earthquake (in http://earthquake.usgs.gov).

According to the inventory made by Frat [6] for damage assessment of the historical Istanbul minarets during the Kocaeli 1999 event, none of the selected minarets for this study withstood any particular damage, although two of the mosques presented vertical cracking in the inner core (Sleymaniye #3 and Sehzade). Historical structures, such as the mosques, with heavy and stiff walls are subjected to larger lateral earthquake forces due to their short periods of vibration, when compared with the minarets which are less stiff and have longer periods of vibration [11].
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

23

From many observations and studies [12] for tall minarets of either masonry or RC the most critical zones are the ones above referred. Analytical studies made on a few representative minaret shapes [1] for both masonry and RC structures got essentially to the same conclusions.

4. SURVEY FOR GEOMETRIC AND MATERIAL PROPERTIES The geometric characterization of the minarets under analysis is a very difcult task due to the lack of detailed information, especially for the absence of drawings for most of them. In one single case, the Mihrimah minaret (Figure 2(g)) we were offered the detailed AutoCad drawings, due to the recent reconstruction works undergone. For Hagia Sophia (Figure 2(a)(c)) and Sleymaniye (Figure 2(e)) there were old drawings with general outer dimensions of their minarets. For all the others we did not have any information. To proceed in an approximate way, we analyzed pictures taken by ourselves, and looked at old engravings and all possible material available. To avoid parallax errors we used pictures taken from the top of other minarets. However, in all cases, during the in situ tests we collected the most easily available geometric data such as the number of steps to important locations, step size, thickness of walls, step width, inner and outer diameters at the balcony area, etc. All the gathered information was confronted in order to reduce the errors in the evaluations. Only with a detailed surveying campaign it would have been possible to reduce geometric errors to a minimum. In this work, all but the minarets with complete drawings have some geometric errors. For certain parameters, such as heights and dimensions of the pulpit, errors could be larger than 10%. In fact, the geometric characteristics of the pulpit changed quite signicantly from case to case. In a few cases, the pulpit is part of the main building and it is well conned. This is the case of Sleymaniye #4 and Rstem Pasa. In all the cases, the pulpit functions structurally as a heavy foundation, with not much deformation capacity. The vibration records made in parts till the top of the pulpit show almost no movement as compared with the ones obtained at higher levels, indicating that the exibility of the structure starts at the transition zone. According to the parameterization presented in Figure 5, a database with the geometric characteristics of each minaret was created. In Table I the main characteristics are presented. Most of the minarets were made of stone masonry

Figure 5. Minaret geometrical parameterization.


Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

24

C. S. OLIVEIRA ET AL.

Table I. Main geometric characteristics of the studied minarets (in m) following the parameterization of Figure 5. The seven minarets highlighted in gray are the ones that were modeled (some of the values presented may be subjected to about 10% error).
Hagia Hagia Hagia Yeni Rstem Sophia Sophia Sophia Sleym. Cami Akbyk Mihrimah Pasa Sleym. Sehzade Minaret 12 Total height (m) 66.55 Balconies 1 Body height (h2+h3+h4) (m) 37.90 Wall thickness (wallth 3) (m) 1.00 Body diameter (ext 3) (m) 4.70 3 63.20 1 39.70 0.55 3.10 4 44.96 1 33.60 0.85 3.30 3 74.40 3 54.80 1.13 4.06 1 54.90 3 38.00 0.65 3.00 1 23.02 1 15.10 0.43 2.33 1 38.65 1 30.74 0.30 2.30 1 41.6 1 22.00 0.45 2.48 1 51.7 2 34.80 0.83 3.20 1 48.7 2 30.90 0.60 2.90

(stone blocks cut to measure and assembled on site). However, the Hagia Sophia #4 was made of brick masonry (brick and mortar), with wooden stair steps in some parts. The Akbyk minaret has a mixed system. It has a core of stone masonry and stone stairs. The exterior wall is of brick masonry. Material properties are very difcult to estimate, due to the anisotropy of the masonry and its dependence on the region providing the materials and the construction techniques. While the general weight ( = 2225 kN/m3 ) did not cause great controversy, the modulus of elasticity (E) and the stress limits ( ) in both compression and tension were very difcult to assess. Direct measurement of the mechanical properties is a difcult task due to the impossibility to perform intrusive techniques in historical structures. Only tests such as seismic tomography, radar, etc. can be used. They will produce some local indications as far as the modulus of elasticity of the material is concerned, but overall information is almost impossible to obtain. On ultimate stresses and strains (compression and traction) the problem is even more complex. We proceed in a different way: the modulus of elasticity was calibrated for each case, based upon the results of the dynamic characterization tests. And, based on some of the most recent values published in the literature [13, 14] we considered the compression strength as 26 MPa and the tensile strength was assumed as 10% of that value (2.6 MPa). These values are only indicative and were used to evaluate qualitatively the results obtained through the computer model. The values for the modulus of elasticity present a wide range of values from case to case. We considered as reference for this study, the values used in studies on the Sleymaniye [15, 16] and Hagia Sophia mosques [17, 18]. In these cases, the modulus of elasticity was considered as 3.55.0 GPa for brick masonry and 9.014.0 GPa for stone masonry. The concentrated mass for a balcony such as in Hagia Sophia #2 would be around 38 tons, adding up to the wall weight. The spire mass with a 5-mm-thick lead would be in general of the order of 23 tons.

5. SOIL CLASSIFICATION Although the analyzed minarets were located in the same area with inter separations of not more than a few kilometers (Figure 1), the soil characterization according to the latest studies [19, 20] indicates that there are a few differences as far as geotechnical properties are concerned. In terms of the NEHRP soil classication [21], the minarets analyzed are founded in soils of types C (shearwave velocity between 750 and 350 m/s) and D (shear-wave velocity between 350 and 200 m/s) as presented in Table II. We are aware that in the case of relative soft soil (CD in Table II) and for tall and heavy structures, soilstructure interaction may be very important to dene the rotation at the foundation level. Further studies should be developed to analyze this effect.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

25

Table II. Soil classication of the sites of the minarets under study (courtesy of Zulkar [20]).
Num 1 2 3 4 5 6 7 Minaret Hagia Sophia Yeni Cami Rstem Pasa Sleymaniye Sehzade Akbyk Mihrimah NEHRP soil class C D C C C C D EC8 equivalent soil class B C B B B B C

6. AMBIENT VIBRATION SURVEY A series of in situ ambient vibration tests have already been carried out in different types of minarets [2225] for frequency as well as for mode identication studies. This alternative technique to using other excitation sources has been applied in many different situations and, for exible structures such as minarets, the results are very robust [26, 27]. The main idea of using ambient vibration tests to determine the dynamic characteristics of a structure is essentially based on the principle that this vibration acts at the foundation level as white noise being ltered by the structure along the height. So, the input signal is amplied at the frequencies of the structure at locations where modal shapes exhibit higher expression. By locating instrumentation in several places and recording the response simultaneously in these places, it is possible to identify both the frequencies and the corresponding mode shapes [28]. For slender structures, the response measured only at the top is enough to indicate their fundamental frequency. The results from ambient vibration tests compare very well with alternative techniques which use other excitation sources, such as earthquake ground motion recorded by permanent monitoring networks [23]. In the study we performed identication of both frequencies and mode shapes using three accelerometric high-resolution instruments with three components and common timing. It was possible to obtain for all 11 minarets the rst nine mode shapes, their frequencies and damping ratios through the use of ARTeMIS Extractor software [29]. We will illustrate the used method on the Mihrimah minaret (Figure 2(g)). The others were treated in a similar manner. 6.1. The Mihrimah minaret We used several techniques to characterize the dynamic properties of the minarets applying the ARTeMIS extractor software, already engaged in similar studies in the past [22]. Frequency domain decomposition (FDD), enhanced frequency domain decomposition (EFDD), and stochastic subspace identication (SSI) were used in the estimation of natural frequencies and damping ratios The tests on all minarets were carried out with three Gralp Systems CMG-6TD instruments with three channels each, two for orthogonal horizontal directions, and one for the vertical direction, and 100 Hz sampling frequency. The accelerometers were distributed along the height of the minaret. The rst instrument was at the ground level, the second one was at approximately mid-height between the balcony and ground, and the third instrument was at the highest balcony. In parallel to the ambient vibration recordings, instruments were calibrated during the Mihrimah minaret testing in order to compare the recordings of the three instruments used and to detect any abnormality among them. For the peak of resonance corresponding to the frequencies of vibration of the structure, the three recordings presented the same amplitude, conrming that the three instruments were well calibrated. The test on the Mihrimah minaret, done on February 6, 2009, was the best in terms of quality control with almost 30 min of simultaneous and uninterrupted data from three stations (ground,
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

26

C. S. OLIVEIRA ET AL.

Figure 6. Example of the SVD lines of Mihrimah minaret with the picked mode technique.

Figure 7. SSI stabilization diagrams of Mihrimah minaret.

stair case, and balcony), which led to very stable results in terms of mode shape identication. Straight lines were used to connect the locations of the instruments in drawing the mode shapes. The data are decimated by 5, which results in a decreased Nyquist frequency to 10 Hz. 1024 frequency lines were chosen. In Figure 6 singular values of the signals, and an average of all the signals obtained from the EFDD technique, are shown. It is clearly seen from the gure that natural frequencies are close to each other, and the frequency span for the rst eight modes is 09 Hz. In Figure 7, the stabilization diagram of estimated state space models obtained from the SSI technique is shown. Frequency values obtained from the SSI technique are nearly the same as those found in the EFDD technique. As seen in the peak picking window (Figures 6 and 7), there are usually two coupled peaks, probably indicating two orthogonal modes. In the case of the rst very dominant modes they are found at two different spectral density lines. There are three pairs of peaks to nd before a fourth single peak appears; most likely indicating torsion. That is not in accordance with previously reported results [1], where higher frequencies were obtained for torsional modes. The rst two modes are picked on two different frequency lines, although they are really close. Nevertheless, they are denitely rst modes, as they present the same shape, but in almost
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

27

Figure 8. First two mode shapes: (a) f 1 = 0.801 Hz and (b) f 2 = 0.840 Hz. Table III. Comparison of frequencies and damping ratio obtained from different methods (Mihrimah minaret).
EFDD SSI FDD f (Hz) f (Hz) Damp. (%) f (Hz) Damp. (%) 0.80 0.84 3.22 3.57 4.26 4.53 5.57 0.81 0.84 3.21 3.57 4.25 4.53 5.57 1.14 1.09 1.82 1.24 0.99 0.57 0.81 0.84 3.27 3.59 4.25 4.53 5.57 0.93 0.68 1.88 3.45 1.30 1.25 0.59

Mode 1 2 3 4 5 6 7

Description First-mode NW direction First-mode NE direction Second-mode E direction Second-mode S direction Third-mode E direction Third-mode S direction Torsion

orthogonal directions. The rst two mode shapes of the minaret are shown in Figure 8. The small displacement at about mid-height (point 2) is the result of the inclusion of the minaret into the structure of the mosque up to some height between points 1 and 2. The rst two peaks could be veried using the stochastic subspace identication method. The selected model showed the same results for the rst two mode shapes, but also, the stabilization diagram revealed that there are no stable modes, for some peaks obtained from FDD. Now, both estimators agree with each other and the estimation of the rst seven modes can be compared regarding frequencies and damping ratios (Table III). Actually, it was possible to extract two more modes, #8 and #9 for most of the minarets. For more details see [23]. Modal parameters (frequencies and damping ratios) obtained from both vibration tests and analytical models (see Section 7) are close to each other, which conrm the reliability and robustness of the vibration tests. 6.2. Other minarets The other minarets were subjected to a similar in situ ambient vibration and subsequent data analyses. We placed the instruments at different locations along the height, sometimes placing two at the same balcony (opposite sides) to enhance the torsion modes. From top to bottom the accelerometers were positioned at the upper most balcony, at about mid-height between the ground level and at the ground level. The results are summarized in Table IV. An analysis of Table IV shows that the frequencies obtained are similar in all three techniques used. It can be said that the results in terms of frequency identication seem quite robust, although damping estimates vary considerably. On average, the damping coefcient is around 1% for ambient vibration amplitudes of the order of 2 mg at the top of taller minarets. 6.3. Empirical formula for computing the fundamental frequency of minarets Based on the data collected we developed a simple formula to estimate the rst frequency of vibration known the inertia of a general cross-section and the height of the minaret. According to
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

28

C. S. OLIVEIRA ET AL.

Table IV. Frequencies and damping ratios of the rst seven modes using different approaches.
EFDD SSI EFDD SSI FDD FDD Mode f (Hz) f (Hz) Damp. (%) f (Hz) Damp. (%) Mode f (Hz) f (Hz) Damp. (%) f (Hz) Damp. (%) 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1.32 1.37 3.81 4.00 8.50 8.89 12.6 1.02 1.05 2.71 3.08 4.32 4.59 7.37 1.37 1.38 2.98 3.41 4.93 5.24 7.80 0.95 1.17 4.91 5.18 6.88 7.06 10.86 1.68 1.74 3.47 3.49 5.14 5.15 6.97 Hagia 1.27 1.38 4.09 8.77 8.85 12.6 Hagia 1.03 1.04 2.76 3.01 4.35 4.58 7.44 1.36 1.38 2.97 3.41 4.94 5.24 7.76 Sophia 1 1.44 1.24 2.38 1.38 3.79 7.30 3.99 2.22 8.54 6.25 9.02 0.70 12.58 Sophia 3 1.41 1.03 1.58 1.04 3.51 2.94 0.71 3.13 0.95 4.30 1.51 4.62 0.27 7.21 1.35 1.38 2.94 3.48 4.96 5.35 7.75 1.24 0.88 0.98 1.21 2.57 3.70 3.21 0.75 0.74 2.41 4.93 4.20 1.99 2.21 0.77 0.64 4.69 4.30 2.83 2.01 0.91 2.23 1.01 1.43 1.01 1.59 0.92 2.34 0.91 2.39 1.34 0.43 0.69 0.75 0.94 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1.17 1.27 3.37 4.05 9.64 10.64 12.79 1.03 1.05 3.55 3.81 5.57 5.80 7.46 1.18 1.31 3.31 3.48 5.79 6.11 7.73 0.83 0.84 3.68 3.84 5.49 5.80 6.41 0.63 0.66 2.98 3.08 5.79 6.05 7.54 Hagia 1.17 1.28 3.46 3.75 9.58 10.66 12.78 Hagia 1.04 1.06 3.48 3.75 5.48 5.80 7.47 1.18 1.31 3.31 3.47 5.86 6.12 7.73 Sophia 2 1.94 1.18 1.51 1.28 0.95 3.38 1.57 3.75 3.47 9.64 0.13 10.7 0.47 12.8 Sophia 4 1.56 1.05 0.18 1.08 0.66 3.5 0.42 3.76 1.63 5.51 0.43 5.8 0.20 7.73 1.18 1.31 3.33 3.45 5.84 6.18 7.72 1.22 0.72 1.63 1.41 1.68 1.31 1.85 0.93 1.00 1.60 0.69 2.46 1.01 3.68 0.93 1.16 1.32 1.74 2.15 1.01 1.34 0.21 1.20 0.91 1.7 1.78 0.33 1.83 0.85 1.25 1.48 2.90 4.21 3.69

Rstem Pasa 0.67 0.58 0.64 1.08 1.27 0.99 0.86

Sehzade 0.73 1.19 1.00 0.86 0.91 0.68 0.41

Sleymaniye 1 0.96 1.85 0.99 1.32 1.66 1.18 4.91 0.41 4.90 5.16 0.73 5.10 6.90 0.40 6.95 7.14 0.48 7.13 10.91 1.74 10.96 1.69 1.73 3.46 3.48 5.14 5.16 6.98 Akbyk 0.75 0.63 0.37 0.43 0.13 1.09 0.50 1.68 1.77 3.45 3.59 5.12 5.20 6.86

Sleymaniye 3 0.84 0.85 3.68 0.30 3.7 3.83 0.29 3.86 5.49 0.23 5.5 5.80 0.23 5.83 6.41 0.16 6.42 0.65 0.66 2.97 3.08 5.77 5.95 7.53 Yeni Cami 2.67 2.52 0.91 0.69 0.89 0.56 0.55 0.65 2.97 3.07 5.81 6.26 7.44

Note: FDDFrequency Domain Decomposition; EFDDEnhanced Frequency Domain Decomposition; SSI Stochastic Subspace Identication.

Clough and Penzien [30] the rst frequency of vibration for a cantilever is given by f= EI 1 (1.875)2 2 m L4 (1)

where E is the modulus of elasticity, I the second moment of area, m the mass per unit of length, and L the total height of the cantilever. Therefore, the fundamental frequency of the minaret is expected to be a function of the second moment of area and the height of the minaret (E and m are material properties). The formulation herein considered that the height of the minaret, H relevant for the rst frequency of vibration, was
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

29

Figure 9. Fundamental frequency of vibration (experimental versus empirical) for all the minarets analyzed.

given by the length from the top of the pulpit to the base of the spire (h2+h3+h4 in Figure 5). The relevant cylindrical cross-section was the base of the shaft (Table I). Based upon these parameters a new equation (Equation (2)) was developed to account for this parameterization. f = .A0.5 .I .H (2)

where A is the area of the cylindrical cross-section and I the second moment of area as dened by Equations (3) and (4). The constants were dened as = 38, = 0.7, and = 1.1. These values were determined through an iterative process that minimizes the difference between the experimental results and the empirical equation, considering an average density of 2 ton/m3 and an elasticity modulus E = 10 GPa, as referred in previous studies [1518]. A = [
2 (
ext3 2wallth3.1 )2 ] 4 ext3 I=
ext3 2
4

(3)
4

ext3 wallth3.1 2

(4)

Note: wallth3.1 is the wall thickness at the base of the cylindrical body. In Figure 9 we compare the results obtained with Equation (2) and the experimental in situ values, and errors range to a maximum of 30%. A similar error was observed for the Fatih minaret, which has a height (H ) of around 55 m and an ambient vibration frequency of 0.50.55 Hz [31]. The natural frequency calculated by Equation (2) was 0.72 Hz. If we have more data points from measured minarets, the parameter in Equation (2) could be disaggregated into two other parameters, reecting the mass and modulus of elasticity, and the errors certainly would reduce. That is what we visualize with Mihrimah and also with Mustafa Pasha which are minarets made of a high-quality block masonry (E higher than average). For these minarets, the empirical formula would lead to higher frequency values, reducing the distance to the experimental in situ values (Equation (2) applied to the geometry of Mustafa Pasha Minaret leads to f = 0.68 Hz against 1.04 Hz from ambient vibration). 7. NUMERICAL MODELING We performed linear dynamic analysis of seven minarets selected from the 11 under study, which presented the most distinctive geometry (Figure 2). SAP2000 [32] was the software to
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

30

C. S. OLIVEIRA ET AL.

Figure 10. Finite element model of the Hagia Sophia minaret: (a) 3D view; (b) vertical cross-section; and (c) details of stair modeling.

compute: (i) the dynamic properties, after calibration with the numerical values of the in situ testing; (ii) the dead loads; and (iii) the seismic response to various ground motion scenarios. 7.1. Model description The geometry of the models was dened according to the data collected from the minarets described in Section 4. We started with the Mihrimah minaret for which we had good geometric information and preceded in the same manner to the other minarets. Four main structural element types were used: (i) shell elements to simulate the walls; (ii) shell elements to simulate the stair steps; (iii) stick/frame elements to model the inner core and connect all steps; and (iv) solid elements to simulate the pulpit, whenever it exists. The cap mass was concentrated on the top of the stone shaft, due to the stiffness of the spire being much lower than the remaining structure. This could lead to local modes of vibration with no interest for this analysis. The cap was only placed in the model for visual reasons. No soilstructure interaction was considered. The minarets had the translations and rotations restrained at the foundation level. Figure 10 shows the main elements described above. After the geometrical denition of each model, it is necessary to adjust its main mechanical properties (m, E) so that the values obtained with the numerical model can be compared with the ones in the in situ testing. For this purpose, a few parameters were selected to change the models dynamic characteristics and modify its frequencies of vibration. The density of the masonry was set as 2 ton/m3 , which is slightly below the value for individual stone blocks, to account for the mortar and openings. The frequency was deemed to have little dependence on the stiffness of the steps. Therefore, the value of 10 GPa was considered for the modulus of elasticity (E) of the shell elements used to simulate the stairs, in all minarets studied. The modulus of elasticity of the exterior walls was the most important parameter to calibrate each model. An iterative process was used, which tested different values of E until the numerical frequencies are considered close enough to the experimental results. The values used for each minaret are given in Table V. The dispersion of E in Table V should not be evaluated as an approximation to the real values. In fact, the modulus of elasticity is accounting for other phenomena that were not considered directly
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

31

Table V. Modulus of elasticity used for each minaret.


Hagia Sophia #2 E (GPa) 9.5 Hagia Sophia #3 11.5 Hagia Sophia #4 4.0 Yeni Cami 4.5 Sleymaniye #3 20.0 Akbyk 2.5 Mihrimah 8.5

1st mode

2nd mode

3rd mode

Torsion mode

Vertical mode

Figure 11. Typical analytical mode shapes.

through the model, such as the degradation of the masonry, the elasticity of the foundations, or the effect of the adjacent main mosque building, among others. In this way, the process of calibration is simplied, as there are no more experimental data to be accounted for. The validity of the following analyses is not an issue, as only a linear analysis is performed and it is directly dependent on the modal analysis, whose results were calibrated. Nevertheless, the values obtained are close to the ones considered in studies in similar structures, as referred in Section 4 [1518], with only a few exceptions. Just as a note we should mention that the difference in modulus of elasticity of Hagia Sophia minarets is due to the fact that minaret #4 is made of brick masonry with wooden stairs, as opposed to the stone masonry in the other two minarets. The only two minarets whose modulus of elasticity does not follow the common trend are Yeni Cami and Sleymaniye #3, which could be related to the interrelation with the mosque. Further studies should be conducted to clarify this situation. 7.2. Modes of vibration Mode shapes are more or less similar in all modeled minarets (Figure 11). They are in pairs (orthogonal in the plane X Y ), and torsion appears only in higher frequencies. Sometimes, we also obtained the vertical mode, also at a higher frequency. The model frequencies obtained for the minarets under study are presented in Table VI, where the comparison with the experimental results is also made. Each model produces pairs of orthogonal modes with almost the same frequency. In fact, the only source of non-axi-symmetry is the stairway, which causes very little effect. Only torsion and vertical modes appear alone. This simple behavior is not exactly the same in the real minarets, due essentially to the connections or links present at the pulpit level. For higher modes, the deformation of the pulpit is not negligible, as well as in torsion and vertical vibration.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

32

C. S. OLIVEIRA ET AL.

Table VI. Comparison between experimental analytical frequencies of vibration, in Hz.


Mode of Vibration First Second Third Fourth Fifth Sixth Seventh Model Exp/mod Model Exp/mod Model Exp/mod Model Exp/mod Model Exp/mod Model Exp/mod Model Exp/mod Hagia Sophia #2 1.18 0.99 1.21 1.05 3.96 0.85 4.01 1.01 8.63 1.12 9.25 1.15 9.42 1.36 Hagia Sophia #3 1.00 1.03 1.00 1.05 4.66 0.93 4.66 0.99 9.49 0.78 Hagia Sophia #4 1.07 0.97 1.07 0.99 5.21 1.07 5.21 1.11 8.07 0.92 Yeni Cami 0.66 0.96 0.66 1.00 2.79 1.07 2.79 1.07 6.15 0.94 6.15 0.98 7.23 1.04 Sleymaniye #3 0.86 0.96 0.86 0.99 4.34 0.85 4.34 0.89 10.45 0.72 Akbyk 1.70 0.99 1.70 1.02 7.69 0.91 7.69 1.02 10.71 0.88 Mihrimah 0.78 1.03 0.78 1.08 4.52 0.94 4.52 1.00 9.58 0.58

In the cases of Hagia Sophia #2 and Yeni Cami, the vertical modes occur at 11.49 and 9.70 Hz, respectively. This mode of vibration was not detected experimentally as the vertical motion was not analyzed. During the stage of calibration of the model, several hypotheses were tested to evaluate the contribution of each structural element. The removal of the stairs has little effect on the modes of vibration dealing with translation. When the pulpit was part of the mosque, the restrain of the pulpit until the height of the building was tested. Nevertheless, the effects are also negligible, as the pulpit has little movement. Especially, in the rst pair of frequencies (mode shape), the pulpit almost does not move, as it was also observed in the experimental records. It is important to point out the existence of some pairs of frequencies which were detected in the experimental campaign, but the model did not detect them. This is happening in the cases of Hagia Sophia #3 and #4, Akbyk and Mihrimah, between the rst and the second pairs of modes in the numerical model. Structural interaction with the building might be responsible for in situ modes, which do not appear in the model. The frequencies measured in the Mihrimah mosque support this hypothesis [25]. The modal mass participation ratio was compared for the Akbyk, Hagia Sophia #2, and Yeni Cami. It was observed that the modal mass participation of the rst mode was 33, 26, and 22%, respectively. The cumulative sum of the modal mass participation ratio reaches 70% for the 5, 6, and 10th modes, for each of the abovementioned minarets. It is possible to conclude that for slender minarets with low frequencies, the higher modes have a greater participation in the dynamic behavior.

8. STRUCTURAL RESPONSE Istanbul is located in an area with intense seismic activity, where numerous destructive earthquakes have occurred due to its proximity to the North Anatolian Fault. The structural response of the minarets to seismic loads was evaluated in two stages. First, a linear response spectrum analysis was conducted, based upon recent studies on regional earthquake hazard. Then, seven ground motions were used as input for a linear time-history analysis. The rst two ground motions are real records from the Kocaeli earthquake, one representing a near-fault source and the other a distant source. The remaining ve time histories are synthetic, i.e. simulated for ve rupture scenarios involving segments of the North Anatolian fault in the Sea of Marmara. These analyses were performed in Hagia Sophia #2, Yeni Cami, and Akbyk, three minarets presenting the most dissimilar geometries. The effects of wind and temperature are not considered, neither soil nor foundation problems. The live load is considered negligible for these analyses.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

33

Table VII. Probabilistic earthquake hazard for a representative site in the Old City of Istanbul for different RPs.
PGA (g) RP= 72 yr RP= 475 yr RP= 2475 yr 0.214 0.438 0.611 SA(0.2 s) (g) 0.460 0.926 1.408 SA(1.0 s) (g) 0.193 0.437 0.687

Figure 12. Response spectrum for a 475-yr RP probabilistic seismic hazard in Istanbul co-plotted with the spectra of Yarmca and Fatih records.

Figure 13. Ground motions records from the 1999 Kocaeli earthquake: (a) Yarmca and (b) Fatih stations (Cosmos, KOERI 1999, [20]).

8.1. Seismic loads The recent destructive events in the Marmara region of Turkey demanded for a proper reassessment of the earthquake hazard in Istanbul. Probabilistic time-independent and time-dependent studies were conducted in this area to dene the peak ground acceleration (PGA) and spectral accelerations (SA) at 0.2 and 1 s periods, for 50, 10, and 2% probabilities of exceedance in 50 years (return periods (RP) of 72, 475, and 2475 yr, respectively) [19]. According to Erdik [33], the seismic motion that can affect the sites of these minarets is summarized in Table VII. The linear response spectrum considered is presented in Figure 12, for an RP of 475 yr and a soil type B according to EuroCode 8 [34, 35]. Two time-history analyses were made with ground motions from the 17th August 1999 Kocaeli earthquake: the Yarmca record (near the fault) with a PGA of 0.322 g; and the Fatih record, obtained in the vicinity of the minaret sites, far from the fault and with a PGA of 0.189 g. Both signals are shown in Figure 13, for the more severe component, as retrieved from the Cosmos database and recorded by the Kandilli Observatory and Earthquake Research Institute (KOERI). In Figure 12, the response spectra of the abovementioned records are plotted together with to the code dened response spectrum for a 475-yr RP.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

34

C. S. OLIVEIRA ET AL.

Additionally, ve simulated ground motion time histories were used as the input. They are broadband hybrid simulations due to ve rupture scenarios to take place on the central Marmara (two ruptures of M = 7.4) and northern boundary segments of the North Anatolian Fault (three ruptures of M = 7.0) in the Marmara Sea. The simulations are for a representative site in the historical peninsula of Istanbul in close vicinity of the minarets. The details of the simulation approach and of physical properties related to tectonic and geological setting can be found in [29]. The largest simulated PGA is 0.39 g. 8.2. Results We subjected the three models to the seismic loads described in the previous section. The actions were applied only in one direction, as the minarets have radial symmetry. Both the response spectrum analysis and the time-history analyses used a linear modal superposition method. When a linear elastic model is being analyzed both the linear modal superposition method and the direct integration method give out approximately the same results. Nevertheless, the rst one is faster as less computation is involved. For simplication, modal damping of 5% (critical) was considered for all modes, a value much higher than the one determined through the dynamic characterization tests, but which could represent the inuence of amplitude input motion. Stresses and horizontal displacements for vertical loads and for earthquake loads are shown in Table VIII. The total vertical load at each section requires the integration of all values around the circle. The average values presented consider the integration in for the interval [45 , +45 ], centered at the point of peak value. In Figure 14, the maximum tensile stresses are shown for response spectrum analysis and time-history analyses of Yarmca and Fatih records. This structural analysis allowed us to make a rst evaluation of the effect of an earthquake on a minaret according to its dynamic properties. The preliminary conclusion is that the main focus of concentration of stresses is next to the transition section, where a reduction in section area and the deviation in the load path occur. Near the balconies, an increase in forces is also observed, due to the increase in mass. Hagia Sophia #2 is the minaret with the highest mass. Its frequency is higher than the Yeni Cami, due to its much higher stiffness. This has a direct effect on the results of response spectrum analysis. The Hagia Sophia and the Akbyk are subjected to a pseudo-acceleration close to the maximum and considerably higher than Yeni Cami (Figure 14). As a note for comparison, Tankut and Pnarbas [35] when monitoring of a leaning minaret 35 m high found that the top of minaret moves during the year due to temperature changes around 30 cm without causing a critical situation (4 MPa compression stresses). Doubling these values, according to these authors, the collapse might be imminent. In relation to the ground motion analysis, it is important to mention that the real records considered present different properties, as one is near the fault (Yarmca) and the other is distant (Fatih). This way, the stresses in Hagia Sophia are higher than the ones in Yeni Cami for the Fatih ground motion and the opposite for the Yarmca. Nevertheless, the Yarmca ground motion presents higher stresses in both cases, as it has a higher PGA. The Akbyk minaret presents contrasting results, as the Fatih ground motion causes higher stresses than Yarmca. Although, for Akbyk the results are similar for all the analysis considered, the response spectrum analysis leads always to higher values in the other minarets. Another important point that needs to be mentioned is that the compression level is at least 2.5 times lower than the reference value for stones (26 MPa), but the tensile stresses are much higher than the reference value (2.6 MPa). The stresses in the stairs are considerably smaller than in the exterior walls, almost 10 times lower. The time-history analysis results of simulated accelerations are essentially parallel to those of real records. The Yeni Cami minaret experiences similar top displacements when analyzed under simulated accelerations and under Yarmca record. The top displacements of Hagia Sophia minaret are in the same order under response spectrum analysis and when analyzed using the Yarmca record. The Akbyk minaret on the other hand experiences largest displacements in the response spectrum analysis. In terms of maximum tensile stresses, the Yeni Cami minaret experiences
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

35

Table VIII. Structural analysis results for the four load cases (dead-load and three input ground motions) for three minaret models (shaded is the equivalent seismic coefcient).
Maximum Maximum Average Average Top Vertical Horizontal Equivalent compression tensile compression tensile displacement reaction reaction seismic vertical vertical vertical vertical (m) (kN) (kN) coefcient (kPa) (kPa) (kPa) (kPa) Dead Load Response spectrum Ground motion 1 (Fatih) Ground motion 2 (Yarimca) Hagia Dead Sophia Load #2 Response spectrum Ground motion 1 (Fatih) Ground motion 2 (Yarimca) Akbyk Dead Load Response spectrum Ground motion 1 (Fatih) Ground motion 2 (Yarimca) Yeni Cami NA 0.398 0.10 0.68 NA 0.254 0.14 0.30 NA 0.131 0.13 0.11 9916.9 9919.2 9917.3 9917.5 45 305.0 NA 2548.8 1154.4 2373.9 NA NA 0.26 0.12 0.24 NA 0.34 0.19 0.25 NA 0.41 0.33 0.36 705 6417 2373 10 595 764 9084 5241 9991 321 3755 4019 3048 NA 5011 1074 8811 NA 7701 3910 8607 NA 3386 3465 2823 705 5579 2121 9138 652 7347 4246 8009 321 3362 3483 2444 NA 4172 816 7403 NA 6103 3133 6763 NA 2880 2933 2368

45 594.0 15 389.1 45 340.2 8534.7

45 348.6 11 336.4 2048.0 2048.0 2048.0 2048.0 NA 848.0 666.0 737.0

similar levels of stresses under simulated accelerations and Yarimca record accelerations. For Hagia Sophia minaret as well simulated and Yarimca accelerations are more critical. For the Akbyk minaret similar levels of stresses are obtained from response spectrum analysis and from simulated accelerations. The Yeni Cami and Hagia Minarets are more sensitive to near-eld ground motions. They experience considerable top displacements and compressive and tensile stresses. The Akbyk minaret presents a higher response for ground motion in the far eld. It appears whether a destructive future earthquake will take near the city or far away from it will lead to different damage patterns in the minarets in Istanbul. The observation made for the Yarmca and Fatih records regarding the stresses that they cause in the minarets depending on whether they are recorded in the near eld or far eld, holds for the simulated records as well.

9. CONCLUSIONS AND FINAL CONSIDERATIONS An extensive dynamic characterization campaign was performed in 11 minarets in Istanbul, which allowed for the determination of frequencies, modes of vibration and damping. The nite element modeling and analysis of seven of these minarets were performed. Linear dynamic structural analyses were conducted to access their earthquake risk level.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

36

C. S. OLIVEIRA ET AL.

Figure 14. Maximum tension stresses in the exterior wall. The presented values are the extreme values of the colored scale, in kPa.

Ambient vibration tests were conducted in all minarets recording the natural excitation at three levels. Dynamic characteristics were determined through the FDD, EFDD, and SSI techniques, which permitted the identication of the rst seven frequencies. Using all the data gathered, an empirical formula was developed to estimate the rst frequency of vibration. This formula considers that the frequency is proportional to the square root of the second moment of area and almost inversely proportional to the square root of the height of the minaret. The formula leads to an upper limit with an error of 30%. Performing in situ testing in other minarets will contribute to validate the empirical formula, probably reducing the errors associated with it by considering the modulus of elasticity as well. This formula has the importance to offer an expedite tool for the structural engineer to have a rst estimate of the minaret response to a given earthquake input.
Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

37

By calibrating the modulus of elasticity of the exterior walls, seven nite element models of the minarets were created, obtaining modal results similar to the experimental ones. Translation, torsion, and vertical modes of vibration were obtained. Effects of the connection to adjacent buildings (mosque) or exibility of the foundations should be accounted for in further studies. After the models calibration, the seismic behavior three minarets, that presented the most distinguishing geometry, was investigated. For this purpose, linear analyses were conducted. A response spectrum analysis, based upon the most recent studies in earthquake hazard levels, was conducted. Additionally, two ground motion records, one representing a near-fault site and the other a distant site and ve simulated accelerations were used in the time-history analysis. Owing to each minarets distinctive dynamic characteristics, the seismic response is different from case to case. Nevertheless, the compression stresses obtained in all studied cases were below the reference maximum value. The tensile stresses however were considerable, as the minarets do not have high dead-loads to balance the tensile forces created by earthquake loads. In a real seismic event, such high values of tensile stresses would probably not be achieved, as cracking would lead to some energy dissipation and eventually it would cause the minaret to collapse. This can only be further investigated if nonlinear models are used. But with the linear modeling we can see that the main areas of stress concentration are at the end of the transition phase, where a reduction in wall thickness occurs, and at the balconies, due to the increase in mass. Several retrot techniques can be considered to correct the observed deciencies. One possibility is to increase the resistance of the masonry through the connement with exterior steel rings, as in chimneys. Owing to the low level of compression forces, a post-tension rod could be placed in the central core, increasing its stability. Another solution to increase the minaret resistance would be to use a helicoidal-spiral (wire-mesh or carbon-bre) in the inner side of the outer walls. One last idea would be to place tuned mass dampers inside the spire. For each case the effects of retrot must be carefully evaluated, in terms of stresses (or bending moments, shear, compression, and tension) and displacements developed. Further studies would include analysis under site-specic ground motion and consequently a nonlinear modeling of the mechanical properties of the masonry components. Also, the effect of soilstructure interaction should be analyzed. Permanent instrumentation is desirable in at least one minaret consisting of strong motion accelerometers and displacement transducers [36]. The construction materials should be characterized by ultrasonic testing and other tests, to determine their mechanical, physical, and chemical properties. Also, the use of shaking table to calibrate the numerical model could be of great interest. This way the performance of proposed retrot methods could be theoretically assessed and experimentally veried.

ACKNOWLEDGEMENTS

CSO acknowledges the partial nancial support given by Fundao para a Cincia e a Tecnologia, Portugal (FCT) through its Pluri-Annual Programme. Ana Mateus made a few initial tests on the numerical modeling of the minarets. This work was made under the State Planning Organization through the Project 2003 K120250 Development Emergency Retrotting Techniques for Historical Structures. To Mr Naz Kafadar, Ahmet Korkmaz, Gkhan Kesti and Orhan Ersah for handling the instrumentation and pre-processing the instrumental data. To GeneralDirectorate of Foundations, Mr Yucel Sezek, Istanbul RegionalDirectorate, Istanbul Directorate for Monuments and Reconstruction of Turkish Ministry of Tourism and Culture, for allowing the eld work in the minarets. To Prof. M. Erdik for all support in this research.

REFERENCES 1. Do angn A, Akar R, Sezen H, Livaoglu R. Investigation of dynamic response of masonry minaret structures. g Bulletin of Earthquake Engineering 2008; 6(3):505517. 2. Motosaka M, Somer A. Ground motion directionality inferred from a survey of minaret damage during the 1999 Kocaeli and Dzce Turkey earthquakes. Journal of Seismology 2002; 6:419430. Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

38

C. S. OLIVEIRA ET AL.

3. Parsons T. Recalculated probability of M 7 earthquakes beneath the Sea of Marmara, Turkey. Journal of Geophysical Research 2004; (109):B05304. DOI: 10.1029/2003JB002667. 4. Do angn A, Livao lu R, Acar R. A study on seismic behavior of minarets considering soilstructure interaction. g g Proceedings of the International Earthquake Symposium, Kocaeli, 2224 October 2007. 5. Do angn A, Sezen H, Tuluk O, Livaoglu R, Acar R. Traditional Turkish masonry monumental structures and g their earthquake response. International Journal of Architectural Heritage: Conservation, Analysis and Restoration 2007; 1(3):251271. 6. Firat YG. A study of the structural response of minarets in the 1999 Anatolian earthquakes. M.Sc. Thesis, Purdue University, 2001. 7. Uluengin F, Uluengin B, Uluengin MB. Classical Structural Details in Ottoman Monument Architecture (in Turkish). Yem Publication: Istanbul, Turkey, 2001. 8. Ambraseys N, Finkel C. The Marmara Sea Earthquake of 1509. Terra Nova 1990; 2(2):167174. 9. Ambraseys N, Finkel C. Long Term Seismicity of Istanbul and of the Marmara Sea Region. Terra Nova 1991; 3(5):527539. 10. Mller-Wiener W. Bildlexikon Zur Topographie Istanbuls (in German). Deutsches Archaologisches Institute: Verlag Ernst Wasmuth Tbingen, 1977. 11. Sezen H, Firat GY, Sosen MA. Investigation on the performance of monumental structures during the 1999 Kocaeli and Duzce Earthquakes. The 5th National Conference on Earthquake Engineering, Istanbul, Turkey, 2630 May 2003. 12. El-Attar AG, Saleh AM, Osman A. Seismic response of a historical Marnluk style minaret. Earthquake Resistant Engineering Structures 2001; 57. DOI: 10.2495/ERES010701. 13. Oliveira CS. Seismic vulnerability of historical constructions: a contribution. Bulletin of Earthquake Engineering 2003; 1(1):3782. 14. Tassios TP. Seismic Engineering of Monuments. Earthquake Engineering in Europe, Chapter 1. Springer: Berlin, 2010; 142. 15. Selahiye A. Determination of dynamic properties of Sleymaniye Mosque. M.Sc. Thesis, Bo azii University, g 1994. 16. Kaya SM, Yuzugullu O, Erdik M, Aydinoglu N. Earthquake performance of Sleymaniye Mosque. The 12th European Conference on Earthquake Engineering, Paper number 851, London, 2002. 17. Kirlangi A. Re-evaluation of earthquake performance and strengthening alternatives of Hagia Sophia. M.Sc. Thesis, Bo azii University, 2004. g 18. Durukal E. A study on structural identication and seismic vulnerability of Aya Sophia. M.Sc. Thesis, Bo azii g University, 1992. 19. Erdik M, Demicioglu M, Sesetyan K, Durukal E, Siyahi B. Earthquake hazard in Marmara Region, Turkey. Soil Dynamics and Earthquake Engineering 2004; 24:605631. 20. Zulkar C. Personal communication, 2009. 21. NEHRP. Recommended provisions for new buildings and other structures, FEMA-450, prepared by the Building Seismic Safety Council for the Federal Emergency Management Agency: Washington, DC, 2003. 22. Bayraktar A, Altunisik AC, Sevim B, Trker T, Akkse M, oskun N. Modal analysis, experimental validation, and calibration of a historical masonry minaret. Journal of Testing and Validation 2008; 36(6). DOI: 10.1520/JTE101677. 23. Sayin MK. Structural identication of the Edirnekapi Mihrimah Sultan Mosque through dynamic response studies. M.Sc. Thesis, Bo azii University, Istanbul, Turkey, 1999. g 24. Selahiye A. A study on identication of natural vibration frequencies of the Suleymaniye Mosque. M.Sc. Thesis, Bo azici University, Istanbul, Turkey, 1994. g 25. Stengel D. System identication for 4 types of structures in Istanbul by frequency domain decomposition and stochastic subspace methods. Diploma Thesis, co-submitted to Karslruhe University and Bo azii University, g 2009. 26. Oliveira CS, Navarro M. Fundamental periods of vibration of RC buildings in Portugal from in situ experimental and numerical techniques. Bulletin of Earthquake Engineering 2010; 8:609642. 27. Gallipoli MR, Mucciarelli M, Vona M. Empirical estimate of fundamental frequencies and damping for Italian buildings. Earthquake Engineering and Structural Dynamics 2009; 38(8):973988. 28. Hans S, Boutin C, Ibraim E, Roussillon P. In situ experiments and seismic analysis of existing buildingspart I: experimental investigations. Earthquake Engineering and Structural Dynamics 2005; 34(12):15131529. 29. ARTeMIS Extractor . Ambient response testing and model identication software. Structural vibration solutions, A/S, 2009. 30. Clough R, Penzien J. Dynamics of Structures. McGraw-Hill: U.S.A., 1975. 31. SDR. Microtremor measurement of Fatih Mosque. System and Data Research Co. (SDR20060124), Japan, 2006. 32. SAP2000 . Integrated Software for Structural Analysis and Design. Computers and Structures, Inc: CA, 2008. Copyright 2011 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

MINARET BEHAVIOR UNDER EARTHQUAKE LOADING

39

33. Erdik M, Durukal E, Sesetyan K, Demircio lu MB. Assessment of the earthquake resistant design basis ground g motion for Hali Metro Bridge. Report, Bo azii University, Istanbul, Turkey, 2006. g 34. Eurocode 8, EN 1998-1. Design of Structures for Earthquake Resistance Part1: general rules, seismic action and rules for building. European Standard, European Committee for Standardization. CEN: Brussels, Belgium, 2004. 35. Eurocode 8, EN 1998-6. Design of structures for earthquake resistancepart 6: Towers, masts and chimneys. European Standard, European Committee for Standardization. CEN: Brussels, Belgium, 2005. 36. Tankut AT, Pnarbas S. Monitoring of a leaning minaret. WCCE-ECCE-TCCE Joint Conference, Istanbul, 2224 June 2009.

Copyright

2011 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2012; 41:1939 DOI: 10.1002/eqe

Vous aimerez peut-être aussi