Sie sind auf Seite 1von 10

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

net/publication/251230815

Structural behaviour of Hagia Sophia under dynamic loads

Conference Paper · September 2013

CITATIONS READS

4 5,842

5 authors, including:

Umut Almac Karl Schweizerhof


Istanbul Technical University Karlsruhe Institute of Technology
20 PUBLICATIONS   12 CITATIONS    554 PUBLICATIONS   3,130 CITATIONS   

SEE PROFILE SEE PROFILE

Gunther Blankenhorn
Livermore Software Technology Corporation (LSTC)
44 PUBLICATIONS   180 CITATIONS   

SEE PROFILE

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

Mechanical Models for Cables and Wire Ropes - Mechanics of Ropes, Knots and Rope-Nets View project

Plagiarism in Science View project

All content following this page was uploaded by Karl Schweizerhof on 01 June 2014.

The user has requested enhancement of the downloaded file.


Vienna Congress on Recent Advances in
Earthquake Engineering and Structural Dynamics 2013 (VEESD 2013)
C. Adam, R. Heuer, W. Lenhardt & C. Schranz (eds)
28-30 August 2013, Vienna, Austria
Paper No. 475

Structural behaviour of Hagia Sophia under dynamic loads


U. Almac1, K. Schweizerhof 2, G. Blankenhorn3, C. Duppel4, F. Wenzel5
1
Faculty of Architecture, Istanbul Technical University, Istanbul, Turkey
2
Institute for Mechanics, Karlsruhe Institute of Technology, Karlsruhe, Germany
3
Livermore Software Technology Corp., Livermore, CA, USA
4
B+G Ingenieure Bollinger und Grohmann GmbH, Frankfurt am Main, Germany
5
Karlsruhe Institute of Technology, Karlsruhe, Germany

Abstract. Hagia Sophia is one of the most well known monuments in the history of architecture. It is a very
innovative structure within its construction period both from architectural and engineering standpoints. The
construction of the masonry structure was completed in the 6th century. Besides the reconstruction of the dome,
which collapsed only 20 years after completion of the monument, the structure survived more than 1500 years
with only secondary damages and partial collapses. Due to this outstanding performance against numerous
earthquakes in its long history, it is important to provide engineering information in order to understand the
structure itself as well as its vulnerabilities, which might be regarded as initial steps to develop efficient
proposals if required.
The paper presents the preliminary results of the research carried out with the support of HPS-Europa2
Transnational Access Programme. In order to accomplish the engineering tasks, FE modelling with high
performance computing was used during the research. Large and detailed FE models were received from the
Institute of Mechanics at the KIT Karlsruhe to represent the structural system. Within the framework of this
paper, a series of analyses concerning the primary and secondary structure will be presented.

Keywords: Hagia Sophia; Earthquake behaviour; Numerical analysis

1 INTRODUCTION

The church of Hagia Sophia in Constantinople was built in 6th century during the reign of Justinian.
The construction lasted in a fairly short time in comparison with its huge scale, only five years and ten
months between 532 and 537 (Procopius 2006). The architects Anthemius of Tralles and Isidorus of
Miletus established a unique structure, by means of combining architectural and structural innovative
arrangements, which astonishes in the very first sight even today(ibid).
However, due to the high seismicity of the Marmara region, Hagia Sophia has suffered from numerous
earthquakes during its long history. In 558, only 20 years after completion of the church, the central
dome, including its supporting elements, partially collapsed due to an earthquake in 557. Nephew of
Isidorus was assigned for the repair works and a new dome with an altered form was completed in 563
(Müller-Wiener 2007). However, this second dome of young Isidorus was also damaged and partially
rebuilt on two occasions (Van Nice 1963). First, the western part of central dome with its adjacent
parts including the western arch and semidome collapsed in 989. Second in 1346, the eastern part of
central dome and the adjacent main arch and semidome partially collapsed (Mainstone 2006).
During Byzantine and Ottoman eras, there had been some later additions to the original texture, which
altered the exterior appearance of the church. Those additions included structural elements such as
buttresses in order to avoid further damages from the earthquakes.
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 2

2 STRUCTURAL SYSTEM OF HAGIA SOPHIA

The layout of the monument is a combination of longitudinal basilica (approx. 31 m. x 80 m.) and a
central plan with a dome (Figure 1). The central dome rests on a circular base, which is formed by four
main arches and pendentives. It rises about 56 m. from ground level (Van Nice 1963) and has a
diameter of 31 m. The massive piers in the central space support the main arches and pendentives in
order to transfer loads to the foundations. The buttress piers on the transversal and the secondary piers
on the longitudinal direction are attached to the main piers to improve the stiffness of this structural
integrity. The primary structural system includes all of these elements. The interior space extends on
the longitudinal axis by two semi domes, which are articulated to the structure from the eastern and
western sides. The semidomes and exedrae could also be included in the primary structural system.
The arches and vaults of the narthexes, the bays of the aisles and galleries, the walls and columns in
those areas could be identified as secondary structural system and separated from the primary
structural system as Mainstone suggested.

Figure 1. Hagia Sophia, isometric view (Mainstone 2006)

3 SEISMIC RISK AND PRECEDING STUDIES

The North Anatolian Fault is one of the largest active faults in the world and it is responsible for the
high seismicity of Marmara region. This fault has produced many large earthquakes throughout the
history. Within the last 2000 years, more than 30 earthquakes with magnitudes larger than 7 have
occurred in this region (Ambraseys 2006) The “Earthquake Master Plan for Istanbul” which was
prepared by the Istanbul Metropolitan Municipality in collaboration with universities, has determined
the major risk as a large earthquake with 60% probability of occurrence in 30 years, based on Parsons
Predictions (Parsons et al. 2000) This plan assumes a M≥7.5 earthquake scenario.
Under these circumstances, the protection of Hagia Sophia is an important task in order to conserve
the world heritage of humanity. Consequently, several scientific attempts to understand the behavior
of the structure, both in static and dynamic conditions, have been made within the last 20 years (Croci
and Viskovic 1999; Croci 2007; Çakmak et al. 2009; Durukal et al. 1997; Durukal et al. 2003; Duppel
2009; Erdik et al. 1998; Kato et al. 1992; Ozkul and Kuribayashi 2007). These attempts should and
will be continued due to the complexity of the structure and to the developments of techniques and
tools for structural analysis of architectural heritage. This contribution is planned to serve as part of
such actions.
A comprehensive research project entitled “Engineering Studies of the Main Dome and Main Pillars
of the Hagia Sophia in Istanbul” (Duppel 2009) was launched in 2004. Initiated and directed by
Wenzel, this project received significant support from the expert commission for securing the Hagia
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 3

Sophia, appointed by the Turkish government, from UNESCO, and from the German Research
Foundation (Deutsche Forschungsgemeinschaft; DFG). The project was aimed to expand and amend
the existing knowledge about the structural characteristics of Hagia Sophia, to evaluate its structural
behaviour based on exact data, and to provide the mathematical parameters required for a realistic
assessment of the structure’s stability – particularly in case of an earthquake.
The examinations were carried out with non-destructive geophysical study methods, completed very
successfully and yielded significant new findings with regard to the building’s geometry as well as to
its structural and material characteristics (Duppel 2010a; Duppel 2010b; Duppel 2011; Wenzel and
Duppel 2010) The follow-up project, which is described in this paper, is based on these findings and
will focus on a dynamic analysis of the Hagia Sophia.

4 FINITE ELEMENT MODELLING

4.1 Background of the numerical model

Schweizerhof has worked on large and accurate FE models in a DFG project in the Institute of
Mechanics KIT Karlsruhe (SCHW 307/23-1), together with Wenzel and his team, KIT Faculty of
Architecture (DFG WE 501/13-1). Those studies provided comprehensive research on the geometry
and material properties of the building and integrated all findings into a detailed CAD model
(Blankenhorn and Schweizerhof 2010). The models, derived from this work, have been designed to be
extendable and are created with common CAD and pre-processor programs to allow model enhance-
ments and geometry modifications in a fairly straightforward manner. The current model is very
detailed concerning the geometry and is completely discretized with solid elements allowing a
mechanically consistent modelling of the predominantly solid structure (Figure 2). A further detailing
was performed by Almac in 2011 (Almac 2013)
In this study, HyperMesh (HyperMesh 2009) for the pre-processing of the model and LS-DYNA
(Hallquist, and Tsay 2009) for the finite element analysis were used.

Figure 2. 3D model of Hagia Sophia including tertiary structural elements.

The structure is composed of three main parts; primary, secondary and tertiary. The first set of 3D
models contained the primary and the secondary structural parts (PS), while the second set of models
included also tertiary structural parts (PST). These sets of models were then divided into two branches
depending on their element number; coarse mesh (A1) and refined mesh (A2). In this paper, coarse
mesh model including the primary and the secondary structural parts is under investigation (PS_A1).
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 4

4.2 Computer resources

The use of high performance computers is vital to improve the quality of results during the
investigations concerning the structural behavior of Hagia Sophia under dynamic loading such as
earthquakes. For the required computer resources during the analysis, high performance clusters were
used like the HP XC3000/4000 Systems at the SCC in KIT Karlsruhe as well as the Nehalem Cluster
in the HPC computing facilities of the HLRS in Stuttgart. Those clusters were very well suited for the
so-called explicit transient analysis as well as for the implicit analyses, which were necessary for the
extraction of eigenvalues and eigenmodes and for quasi-static analyses. For both types of machines,
LS-DYNA is available in parallelized versions. During the analysis process up to 32 processors were
used in order to have the computational times as short as possible. The parallel computing information
of the analyses was also investigated to have optimum computational performance for future studies.

4.3 Material properties

In this study, masonry was modelled as a homogeneous material, which behaves within the elastic
range under compressive and tensile stresses. However, due to the different types of materials, dating
from the original construction as well as the repair works dating after the earthquakes in the 10th and
14th centuries, different material parameters were defined in order to represent the existing state of the
structure (Table 1).

Table 1. The initial material properties of the numerical model (Duppel 2009)
Materials Elasticity Poisson ratio Mass (KN/mm3)
modulus (N/mm2)
Brick masonry 6th century 1800 0,2 1,8
Brick masonry 10th century 3500 0,2 1,8
Brick masonry 14th century 4900 0,2 1,8
Stone masonry 15200 0,2 2,9

The 6th century brickwork material properties were assigned to the south and the north quarters of the
central dome, four main arches, pendentives, two semi domes and the bays of the aisles and galleries.
For the massive piers in the central space, buttress piers on the transversal and the secondary piers on
the longitudinal direction, stonework material properties were given. The 10th century brickwork
material properties were assigned to the western quarter of the main dome. The 14th century
brickwork material properties were assigned to the eastern quarter of the main dome (Figure 3).

Figure 3. Material assignment of numerical model


U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 5

4.4 Boundary conditions and element types

The foundation was assumed to be rigidly fixed to the bedrock, preventing all the translational
displacements on the boundary nodes. This assumption could be reviewed depending on the behaviour
of the structure for the future studies.
The relationship between structural parts of the building is another issue regarding to the general
behaviour of the structure. All structural parts were continuously connected to each other for all types
of analyses. A mesh of solid elements was used for the numerical analysis. The model with coarse
meshed primary and secondary parts has 92.000 nodes and 59265 elements.

5 STRUCTURAL ANALYSIS

Our numerical analyses of Hagia Sophia began with linear static analysis under dead load and
continued with an eigenvalue analysis. However, before taking these first steps, the structure was
exposed to inverse-gravity considering that only the current geometry with acting dead load is known.
By this analysis, the geometrical data of the structure including nodal coordinates of the elements were
captured in order to represent the shape of the structure for subsequent computations. In this way, the
present state of the structure including its already pre-stressed condition was taken into consideration.

5.1 Linear static analysis

Starting with this unstressed initial geometry for the analysis of the structure under its self-weight,
standard implicit solution schemes taking all nonlinearities into account were employed with an
adequate convergence tolerance and incremental loading. The results show compatible deformation
shapes in comparison with the preceding studies of the monument (Croci and Viskovic 1999; Croci
2007; Durukal et al. 1997; Duppel 2009; Erdik et al. 1998; Ozkul and Kuribayashi 2007a). Tensile and
compressive stress values are significantly below the strength of masonry. Therefore, the values are
not sufficient to provoke damages that would lead the structure into a vulnerable condition under its
self-weight.
Regarding the displacement scheme of the structure, the crowns of west and east main arches deform
inward, while the crowns of north and south arches lean outward (Figure 4). Apart from that, the
displacement values of east and west arches are remarkable. The area affected by maximum
displacements at the eastern part is more prevalent in comparison with the western part (Figure 4).

Figure 4. Total displacement of the structure under its self-weight – top view

The compressive stress distribution of the structure under its own weight indicates maximum values at
the main piers at the ground level. Regarding the secondary structural system, the columns of the
aisles between main piers have also high stress values (approximately 2,25 MPa) in comparison with
the other parts of the structure.
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 6

At the very first sight, the most important part of the tensile stress distribution of the structure
concerns two zones. On the ground floor, the connection between buttress piers and main piers on the
transversal direction seems to be the first area, which attracts attention. Due to the thrust of the east
and west main arches, main piers and buttress piers tend to incline outward and therefore the
connections between those load bearing vertical elements are put under stronger tensile straining. The
vaults of the aisles, the gallery and the barrel vaults adjacent to the semi domes on the longitudinal
direction are other zones where major tensile stresses appear.

5.2 Eigenvalue analysis

In order to determine vibration characteristics of the numerical model, eigenvalue analyses were
carried out. The first and the second modes correspond to horizontal translations in the longitudinal
(E-W) and transversal (N-S) directions. The third mode shape is clearly dominated by torsional motion
(Figure 5).

Figure 5. First three mode shapes of the structure

The natural frequencies of the FE model were compared with ambient vibration data and the initial
material properties of the model were modified according to the first 5 modes of the structure
(Durukal et al. 1997). Within this context, only the elasticity modulus of stone masonry was reduced.
With the adjustment of material properties, the structure corresponds to the ambient vibration
frequencies quite satisfactorily (Table 2). The modal effective mass chart – not documented in this
paper - presents the contributions of the higher modes to the behaviour of structure, which also
indicates the structural complexity of the monument.

Table 2. Modal frequencies of Hagia Sophia – comparing measured and computed eigenfrequencies
Modes Ambient Numerical model
vibration frequencies* vibration frequencies
1 1.85 1.92
2 2.10 2.08
3 2.35 2.35
4 2.50 2.50
5 2.70 2.65
* data from the surveys (Durukal et al. 1997).

5.3 Dynamic analysis

The dynamic analysis started after the static initialization. Thus, it began with a static implicit step for
the gravity loading and then switched to explicit time integration to perform transient analysis of the
structure. For the earthquake simulation, a prescribed displacement time history was imposed to the
structure through the nodes on the connection between ground and super-structure.
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 7

The stress distribution of the structure under lateral loads demonstrates the east and west main arches,
together with adjacent portions of semi domes and central dome, as the most vulnerable areas of the
structure (Figure 6). These critical areas are also coherent with the partial collapses in the10th and 14th
century earthquakes.

Figure 6. Displacements on E-W direction under longitudinal excitation

During the earthquake loading imposed on the transversal direction, north and south arches oscillate
coherently with the loading direction. However, east and west arches of the structure seem to deform
not only in the transversal direction but also in the longitudinal direction (Figure 7). This deformation
pattern could also be tracked in the second vibration mode of the structure regarding the eigenvalue
analysis.

Figure 7. Displacements on vertical and longitudinal directions of east and west main arches

One of the issues that get attention is the behaviour of tympanum walls. According to Mainstone, the
tympanum walls were reconstructed after the earthquake of 869. As these walls are slender because of
their limited thickness in comparison with their height, they deform remarkably due to the out of plane
mechanism (Figure 8). Since these elements are infill walls between main piers and the north-south
main arches, they are not the major focus in attention concerning stresses or failure. However, they
limit the displacements of north and south main arches in longitudinal and vertical directions.

Figure 8. Displacements on tympanum walls due to out of plane mechanism


U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 8

6 CONCLUSIONS AND FUTURE STUDIES

Linear static analyses indicate that tensile and compressive stress values are not sufficient to provoke
damages that would lead the structure into a vulnerable condition under its self-weight. Regarding to
the dynamic characteristics of the structure, the numerical model demonstrates very good
correspondence with the ambient vibration frequencies. Mode shapes and mass participation factors
indicate the complexity and irregularity of the structure. According to the dynamic analysis, the east
and west main arches, together with adjacent portions of semi domes and central dome are the most
vulnerable areas of the structure.
Forthcoming studies will be focused mainly on two issues. The first step will be on the usage of
appropriate earthquake data for the dynamic analysis. Obtaining the most suitable earthquake data
related to the seismic characteristics of the zone is a very delicate issue and needs high expertise.
The second issue is related to the interaction between main arches and semi domes. For this particular
purpose, the current numerical model assuming continuous connections will be modified and between
those particular structural parts, contact conditions will be defined. These contact elements will allow
separation and repeated contact between the surfaces of structural parts during dynamic simulations.
In the case of pounding between the surfaces, the reaction forces could be also captured through the
contact elements.
The contribution of tertiary structural parts on the behaviour of monument under dynamic loading is
another question to be verified. So far, none of the preceding studies regarding to the structural
behaviour of Hagia Sophia considered tertiary structural elements. However, the answer could not be
satisfied, unless input parameters on the interaction between buttresses with the main structure are
provided. A detailed site survey regarding to the efficiency of the bond between each buttress and
exterior façade of the structure is needed.
The interaction between structural elements having different construction dates is another important
issue regarding to the vulnerability of the monument under earthquake excitations. During the initial
steps of this research activity, the 3D mesh setup of the numerical model was somehow adjusted to
accomplish these kinds of particular analyses. However, site surveys are crucial in order to provide
reliable data for the FE-models before taking any further computational effort. Further aspects could
concern local material properties, thus a larger variation throughout the structure as well as the
foundation could be obtained beyond the current simple assumptions. It would be very interesting to
take a larger variation of the currently assumed properties into account.

ACKNOWLEDGEMENTS

It is a pleasure to acknowledge the scholarship generously provided by HPS-Europa2 Transnational


Access Programme (the project was funded by the European Commission - DG Research in the
Seventh Framework Programme under grant agreement n° 228398, http://www.hpc-europa.eu/) and
HPC computing facilities of the HLRS in Stuttgart as well as SCC in KIT Karlsruhe in 2011. The
author would also like to thank to all members of the Institute of Mechanics at KIT Karlsruhe for their
great hospitality.

REFERENCES

Almac, U. (2013). Numerical Analysis of Hagia Sophia, HPS-Europa Report, Stuttgart, Germany.
Ambraseys., N.N. (2006). Comparison of frequency of occurrence of earthquakes with slip rates from
long-term seismicity data: the cases of Gulf of Corinth, Sea of Marmara and Dead Sea fault zone.
Geophysical Journal International, 165: 516–526.
U. Almac, K. Schweizerhof, G. Blankenhorn, C. Duppel, F. Wenzel / VEESD 2013 9

Blankenhorn, G. and Schweizerhof, K. (2010). Numerische Modelle der Hagia Sophia in Istanbul,
Institut für Mechanik, Karlsruher Institut für Technologie. [online] Available at:
<http://www.ifm.kit.edu/english/19.php>.
Cakmak, A., Taylor, R., Durukal, E. (2009). The structural configuration of the First Dome of
Justinian’s Hagia Sophia (A.D. 537-558): an investigation based on structural and literary analysis.
Soil Dynamics and Earthquake Engineering, 29: 693-698.
Croci, G., and Viskovic, A. (1999). Hagia Sophia in Istanbul: study, research and mathematical
models for the analysis of the structural behaviour program of investigation, Report, Istanbul,
Turkey.
Croci, G. (2007). Santa Sophia, Istanbul, Il Comportamento Strutturale. Recupero e Conservazione,
78: 36-41.
Duppel, C. (2010a). Ingenieurwissenschaftliche untersuchungen an der Hagia Sophia in Istanbul – teil
1: das konstruktionsgefüge. Bautechnik, 87, H. 11: 708–716.
Duppel, C. (2010b). Ingenieurwissenschaftliche untersuchungen an der Hagia Sophia in Istanbul – teil
2: zum tragverhalten. Bautechnik, 87, H. 12: 790–799.
Duppel, C., (2009). Ingenieurwissenschaftliche untersuchungen an der hauptkuppel and den
hauptpfeilem der Hagia Sophia in Istanbul. Ph.D. Karlsruhe Institute of Technology.
Duppel, C. (2011). On the Structural Characteristics and Present Stability of the Hagia Sophia in
Istanbul. In: WCCE-ECCE-TCCE Joint Conference 2, Seismic Protection of Cultural Heritage,
Conference Proceedings; October 31-November 1, 2011, Antalya, Turkey. 175–188.
Durukal, E., Erdik, M., Çakmak, A.Ş. (1997). Assessment of the earthquake performance of Hagia
Sophia, In: Arun, G., Seckin, N. (eds), Proceedings of the International Conference on Studies in
Ancient Structures, Istanbul, Turkey. 407–415.
Durukal, E., Cimilli S., Erdik, M. (2003). Dynamic response of two historical monuments in Istanbul
deduced from the recordings of Kocaeli and Duzce earthquakes. Bulletin of the Seismological
Society of America, 93: 694–712.
HyperMesh®, Vers. 10.0. (2009). Altair Engineering GmbH, Böblingen, Germany.
Hallquist, J.O., Tsay, C.S. (2009). LS-DYNA®, Nonlinear dynamic analysis of solids in three
dimensions, Users Manual, Livermore Software Technology Corp., Livermore. Used LS-DYNA
version 970.
Kato, S., Aoki, T., Hidaka, K., Nakamura, H. (1992). Finite element modelling of the first and second
domes of Hagia Sophia. In: Mark, R., Cakmak, A. (eds), The Hagia Sophia from the age of
Justinian to the present. Cambridge University Press.
Mainstone, R. J. (2006). Hagia Sophia: architecture, structure and liturgy of Justinian’s great church.
Thames and Hudson.
Müller-Wiener, W. (2007). İstanbul'un tarihsel topografyası: 17. yüzyıl başlarına kadar Byzantion -
Konstantinopolis – İstanbul. Yapı Kredi Yayınları.
Ozkul, T. A., Kuribayashi, E. (2007a). Structural characteristics of Hagia Sophia: I-A Finite element
formulation for static analysis. Building and Environment, 42: 1212-1218.
Ozkul, T. A., Kuribayashi, E. (2007b). Structural characteristics of Hagia Sophia: I-A Finite element
formulation for dynamic analysis. Building and Environment , 42: 2100-2106.
Procopius. (2006). Buildings. Translated by H.B. Dewing, Harvard University Press.
Parsons, T., Toda, S., Stein, R.S., Barka, A., and Dieterich, J.H. (2000). Heightened odds of large
earthquakes near Istanbul: an interaction-based probability calculation. Science, 288: 661-665.
Van Nice, R. L. (1963). The structure of Hagia Sophia. Architectural Forum, 18:131-139
Wenzel, F., Duppel, C. (2010). Investigations into the construction and repair history of the Hagia
Sophia. Annual of Ayasofya Museum, 13.

View publication stats

Das könnte Ihnen auch gefallen