Structural Ana/ysis of Historica/ Constructions - Modena, Loureno & Roca (eds)

2005 Taylor & Francis Group, London, ISBN 04 1536 379 9

Earthquake performance of Suleymaniye Mosque

S.M. Kaya, O. Yuzugullu, M. Erdik & N. Aydinoglu

Bogazici University Kandilli Observatory and Earlhquake Research InslitUle

ABSTRACT: In a previous research project on the identification ofthe structural configuration ofSuleymaniye
Mosque, earthquake performance, natural frequencies and mo de shapes were determined by both ambient vibra-
tion tests and finite element analysis. In the present study the three dimensional finite element model prepared
during this research was refined and improved as another step ofthe ongoing research activities for the edifice
by adding the small domes to the model and increasing the number of elements to achieve maximum precission
in the analysis. Non-destructive material tests were carried out in order to determine the material properties. The
etfects of different materiais and different boundary conditions were also combined . A satisfactory correlation
is observed between the previous and present analytical results.

INTRODUCTION The structural elements of Suleymaniye are com-

posed of domes, transition elements, arches, counter
Suleymaniye Mosque was built between 1549- weight towers, piers, walls and butresses and founda-
1557 by the great Turkish Architect "Mimar Sinan" tions. The Suleymaniye Mosque is the main building
and named after the legendary Ottoman Emperor ofthe Suleymaniye complex. It's plan dimensions are
Suleyman the Magnificient. 61 m in the south direction and 73 m in the perpen-
Considered to be the masterpiece of the Ottoman dicular direction. The mosque has an axis of symme-
architecture, Suleymaniye Mosque has successfuly try in the north-south directions. The main dome is
sustained several major earthquakes without any 26.2 m in diameter and height reaches 49.5 m from the
severe damage. ground.
It is a fact that research studies carried out up The circular base dome is transferred to the square
today towards the determination ofthe dynamic char- geometry via the four decorated pendantives. The main
acteristics of this masterpiece of Ottoman Turkish arches are connected to the main piers at an elevation
Engineering is very limited. Within the framework of of 32 m from the ground.
research activities being carried out at the Earthquake Semi-domes existing in the north and south parts
Engineering Department ofKandilli Observatory and of the structure are connected to the main arches and
Earthquake Research Institute for the historic edifices rest upon the two exterior buttress piers. On either
which was initiated with Hagia Sophia, the present side ofthe semi-domes, an exedra semi-dome enlarge
study on historic Suleymaniye Mosque is aimed to the internai volume of the mosque.There are also
explore more on the earthquake performance and five smaller domes both on the west and east of the
dynamic characteristics ofthe monumento structure.
The huge dimensions of the main four piers are
hidden by means of the detailing provided in the cir-
2 STRUCTURAL SYSTEM or cumference of the piers. These piers are connected to
SULEYMANIYE the exterior buttresses via the double arches, at an ele-
vation of 10m from the ground. The counter weight
Early Ottoman domed buildings (14th and 15th cen- towers resting on the top of the main piers provide
turies) were based either on the concept of a single lateral supports.
dome of medi um size covering the whole inner space The west and east walls, laying under the main
or, on the series of small domes one neighbouring the arches, rest on top of the small arching frame sys-
other at the same leveI. In both solutions, thrusts and tems which spring to the main piers and internai
seismic actions would thus be laterally transmitted to columns at an elevation of 10m from the ground
the massive exterior walls or piers. leveI.

477
3 CONSTRUCTION MATERIALS ANO ultrasonic testing method and Schmidt Hammer test-
THEIR MATERIAL PROP ERTIES ing method were used.

3. 1.1 Stone and brick 3.2 Schmidt Hamm er lests

The structure is mainly constructed by stone and brick.
The major part ofit such as ma in piers, arches, internai Various structural elements were tested in Suley-
secondary arches, buttress piers, and walls are made maniye including main piers, marble columns, infill
of stone. The domes are made of brick and covered walls, floors, exterior stone walls. Among them the
by lead. most reliable measurements are those taken from
Among the various types of stones Architect Sinan exposed (uncovered) surfaces like stones of the main
most widely used is "Kufeki" stone the test results of piers. Therefore, only the measurements belong to the
which is reported in (1). "Kufeki" stone was used as ma in piers shall be given in this study.
structural material on piers, walls, arches or decora-
tive material for covering walls and slabs. The art of 3.3 Ultrasonic pu/se ve/o city measuremenl
using this material can be observed on severa I his- Four series of ultrasonic tests were conducted on the
torical edifices built by Sinan. The pore ratio of this main piers of the mosque. The selected test locations
stone is rather low. The test results indicate that the had exposed surface (without plaster). The test results
mechanical characteristics like compressive, tensile are summarized in Table 2 below.
and shear strenghts of the stone increase as a func-
tion of decrease in pore pressure, and provide better 3.4 Mortar
resistance to externai effects.
Previous structural studies to determine the earthquake
3. 1.2 Non-destructive tests worthiness of monumental buildings in lstanbu l like
In order to determine the parameters suitable for Hagia Sophia (2) have shown that the monllments'
the definition of the static behaviour of the edifice, static and dinamic behaviour depend strongly on the
diagnostic investigations are conducted using only mechanical and chemical properties ofthe mortar and
simple and rapid non-destructive testing techniques. bricks used in their masonary.
Tt must be emphasised, however, that the use of only Estimated Elastic Moduli from in-situ ultrasonic
a preliminary evaluation ofthe mechanical character- tests at variolls brick and mortar locations in Hagia
istics of the masonary by defining their mechan ical Sophia (2):
"quality indices". Even though these tests are unable Brick: Eb = 3.10 Gpa; Mortar: Em = 0.66 Gpa;
to supply the quantitative mechanical parameters, their Composite: Ebm = 1.83 Gpa.
use is very important as they provide information on The modulus of elasticity of the composite mate-
the homogeneity of the material and on the presence rial Ec (brick + mortar) was computed using Ec =
of the structural anomalies. 2Eb Em/(Eb + Em) where Eb = Modulus of elasticity
The only reliable means for the determination of the
parameters that influence the mechanical behaviour of Table I. Compressive strength va lues obtained by Schmidt
the material is that of utilising a slightly destructive Hammertest.
method that require some interventions (coring and
curing). These actions must be studied in such a way Average lndicated
that the disturbance to the member is temporary. At the measured from Hammer
end ofthe tests, there should be no visible signs ofthe strength curve Number of
work remain on the structure. Element (N/mm 2 ) (N/mm 2 ) measurements
For a historic significant monument Iike Suley-
Main pier I 32 226 36
maniye, slightly destructive testing techniques could
Main pier 2 35 310 36
not be used since even slight alteration in any element Main pier 3 40 350 36
is not permitted. Similar non-destructive testing meth- Main pier 4 39 339 36
ods were utilised on the exposed surfaces of Hagia
Sophia with the above mentioned considerations.
Characteristics which may be specified and are Table 2. Average modulus of elasticity.
capable of assesment by non-destructive tech-
niques are: Modulus of
Test location elasticity (N/m 2 )
Structural integrity
Durability Main pier I 2,35
Appearance and tolerance. Main pier 2 2,77
Main pier 3 2,71
In Suleymaniye, for the determination of the Main pier 4 2,30
dynamic moduli and the average compressive strength,

478
 Table 3. Average modulus of elastieity.

Modulus of
elastieity
Time (see) Veloeity (m/ see) (N/m 2 ). 109
Test Length
loeation (em) Il II II

Pier 1 70.0 589.2 1188.0 2.54

Pier I 77.5 632.8 1224.7 2.70
Pier I 93.0 742.2 1253.0 2.83
Pier 1 113.0 838.5 1347.6 3.27
Pier I 120.0 1209.0 992.6 1.77
Pier I 150.0 2038.4 735.9 9.75
Pier 2 70.0 556.6 586.6 1257.6 1193.3 2.85 2.56
Pier 2 100.0 765.0 834.6 1307.2 1198.2 3.08 2.58
Pier 2 130.0 984.4 1121.1 1320.6 11 59.6 3. 14 2.42
Pier 3 70.0 540.4 613.9 1295.3 1140.3 3.02 2.34
Pier 3 100.0 806.0 830.0 1240.7 1204.8 2.77 2.61
Pier 3 130.0 1040.0 1056.0 1250.0 1231.1 2.81 2.73
Pier 4 70.0 509.0 699.0 1375.3 1001.4 3.40 1.81
Pier 4 100.0 1161.0 977.2 1112.4 1023.3 2.23 1.88
Pier 4 130.0 1384.0 1333.0 1119.7 1147.4 2.26 2.37

Table 4. Average modulus of elastieity. as possible which increase the run time of the com-
puter. However, run time was considerably reduced
Modulu s of by the aid of a computer which had a large memory
Material elastieity Poisso n capacity and fast processors.
properties (N/m2 ) . 109 ratio Mass Thiekness

Arehes 3.5 0.20 4.1.2 Struetural mode! properties

Piers 10.0 0.18 The total number of elements and nodes in the model
Domes 3.5 0.18 0.50 were 3989 and 6980 respectively.
Pendantives 3.5 0.20 2.0 In a previous study on Suleymaniye Mosque (3)
Semi-domes 3.5 0.18 0.70 a three dimensional finite element model was con-
structed. During this study, the initial intention was to
use SHELL elements for the domes, SOUO elements
fort the piers, BEAM elements for the lateral dome
buttresse. This selection of elements created problems
ofthe bricks; Em = Modulus of elasticity ofthe morta r.
at the dome leveI such as:
Therefore, Ec = 50000 kg/cm 2 .
The beam elements used at side of the dome win-
dows increased the actual window opening width.
4 STRUCTURAL ANALYSIS BY FfN1TE The use of beam type element for the lateral dome
ELEMENT MOOEL bracing would increase total stress on the connection
node on the arch.
4.1.1 Numerieal modeling The problem of where to connect the dome base
The numerical modelling used for Suleymaniye arose.
Mosque is created by LUSAS package programo
LUSAS (4) is a general purpose finite element anal- Considering the above mentioned problems, and to
ysis system which incorporates facilities for: linear maintain the homogeneity in the element types; lat-
and non-linear; creep, natural frequency, buckling, eral dome bracings and the short columns between the
spectral response, harmonic response, fourier analysis; dome windows were simulated by 3D SOU O elements
steady field and transient field analysis and coupled which is actually the case in the real structure.
thennomechanical analysis.
The finite element model in this study was obtained 4.1 .3 Struetura! idealization
through severa I runs. In order to assure the precission The structural finite element model which was used in
ofthe resuJts, the number of elements were kept as high the previous study (3) 011 Suleymaniye Mosque was

479
Figure I. 3-Dimensional view of Suleymaniye Mosque.

x
J/

Figure 2. Modal analysis mode 1-2-3-4.

480
improved by utilising the following modifications: Foundation leveI was lovered 2.5 m below the
ground leveI.
A detailed study was carried out to determine an
Four different boundary conditions were analysed
optimum number of structural elements. For this
for different combinations of material properties.
purpose the total number of elements were raised to
3989 and the corresponding total number of nodes The frequencies corresponding to the first five
to 6980. modes are given in table 5, (Fig. 2)
Small domes were added to the model which were
previouslyexcluded. 4.2 Analysis
The results ofthe various dynamic runs under differ-
Table 5. Results of modal analysis. ent combinations ofmaterial properties and boundary
conditions are summarized in Table 6.
Modes Frequency (Hz) Four different boundary conditions were analysed
for four different combinations ofmaterial properties.
Mode I (Lateral mode in 3,244 Severa I dynamic runs were carried out and finally
the north-south direction) a model with nearest frequency values to those of
Mode 2 (Lateral mo de in 3,420 ambient vibration tests was selected as the improved
the east-west direction) model which was used for the spectral response anal-
Mode 3 (Torsional mode) 4,305 ysis under a scenario earthquake for istanbul. Table 7
Mode 4 (Diagonal Torsional mode) 4,734
gives the material properties used in the analysis.
Mode 5 (Lateral squeezing in 4,745
east-west and north-south direction) Table 8 below can be used to compare the results
obtained from ambient vibration tests, earthquake

Table 6. Table of dynamic runs.

Natural frequencies
Model lnpul
na me Descriplion FI F2 F3 F4 F5 no

Supl AI Fix in XYZ aI - 5.3 m 3,07 3,23 3,76 4,81 4,93 I

2,84 2,99 3,65 4, 17 4, 18 2
2,89 3,03 3,03 4,28 4,30 3
Sup2 A2 Fix in XY aI - 2.8 m 3,24 3,41 4,02 5,01 5, 12 I
2,96 3,12 3,89 4,32 4,33 2
3,02 3,17 3,91 4,44 4,44 3
B2 Spring in XY aI - 2.8 m A I 3,08 3,29 3,82 4,85 4,91 I
2,85 3,03 3,70 4,20 4,22 2
2,90 3,08 3,72 4,31 4,34 3
Sup3 A3 Fix in XYZ aI - 2.8 m AI 3,52 3,75 4,45 5,35 5,44 I
3, 15 3,34 4,26 4,56 4,60 2
3,24 3,41 4,30 4,7 1 4,73 3
Ambienl tests 3,38 3,44 4,26 4,71 5,85
Sup4 A4 Fix in XYat - 2.8mAl 3, 14 3,33 3,89 4,90 5,03 I
2,89 3,06 3,78 4,25 4,26 2
Fix in XY at - 0.3 m 2,94 3, 11 3,80 4,36 4,38 3
B4 Spring in XY at - 2.8 m A I 3, 14 3,33 3.89 4,90 5,03 I
2,89 3,06 3,78 4,25 4,26 2
Spring in XY aI - 3 m 2,94 3, 11 3,80 4,36 4,38 3
A5 Fix in XYZ aI - 2.8 m AI 3,72 3,74 4,47 5,36 5,45 I
3,16 3,35 4,28 4,56 4,6 2
Fix in XY at - 0,3 m 3,37 3,56 4,59 4,88 4,38 3
B5 Fix in XYZ aI - 28 m A I 3,.53 3,74 4,47 5,36 5,45 I
3, 16 3,35 4,28 4,56 4,61 2
Fix in XY at - 0.3m 3,24 3,56 4,59 4,88 4,93 3
A6 Fix in XYZ at - 28 m 4,04 4,31 5,34 6,04 6,08 1
3,46 3,68 4,94 4,94 5,06 2
3,58 3,79 5,02 5,14 5,23 3

481
Table 7. Material properties used in the analysis. were used in the model and the resu lting frequencies
were compared with those obtained by ambient vibra-
Modulus of e lasticity values (N /m 2 ) . 109 tion tests. By using this comparison, different material
properties were applied to the model in order to
Input no Piers (Stone) Dome (Brick) Arches (Brick) obtain model ofthe best conformity with the ambient
vibration test results.
I 10 5 8,5
2 10 3 3
Four different boundary conditions were ana lysed
3 3,5 3 3 for a number of different combinations of material
4 10 3,5 3,5 properties.
Several dynamic runs were carried out and finally
a model with the nearest frequency values to those of
ambient vibration tests was selected.
Table 8. Comparison of model frequencies.
The improved model showed an acceptable compat-
Earthquake ibility with the test results, the earthquake results of
Ambient results Previous Improved 28.05 .1994 and the results ofthe previous model (3).
test results 28/05/ 1994 model model Unfortunately , precise documents ofthe works are
Mode (cps) (cps) (cps) (cps) not found in the archives. The chronicle usually reports
the payment of the works, but not their location and
1 3,38 3,38 3,26 3,24 technical details which would be very useful for the
2 3,44 3,42 3,65 3,42 analysis. lt is essential to obtain as-built drawings and
3 4,26 4,3 4,58 4,3 full description of the works for the future studies on
4 4,71 5,21 4,73
the edifice.
5 5,85 5,35 4,74

* Could not be computed.

REFERENCES

records and the finite element models . lt is observed Arioglu , E. & Arioglu, N. 1999. Mimar Sinan'in Ta~iyici
that the frequencies of the improved model and fre- olarak Kullandigi Kfeki Ta~inin Mhendislik Gizemi ,
quencies observed during 1994 earthquake are very Mimar Sinan Danemi Yapi Etkinlikler Seminar, istanbul
close to each other. Durukal, E. & Yzgll, O. 1998 . Non-destructive Test-
ing Techn iques of Structural Materiais in Historical
Structures, /NCOMARECH-RAPHAEL 97/E/412 Com-
patib/e Materiais for the Proteclion ofElIropean Cultural
5 CONCLUSrON Heritage Pact 55, 1998, Athens
Selahiye, A. Determination of Dynamic Properties of
The three-dimensional finite element model previ- Sleymaniye Mosque. M.Sc. Thesis, Bogazici Uni-
ously prepared was improved by adding the small versity, Kandilli Earthqllake Observatory and Earth-
domes, lowering the foundat ion levei 2.5 m below the quake Research Im'litute, Department of Earthqllake
previous levei and increasing the number of elements Engineering
to achieve maximum precision. These improvements Papayanni, I. 1997. Repair Mortars Suitable for lnterventions
were realised after the preliminary studies on the ofOttoman Monuments, Studies inAncient Strllcture, Pro-
structural system and the material properties. ceedings of the Interna/ional Conference of Ear/hqllake
July 14- 18, 1997, /stanbul
In order to determine the material parameters suit- Anadol , K. & Arioglu, E. 1973 . Earthquake Resistance of
able for the definition of the edifice, diagnostic inves- Suleymaniye Mosque, Fiflh World Conference on Earth-
tigations were conducted using non-destructive testing quake Engineering, 1973, Rome
techn iques (Schmidt Hammer tests and ultrasonic LUSAS Finite Element System Theory Manuels, FEA Ltd. ,
pulse velocity measurements). The results obtained United Kingdom

482